energies

Review
AI-Driven Innovations in Building Energy Management Systems:
A Review of Potential Applications and Energy Savings
Dalia Mohammed Talat Ebrahim Ali 1 , Violeta Motuzienė 2, * and Rasa Džiugaitė-Tumėnienė 2

1 Lithuanian Energy Institute, Breslaujos Str. 3, LT-44403 Kaunas, Lithuania; dalia.ali@lei.lt

2 Department of Buildings Energetics, Faculty of Environmental Engineering, Vilnius Gediminas Technical
University, Sauletekio Av. 11, LT-10223 Vilnius, Lithuania; rasa.dziugaite-tumeniene@vilniustech.lt
* Correspondence: violeta.motuziene@vilniustech.lt

Abstract: Despite the tightening of energy performance standards for buildings in various countries
and the increased use of efficient and renewable energy technologies, it is clear that the sector needs
to change more rapidly to meet the Net Zero Emissions (NZE) scenario by 2050. One of the problems
that have been analyzed intensively in recent years is that buildings in operation use much more
energy than they were designed to. This problem, known as the energy performance gap, is found in
many countries and buildings and is often attributed to the poor management of building energy
systems. The application of Artificial Intelligence (AI) to Building Energy Management Systems
(BEMS) has untapped potential to address this problem and lead to more sustainable buildings. This
paper reviews different AI-based models that have been proposed for different applications and
different buildings with the intention to reduce energy consumption. It compares the performance of
the different AI-based models evaluated in the reviewed papers by presenting the accuracy and error
rates of model performance and identifies where the greatest potential for energy savings could be
achieved, and to what extent. The review showed that offices have the greatest potential for energy
savings (up to 37%) when they employ AI models for HVAC control and optimization. In residential
and educational buildings, the lower intelligence of the existing BEMS results in smaller energy
savings (up to 23% and 21%, respectively).

Citation: Ali, D.M.T.E.; Motuzienė, V.; Keywords: artificial intelligence; buildings; energy efficiency; saving potential; HVAC; BEMS
Džiugaitė-Tumėnienė, R. AI-Driven
Innovations in Building Energy
Management Systems: A Review of
Potential Applications and Energy
1. Introduction
Savings. Energies 2024, 17, 4277.
https://doi.org/10.3390/en17174277
The operation of buildings accounts for 30% of global final energy consumption and
26% of global energy-related emissions [1], and this sector has remained a priority for
Academic Editor: Constantinos sustainable development for decades. Although minimum performance standards and
A. Balaras
building energy codes are becoming more comprehensive and stringent across countries,
Received: 21 July 2024 and the use of efficient and renewable building technologies is increasing, the IEA reports
Revised: 22 August 2024 that energy consumption in the buildings sector is still an issue, as global growth in floor
Accepted: 23 August 2024 space is more than offsetting the increased efficiency and decarbonization efforts [1]. In
Published: 27 August 2024 addition, there is plenty of evidence that even new buildings do not perform as well as they
were designed to (a problem called the Energy Performance Gap), which is presented in
detail in the recent review of Bai et al. [2]. It is obvious that the sector needs faster change
to get on track towards the Net Zero Emissions (NZE) scenario by 2050.
Copyright: © 2024 by the authors. The European Union’s (EU) increased climate and energy ambition requires all new
Licensee MDPI, Basel, Switzerland. buildings to be zero-emission by 2030 and existing buildings to be zero-emission by 2050.
This article is an open access article
The recent recast of the EPBD [3] pays more attention to the energy efficiency of existing
distributed under the terms and
buildings, as 75% of EU buildings are still energy-inefficient. New policy measures empha-
conditions of the Creative Commons
size the importance of digitalization, monitoring, building automation and smartness (IoT),
Attribution (CC BY) license (https://
data collection, and sharing, which can be listed as follows:
creativecommons.org/licenses/by/
4.0/).

Energies 2024, 17, 4277. https://doi.org/10.3390/en17174277 https://www.mdpi.com/journal/energies

Energies 2024, 17, 4277 2 of 35

• Deployment of High-Capacity Communication Networks: to facilitate smart homes

and well-connected communities.
• Targeted Incentives: to promote smart-ready systems and digital solutions in the
built environment.
• Use of Digital Technologies: for the analysis, simulation, and management of buildings.
• Smart-Readiness Indicator: to measure the capacity of buildings, to use information
and communication technologies and electronic systems, and to adapt their operation
to the needs of occupants and the grid.
• Building Automation and Electronic Monitoring: to improve the energy efficiency
and overall performance of buildings and to provide confidence to occupants about
actual savings.
• National Databases for Energy Performance: to collect data on the energy performance
of buildings and transfer this information to the EU Building Stock Observatory.
All these tools and technologies are familiar to scientists and pioneers in the building
sector. However, regulating them will significantly speed up their practical application.
This includes artificial intelligence, which, while not explicitly mentioned, is inherently
connected to the aforementioned areas.
The practical use of IoT and AI in building systems management is still in its early
stages, but the future is exceedingly promising. Some believe that facilities management
could be the industry to gain the most from AI in the coming years, particularly due to the
high volume of repetitive and time-consuming tasks. With the AI-powered management
of buildings, better energy efficiency is expected first and foremost, accompanied by
additional benefits, such as lower overall maintenance costs, better contractor relationships,
and enhanced asset reliability.
The IEA estimates that digitalization could cut total energy use in residential and
commercial buildings by around 10% by 2040 [4]. But what contribution and potential
does the application of AI offer, bearing in mind that not all buildings can benefit from
sophisticated control due to the low level of intelligence of their systems? Which buildings
have the highest energy-saving potential, and what AI models demonstrate the highest
performance? The answers to these questions will be further provided in this review.

1.1. AI Applications
AI has rapidly permeated many aspects of our lives and has revolutionized industries
and enhanced efficiency in various domains, such as healthcare, education, manufacturing,
finance, and transportation. Additionally, AI has emerged as a powerful tool for achieving
sustainability in the building sector. Its technologies and methodologies have demonstrated
great potential in increasing energy efficiency and reducing costs [5]. In the context of BEMS
(Building Energy Management Systems), AI has been applied in predicting and forecasting
a building’s energy consumption, providing occupant behavior insights, achieving thermal
comfort, improving indoor air quality, as well as enhancing maintenance and operational
efficiency [6]; on top of that, other applications can be found, as presented in several
review papers [7–9]. The difference is not just in the application but also in the models that
are employed.
AI is a broad field that aims to create systems capable of performing tasks that require
human intelligence. The field of AI encompasses a wide range of domains and includes
learning as well as non-learning methods, such as robotics, natural language processing,
autonomous systems, and expert systems. Different AI models are designed to perceive,
analyze, learn, reason, find patterns, and make decisions and predictions based on the
information and data provided [5]. They could be generally categorized into different
types, each serving different uses and objectives and employing various techniques. One
such technique is Machine Learning (ML). ML is a field of study within AI that focuses
on developing algorithms that enable computers to learn from data and improve their
performance over time. While all ML is part of AI, not all AI equals ML. AI is the broader
concept of intelligent machines, whereas ML is a specific approach within AI that focuses
 concept of intelligent machines, whereas ML is a specific approach within AI that focuses
on learning from data [10]. ML has been recently gaining popularity due to its ease of use,
wide applicability, continuous learning, abundant and cheap computation, and the fact
that ML-based models do not require human intervention [11]. The most popular ML al-
Energies 2024, 17, 4277 3 of 35
gorithms are broadly divided into three categories: supervised, unsupervised, and rein-
forcement learning. Supervised learning is further categorized into classification and re-
gression. Classification techniques include Naive Bayes classifiers, decision trees, support
on learning from data [10]. ML has been recently gaining popularity due to its ease of use,
vector machines, random forests, and K-nearest neighbors, whereas regression methods
wide applicability, continuous learning, abundant and cheap computation, and the fact that
include linear regression, neural network regression, decision tree regression, lasso re-
ML-based models do not require human intervention [11]. The most popular ML algorithms
gression, anddivided
are broadly ridge regression. Unsupervised
into three categories: learningunsupervised,
supervised, focuses on clustering techniques
and reinforcement
like K-means clustering, mean shift clustering, and Gaussian mixtures. On the
learning. Supervised learning is further categorized into classification and regression. other hand,
reinforcement
Classificationlearning involves
techniques includemodels
Naivesuch
Bayesasclassifiers,
Q-learning, R-learning,
decision trees,and temporal
support vector dif-
ference learning. Each of these approaches has specific applications and
machines, random forests, and K-nearest neighbors, whereas regression methods include strengths, con-
tributing to the diverse
linear regression, neuralcapabilities of machine
network regression, learning
decision in solvinglasso
tree regression, real-world problems
regression, and
[12,13]. ML is prominent
ridge regression. and extensively
Unsupervised appliedon
learning focuses inclustering
BEMS. A techniques
wide rangelike of K-means
studies ap-
clustering,
plying mean shift
ML models clustering,
have and Gaussian
been discussed in thismixtures. On the other
paper. Therefore, hand, reinforcement
in addition to “Artificial
Intelligence”, “Machine Learning” has been included in the list of keywords to learning.
learning involves models such as Q-learning, R-learning, and temporal difference capture as
Eachrelevant
many of these approaches has specific applications and strengths, contributing to the diverse
studies as possible.
capabilities
Differentofreview
machine learning
papers in solving
were analyzed real-world problems
and compared in [12,13]. ML to
this paper is prominent
find what is
and extensively applied in BEMS. A wide range of studies applying ML
still not identified regarding the application of AI for the improvement of buildings’ models haveen-
been discussed in this paper. Therefore, in addition to “Artificial Intelligence”, “Machine
ergy efficiency. They are presented below.
Learning” has been included in the list of keywords to capture as many relevant studies
as possible.
1.2. Related Reviews
Different review papers were analyzed and compared in this paper to find what is still
notUsing the keyword
identified regardingstring mentioned
the application ofin
AIthe
for“Methodology” section
the improvement in the search
of buildings’ energyen-
gine “SCOPUS”,
efficiency. as of
They are 30 Aprilbelow.
presented 2022, with only review papers included, the number of
review papers after the title and abstract screening was 54. The number of review papers
in1.2.
the Related Reviews
field has been growing, reflecting a growing interest among researchers in this study
Using 1).
area (Figure the keyword string mentioned in the “Methodology” section in the search
engine
The “SCOPUS”, as of 30 April
following paragraphs 2022, with
discuss onlyreview
just the reviewpapers
papers selected.
included,The
the number of
papers were
review papers after the title and abstract screening was 54. The number of review
selected based on their topics being relatively similar to this review paper and having papers in
the field has been growing, reflecting a growing
conclusions that are relevant to the scope covered. interest among researchers in this study
area (Figure 1).
16
14
11
7

2020 2021 2022 2023 2024

Figure 1. Number of Review Papers relevant in recent years [SCOPUS].

Figure 1. Number of Review Papers relevant in recent years [SCOPUS].
The following paragraphs discuss just the review papers selected. The papers were
Reviews
selected on energy
based consumption
on their topics beingbehavior and prediction.
relatively similar to Wu
this et al.’s [14]
review paperreview aimed to
and having
analyze energy
conclusions consumption
that are relevant tobehavior
the scopeand assess the prediction performance and inter-
covered.
pretability
Reviewsof on
ML-based baseline modeling
energy consumption behavior and techniques
prediction. across various
Wu et al.’s major building
[14] review aimed
to analyze
types. energy identified
The authors consumption the behavior andinfluencing
key factors assess the prediction performance
the performance and base-
of energy in-
terpretability of ML-based baseline modeling techniques across various
line modeling for different buildings and investigated and compared building profiles to major building
types. explain
further The authorsthe identified
differences theinkey factors influencing
baseline the performance
modeling results of energy
[14]. Moghimi baseline
et al. [15] re-
viewed ML methods applied to improve building energy consumption in further
modeling for different buildings and investigated and compared building profiles to modern
explain the differences in baseline modeling results [14]. Moghimi et al. [15] reviewed
ML methods applied to improve building energy consumption in modern building data
processing, focusing on their accuracy and efficiency, and found that hybrid ML models
predict energy consumption with an accuracy up to 15% higher compared to that of single
ML models. Farzaneh et al. [9] provided an in-depth review of recent studies on the applica-
tion of AI technologies in smart buildings through the concept of the building management
system and demand response programs. In their research paper, they mentioned some
future directions and recommendations, like the need to improve prediction methods that
Energies 2024, 17, 4277 4 of 35

consider building characteristics and environmental conditions, develop standardized

protocols and policies as guidance for AI technology implementation, and address the
security and privacy concerns associated with the data collected [9].
Thermal comfort and energy efficiency. Merabet et al. [16] discussed the application of AI
and focused mainly on improving thermal comfort and energy efficiency in building control
systems. They concluded that AI technology’s application in building control shows great
promise but remains an ongoing challenge, as the performance of AI-based control systems
is not yet entirely satisfactory, mainly because these algorithms require substantial amounts
of high-quality real-world data, which are often lacking in the building energy sector. From
1993 to 2020, AI techniques and personalized comfort models have demonstrated average
energy savings between 21.81% and 44.36%, with comfort improvements ranging from
21.67% to 85.77% [16]. A review by Ghahramani et al. [8] emphasizes the necessity of a
cohesive system comprising sensors, infrastructure, learning algorithms, and actuators
governed by a central intelligent system to improve comfort and energy efficiency. The
paper concludes that improvements in all aspects of a smart system are needed to achieve a
better determination of the correct combination of systems to increase the system’s overall
efficiency and improve comfort [8].
The IoT and AI for building energy control. A review paper by Broday et al. [7] examines
how the IoT is being used in building control to save energy and monitor indoor environ-
mental quality. Their findings show that the main application of the IoT in buildings is to
reduce energy consumption and that ML methods are mainly used to save energy and un-
derstand occupant behaviour to achieve thermal comfort [7]. A review by Sayed et al. [17]
explores various DL models like CNNs, RNNs, LSTMs, GANs, and autoencoders and
their advantages and limitations in occupancy detection, where the authors concluded that
several directions are provided to reduce privacy problems by employing forthcoming
technologies such as edge devices, Federated Learning, and blockchain-based IoT [17].
Reviews on models and their reliability. Bashir et al. [18] discuss the various models used
for predicting the cooling and heating loads in smart buildings and the role of accurate
load prediction in enhancing energy efficiency by including a detailed analysis of AI
algorithms, such as ANN, SVM, and DL, presenting their advantages and limitations in
load forecasting. The review shows that AI-based models achieve higher accuracy but often
require extensive data and computational resources [18]. A review by Rodrigues et al. [19]
systematically analyzes various modeling techniques for STLF in the residential sector
over the past decade, identifying various modeling techniques and associated algorithms.
Additionally, the paper discusses AI models’ advantages in handling nonlinear problems
and the necessity of adequate historical data for optimal performance [19]. Runge et al. [20]
discuss the application of DL models in predicting energy consumption in buildings,
where their key findings reveal that DL models, such as RNNs, CNNs, and DBNs, offer
better performance in handling large datasets and extracting features compared to the
traditional methods.
Model performance, challenges, and future directions. According to a review by Mo-
hgimi et al. [15], DL models such as DNN and LSTM have shown high accuracy in predict-
ing energy consumption and managing building systems. However, despite their accuracy,
these models demand significant computational resources and extended training times.
These models are evaluated based on their effectiveness in reducing energy consump-
tion and operational costs, particularly in smart building applications [15]. A review by
Runge et al. [20] discussed that white-box models use detailed physical equations to repre-
sent energy systems, offering deep insights into system dynamics but requiring extensive
parameter measurements. Data-driven models, including black-box and grey-box types,
rely on data to establish mathematical relationships without detailed system knowledge,
making them easier to implement but often less interpretable. Performance-wise, deep
learning models like RNNs and DNNs excel in accuracy with large datasets but demand
significant computational resources [20].
Energies 2024, 17, 4277 5 of 35

A review by Mousavi et al. [21] identified some challenges, such as the difficulty
in understanding the reasoning behind the model’s decisions because most AI-based
predictive models are black-box by nature. The review also pointed out the limitations
of supervised learning, stating that the building industry relies heavily on supervised
learning methods, which require labeled data, and that alternative learning methods, such
as semi-supervised learning or reinforcement learning, could eliminate this issue [21].
In the context of short-term household forecasting, Ma et al. [22] stated the importance
of enhancing the generalization ability of DL models, as they are prone to overfitting.
Household forecasting involves many uncertain factors, and integrating these uncertainties
into DL models is a challenging problem [22]. Shaqour et al. [23] provided a review of
the recent advancements in DRL-based BEMS for different building types. The authors
observed that residential and office buildings were the most explored types of buildings.
There is still a clear gap in real implementations and system validations, where only 11%
of the recent works have been reported so far. The authors suggested that future research
should focus on the data efficiency of DRL models due to the lack of real-world validations.
This could be accomplished by using virtual building systems for offline pretraining and
exploring methods for reducing the requirement for large amounts of data [23].
Based on other reviews, it can be concluded that when AI models are used in predic-
tions, the building type matters. Also, in their reviews, different authors demonstrated that
models used in predictions provide different reliabilities in different situations, and many
authors discuss issues related to the quality, reliability, and amount of data. Also, different
authors agree that in real-life applications, the ability to integrate AI-based models into
controllers is still in its infancy and is limited.
This review paper aims to present AI’s contributions to BEMS, identify which types
of buildings, varying in their level of intelligence, have the highest energy-saving po-
tential, and determine which AI models perform the best in this area by estimating the
overall efficiency.
To identify the added value of the paper, it was compared to the outcomes of the
three most similar review papers: Ardabili et al. [24], Yan et al. [25], and Tien et al. [26].
The reviews by Ardabili et al. [24] and Yan et al. [25] both share a similar scope with
our work. They explore various AI models applied to building applications such as
energy consumption prediction, load forecasting, occupant detection, and optimization.
Ardabili et al.’s paper also discussed the evaluation criteria [24,25]. Meanwhile, the review
by Tien et al. [26] also closely aligns with our work. However, compared to these reviews,
the added value of our paper is (1) our review stands out by providing detailed numerical
values and discussing the different building types where these AI models have been
applied, which were not covered in either Ardabili et al. [24] or Yan et al.’s [25] reviews;
(2) our review stands out in its distinct structure of application areas, its inclusion of study
locations, and its more in-depth discussion of evaluation metrics compared to Tien et al.’s
review paper [26].

2. Methodology
The methodology implemented in this review paper is the Systematic Literature Re-
view (SLR). This methodology comprises the following steps: formation of the research
question(s), validation of keywords, setting eligibility and inclusion criteria, systematic
search, screening and exclusion criteria, and analysis and synthesis [27]. To ensure trans-
parency, the review follows the PRISMA (Preferred Reporting Items for Systematic Reviews
and Meta-Analysis) guidelines and checklists.
The search process aims to find relevant studies based on the research question, defined
based on the article’s objective. The research question was formulated following the widely
known PICO method, which covers four elements of a search question: Population (Who?),
Intervention (What?), Comparison (Compared to what?), and Outcome (What are you
trying to accomplish/improve?) [27]. The questions for outlining PICO components are
presented in Table 1.
 widely known PICO method, which covers four elements of a search question: Population
(Who?), Intervention (What?), Comparison (Compared to what?), and Outcome (What are
you trying to accomplish/improve?) [27]. The questions for outlining PICO components
Energies 2024,are presented in Table 1.
17, 4277 6 of 35
The research question of this review is formulated as follows: What AI models are
employed in BEMS, and how do they contribute to energy savings?
Table 1. Questions for outlining PICO components.
Table 1. Questions for outlining PICO components.
P I C O
P I
Population
C
Intervention/ Comparison/ O Outcome
factors Circumstances
Intervention/ Comparison/
Population Compare the Outcome Efficiency and potential
factors Circumstances
reliability/accuracy and savings achieved by
Various AI models
Buildings error rate of Efficiency
AI models and potential
AI-based models and their
Compare the reliability/ac-
and their reliability
Various AI models savings
used to achieve energyachieved by AI-to enhancing
contribution
curacy and error rate of AI in buildings
efficiency BEMS for energy efficiency
Buildings and their reliabil- based models and their
models used to achieve en-
ity contribution to enhancing
The researchergy efficiency
question of thisin buildings
review is formulated
BEMS foras energy
follows:efficiency
What AI models are
employed in BEMS, and how do they contribute to energy savings?

Search Strategy Search Strategy

The initial keywords have been
The initial defined
keywords haveasbeen
follows: Artificial
defined Intelligence,
as follows: Buildings, Buildings,
Artificial Intelligence,
Energy, HVAC,
Energy, HVAC, Control, Control,Forecasting,
Optimization, Optimization,and
Forecasting,
Occupancy andDetection.
OccupancyThe Detection.
key- The key-
word string used in the search engines is
word string used in the search engines is shown in Figure 2. shown in Figure 2.

Figure 2. KeywordsFigure
string.2. Keywords string.

The search enginesThe search engines

“SCOPUS” and “SCOPUS” and “Web
“Web of Science” of Science”
were exploredwere
on explored
30 April on 30 April 2024.
2024.
The inclusion
The inclusion criteria criteria for
for the selected the selectedand
publications publications and the
the keyword keyword
search search
string are string are
as follows: as follows:
• Recent publication: publications not older than five years (2019–2024).
• Recent publication: publications not older than five years (2019–2024).
• Language: any.
• Language: any.
• Publication type: journal articles, conference papers, and books.
• Publication type:
• journal articles,
Geographic conference
coverage: papers, and books.
worldwide.
• Geographic coverage: worldwide.
The selection criteria have been implemented in the selection process to determine
which papers to include in the analysis, as presented in Figure 3. It illustrates the exclusion
criteria for studies in the research selection process. The initial step checks if the study
includes all the necessary information (title, author, abstract). If it does, the next criterion
 The selection criteria have been implemented in the selection process to determine
which papers to include in the analysis, as presented in Figure 3. It illustrates the exclusion
criteria for studies in the research selection process. The initial step checks if the study
includes all the necessary information (title, author, abstract). If it does, the next criterion
Energies 2024, 17, 4277 7 of 35
ensures the study is recent (published within the last 5 years). The third step verifies that
the study is not a review paper, as review papers are discussed separately in the Introduc-
tion. Following this, the study must not cover renewable energy systems, as including
ensures the study is recent (published within the last 5 years). The third step verifies that
them would broaden the scope and slightly divest from the core subject of this paper,
the study is not a review paper, as review papers are discussed separately in the Introduction.
which is AI for energy
Followingmanagement andmust
this, the study optimization in buildings,
not cover renewable without
energy necessarily
systems, as including them
altering the primary
would energy
broadensources of these
the scope buildings.
and slightly Additionally,
divest from the core both areas
subject of present
this paper, which
distinct challenges. Energy
is AI management
for energy management forand
energy efficiency
optimization mainly deals
in buildings, withoutwith chal- altering
necessarily
lenges related to occupant behavior, load control, and reducing consumption using intel- distinct
the primary energy sources of these buildings. Additionally, both areas present
ligent systems. Bychallenges. Energy management
contrast, renewables involvefor energyand
storage efficiency mainly deals with
grid compatibility challenges related
challenges,
which are out of this review’s scope. Therefore, limiting this review to non-renewables systems.
to occupant behavior, load control, and reducing consumption using intelligent
By contrast, renewables involve storage and grid compatibility challenges, which are out
ensures a more explicit comparative analysis of the AI models specifically designed for
of this review’s scope. Therefore, limiting this review to non-renewables ensures a more
energy management systems and a clearer comparison of error metrics and performance
explicit comparative analysis of the AI models specifically designed for energy management
measures. Then, the study
systems and should include
a clearer a developed
comparison of errorAI model.
metrics andFinally, the study
performance must Then, the
measures.
provide performancestudy reliability,
should include savings, error AI
a developed metrics,
model. or similar
Finally, measures
the study must for the AI
provide performance
model. If all thesereliability,
criteria are met, the study is included; otherwise, it is excluded at the
savings, error metrics, or similar measures for the AI model. If all these criteria
corresponding step.are met, the study is included; otherwise, it is excluded at the corresponding step.

Figure 3. Paper selection

Figure criteria.
3. Paper selection criteria.

A PRISMA chart A is PRISMA

a flow diagram
chart is aused
flow to present
diagram thetodifferent
used stages
present the of the
different selection
stages of the selection
process in systematic reviews. It helps visualize the number of studies identified, screened,
process in systematic reviews. It helps visualize the number of studies identified,
and included
screened, and included in the in the review
review [28].[28].
The The PRISMA
PRISMA chart
chart ensurestransparency
ensures transparency in inreporting
re- and
explains how the final set of studies was determined [28]. As can be seen from the PRISMA
porting and explains how the final set of studies was determined [28]. As can be seen from
chart (Figure 4), 1396 records were initially identified through a SCOPUS database search,
the PRISMA chart (Figure 4), 1396 records were initially identified through a SCOPUS
and 414 were identified through Web of Science. After eliminating duplicates, 1615 records
database search, and 414 were
remained. identified
A total through
of 419 records Web
were of Science.
screened basedAfter eliminating
on their dupli- resulting
title and abstract,
cates, 1615 recordsin remained.
the exclusionA total
of 1196of records.
419 records were
Finally, 148screened based on
full-text articles weretheir title and
assessed for eligibility,
abstract, resultingand
in 271
the were
exclusion of 1196
excluded for not records.
meetingFinally, 148 full-text
the selection criteria. articles were as-
sessed for eligibility, and 271 were
The papers excluded
selected for not meeting
are systemized the selection
and analyzed criteria.
from different perspectives in a bid
to find the answers to the raised research question.
2024, 17, xEnergies
FOR PEER 17, 4277
2024,REVIEW 8 of 35 8 of 35

Figure 4. PRISMA Chart.

Figure 4. PRISMA Chart.

The papers3. selected

Results are systemized and analyzed from different perspectives in a bid
to find the answers Theto the raised research
following question.
subchapters compare a wide range of AI models used for various
applications, such as energy consumption forecasting, load forecasting, HVAC control and
3. Results optimization, and occupant detection, in different types of residential and non-residential
The following subchapters
buildings. compare a wide
Key performance range
metrics likeofRMSE,
AI models
MSE,used
MAE, forMAPE, ap- R2 (coefficient
variousand
plications, suchofasdetermination) are provided.
energy consumption These load
forecasting, metrics are crucial
forecasting, for evaluating
HVAC the accuracy and
control and
effectiveness
optimization, and occupantof these AI models,
detection, as well
in different typesas of
potential energy
residential andand cost savings, demonstrating
non-residential
buildings. Keytheir practical metrics
performance benefitslike
in real-world
RMSE, MSE, applications.
MAE, MAPE, and R2 (coefficient of
determination)•are RMSEprovided. These metrics
is a widely are crucial
used metric for evaluating
for measuring the accuracy
the differences betweenandvalues predicted
effectiveness of these AI models, as well as potential energy and cost savings, demonstrat-
by a model and values that are observed. It is sensitive to large errors, providing a
ing their practical benefits in real-world
clear picture applications.
of the model’s performance [29]. RMSE is expressed in the same units
• RMSE is a widelyas theused metric for
dependent measuring
variable the differences
(e.g., kWh or kWh/mbetween2 for energyvalues pre-
consumption). It can
dicted by a model
alsoand values thatinare
be expressed manyobserved. It is sensitive
other units to large fields,
across different errors,including
provid- temperature
in ◦ Cof(Celsius),
ing a clear picture the model’s◦ F (Fahrenheit),
performance or [29].K (Kelvin), pressure ininPa
RMSE is expressed (Pascals)
the same or bars, and
concentration in ppm (parts 3 (micrograms per cubic meter) [29].
units as the dependent variable (e.g., kWhper or million)
kWh/m2orfor µg/m
energy consumption). It
can also be• expressed
MSE is in similar
manyto RMSE
other unitsbutacross
does different
not involve taking
fields, squaretempera-
including roots. It averages the
squares°Fof(Fahrenheit),
ture in °C (Celsius), the errors, emphasizing
or K (Kelvin), larger errors more
pressure in Pathan smaller
(Pascals) or ones.
bars, MSE measures
the average
and concentration magnitude
in ppm (parts of errors
per million) in a 3set
or µg/m of predictions
(micrograms without
per cubic considering their
meter)
[29]. direction. It provides a straightforward interpretation of error magnitude [29]. It is
• MSE is similarexpressed
to RMSE in squared
but does notunits of thetaking
involve dependentsquarevariable
roots. (e.g., (kWh)2the
It averages or (kWh/m2 ) as
squares of thewell as many
errors, other unitslarger
emphasizing in different
errorsfields
more[29].than smaller ones. MSE
• MAPE expresses accuracy as a percentage,
measures the average magnitude of errors in a set of predictions making without
it unit-free [30]. It is useful for com-
considering
paring model performance across different buildings
their direction. It provides a straightforward interpretation of error magnitude or energy systems
[29]. with varying
It is expressed energy consumption
in squared scales
units of the and communicating
dependent variable (e.g.,results
(kWh)to2 ornon-technical
(kWh/m2) stakeholders,
as percentages are easily understood
as well as many other units in different fields [29]. [30].
• •
MAPE expresses R2 indicates
accuracy as how well the model’s
a percentage, making predictions
it unit-freematch
[30]. Itthe
is actual
useful data,
for with values
closer to 1.0 indicating a better pair [31]. R 2 allows for an easy comparison between
comparing model performance across different buildings or energy systems with
Energies 2024, 17, 4277 9 of 35

different AI models. By comparing R2 values of models with different input features,

researchers can assess which building characteristics or environmental factors have
the most significant impact on energy consumption predictions [31].
When choosing error metrics, it is crucial to consider the designed model’s specific
objectives, the data’s characteristics, and the target audience for the results. RMSE and MSE
metrics are scale-dependent, where their values are influenced by the scale of the target
variable. This can make comparing models across different buildings or energy systems
with varying energy consumption scales challenging. In practice, RMSE and MSE are often
used with metrics like MAPE and R2 to evaluate AI model performance in building energy
efficiency applications better [32]. Employing multiple metrics offers a more thorough
evaluation of model performance and aids in identifying potential limitations or areas for
improvement in AI-based building energy efficiency models [32].

3.1. Energy Consumption Forecasting

AI is widely applied to forecast energy consumption, and different methods are used in
the literature for that purpose. Table A1 in Appendix A presents a variety of AI models that
can be applied and assigned to different categories based on their underlying algorithms
and techniques. These models fall broadly into traditional machine learning, deep learning,
and hybrid models.
ML models and reliability. Traditional ML models include algorithms like SVR, Decision
Trees, and ensemble methods like Random Forests and Gradient Boosting. For example,
in reference [33], linear regression, ANN, and regression trees are used in commercial
buildings. Reference [34] also includes models like ANN, SVM, and DNN applied in
a residential building. These models are often more straightforward and require less
computational power than other DL models, making them suitable for smaller datasets
and less complex prediction tasks [35].
Deep learning models, a subset of ML, involve neural networks with multiple layers
that can capture complex patterns in data [11]. Notable examples in the table include CNN
and LSTM. Reference [36] mentions an LSTM neural network used in an educational facility,
while reference [37] applies asymmetric encoder–decoder DL algorithms. DL models, such
as those listed in references [38,39], are effective for handling large datasets, making them
ideal for accurate energy consumption predictions, with the RMSE ranging between 0.07
and 0.09 and the R2 amounting to 0.90, respectively.
Hybrid models combine traditional ML and DL elements to exploit both approaches’
strengths. These models often integrate various techniques to improve prediction accuracy
and robustness. The model in reference [40] combines ANFIS and GDFA applied to an
educational facility. Similarly, the study in reference [41] implemented a hybrid CNN with
LSTM AE used in a commercial building. By combining different methods, these models
can capture both linear and non-linear relationships in the data, providing a comprehen-
sive solution for energy prediction tasks and better accuracy, where nMAE = 0.168 and
R = 95.09% in reference [40], and MSE = 0.19, MAE = 0.31, and RMSE = 0.47 in reference [41].
Building types. Table A1 in Appendix A shows a variety of buildings. However,
residential buildings are the most common building type studied, with numerous studies
applying AI models like LSTM, DRNN, and various ensemble models, where some studies
achieve RMSE values as low as 0.1183 [42] and MAPE improvements of up to 0.54% [43].
Educational facilities are also prominently featured, employing models like LSTM, DNN,
and hybrid methods [36,44,45]. Office buildings are popular as well, with models showing
notable accuracy improvements, with MAPE values as low as 4.97% [46]. Although less
common than residential and educational types, commercial buildings receive attention
with models like DNN and DF [39,47]. Manufacturing facilities are studied less frequently.
Lastly, mosques represent a unique category with fewer studies. To summarize, residential
buildings dominate the research landscape, followed by educational and office buildings,
while manufacturing facilities and mosques are studied less frequently.
Energies 2024, 17, 4277 10 of 35

3.2. Load Forecasting

Energy load forecasting plays a pivotal role in efficient energy management. It con-
tributes to the optimization of energy production, distribution, and consumption. Accurate
energy load forecasts help in the reduction in operational costs and the improvement of
reliability [48].
AI models. AI is also used to forecast energy loads across different building types.
Table A2 in Appendix A showcases an overview of various AI models used in studies,
such as ensemble models combining ML algorithms, ANN, and DT in reference [48], as
well as LSTM, GRUs, and Bi-directional LSTM applied in references [49,50]. Additionally,
hybrid models, such as the ones presented in references [51–53], which combine XGBoost
with LSTM or CEEMDAN with Bi-LSTM, are also applied to enhance the performance of
the models in forecasting energy loads. Bio-inspired algorithms, used in references [54,55],
have also been applied in the research area. Traditional ML algorithms like RF and Gaus-
sian Radial Basis Function Kernel Support Vector Regression, used in references [32,56],
demonstrate their relevance in energy load forecasting.
Reliability and performance. The performance metrics across the studies vary, reflecting
the diverse approaches to evaluating model efficacy. MAPE has been frequently applied,
with values ranging from 0.07% in reference [57] to around 35.9% in reference [58]. This
indicates a significant variation in model performance. RMSE is another common metric
reported in several studies ranging from 0.01 [59] to over 100 kW [60].
R2 was used in references [32,52,61] with values above 0.9, suggesting high model reli-
ability. As presented in reference [62], some studies also employ specialized metrics, such
as accuracy improvement percentages, to highlight specific advantages of their approaches.
Building types. The types of buildings examined in these studies are diverse. Many
references, including [48,50,51,54], focus on residential buildings, highlighting the high
demand for efficient energy management in this sector. Educational facilities are also
examined often, as shown in references [49,62–64], exploring models applied in schools
and universities. Studies in references [32,58] address commercial and office buildings.
Hotels, public buildings, and hospitals are also included, as seen in references [52,65–67],
demonstrating the adaptability of AI models to diverse building types.

3.3. HVAC Control and Optimization

The control of HVAC systems is critical for maintaining indoor comfort and regulating
temperature, humidity, and air quality in buildings while minimizing energy consump-
tion [68]. Optimizing these systems can significantly reduce energy usage and operational
costs, contributing to environmental sustainability [69]. Integrating AI models in HVAC
control presents a promising advancement, allowing for more precise and adaptive energy
management [68].
AI models. Table A3 in Appendix A presents various studies that have utilized different
AI models for HVAC and their reliability across different building types.
The studies feature many AI models in references [68,70–75], such as LSTM, DRL,
FIS, AMADRL, YOLOv5, SVM, RF, and DNN Bilinear Koopman Predictor. This variety
illustrates the potential of AI in enhancing HVAC system efficiency, with each model
bringing distinct advantages.
Reliability and performance. The performance of these models is evaluated using mul-
tiple metrics, including RMSE, MSE, and energy savings percentages. In reference [70],
LSTM and DRL achieved an MSE of 0.0015 and energy savings from 27% to 30%. Shallow
ANN models in reference [76] demonstrated improvements in energy consumption and
thermal comfort, with heating energy consumption reductions ranging from 0.6% to 29%
and thermal comfort improvements of up to 58.8%. YOLOv5 in reference [73] achieved an
accuracy of 88.1%, while the ensemble approach in reference [74] demonstrated reductions
in natural gas consumption (22.2%) and building heating demand (4.3%), with an RMSE
value of 32.1 kW.
Energies 2024, 17, 4277 11 of 35

Building types. The studies cover a range of building types. Educational facilities are
prominently featured, with models like Shallow ANN [76], ANN [77], YOLOv5 [73], and
the ensemble approach [74] showing energy and thermal comfort improvements. Models
like DRL [78] and DQN [79] have been applied to residential buildings, enhancing PM2.5
levels and overall energy consumption.
AI models applied in offices also showed promising results, as presented in refer-
ences [80,81]. Additionally, specialized buildings such as sports halls [82] and churches [71]
were considered, demonstrating the versatility of AI applications in different building
environments.

3.4. Occupant Detection

Occupant detection contributes to reducing energy consumption in buildings. Various
technologies are used to identify and detect people’s presence, number, and activities in
a building. This information is important in optimizing building energy management
systems by ensuring that resources are used efficiently [83].
AI models. Table A4 in Appendix A presents various AI models applied for occupant
detection across different types of buildings, such as CNNs, DMFF, YOLO, LTSM net-
works [83–86], and other advanced machine learning techniques like 1D CNN and RL [87],
as well as traditional methods like MLR [88]. Each model demonstrates certain strengths in
terms of accuracy and performance.
Reliability and performance. The performance of these models is evaluated using differ-
ent metrics, such as accuracy, RMSE, MAE, MAPE, NRMSE, and correlation coefficients.
Metrics specific to occupant detection have also been used, like thermal comfort improve-
ment and CO2 levels. Accuracy is the most common across the studies, where the DMFF
model [84] achieves a high accuracy of 97%, while Faster R-CNN variants achieve ac-
curacies between 78.39% and 98.9% [89,90]. Energy savings is also an important metric
highlighting AI models’ potential benefits. YOLOv5, for example, shows annual HVAC and
lighting energy savings of 10.2% [85], while DMFF reports up to 30% energy savings [84].
RMSE, MAE, and MAPE are used to measure prediction errors. The YOLOv4 model in
office settings has an RMSE of 0.883 and an NRMSE of 0.141, indicating high precision in
maintaining indoor CO2 levels [91]. The GA-LSTM and PSO-LSTM models exhibit high
correlation coefficients (99.16–99.97%), indicating strong predictive capabilities [86].
Building types. The models presented in the studies have been applied across dif-
ferent building types, including residential, office, and educational facilities. A range of
models, including CNN, YOLOv5, Faster R-CNN, and LM-BP, have been employed in
offices [83,90,92,93]. In educational buildings, Faster R-CNN demonstrates high people-
counting accuracy of 98.9% and activity detection of 88.5% [94].

3.5. Other Areas of Application

Table A5 in Appendix A presents various studies implementing AI models in different
applications in buildings, such as thermal comfort prediction, air quality prediction, and
indoor temperature prediction. The discussion below addresses the AI models utilized, the
performance metrics reported, and the types of buildings considered in these studies.
AI models. The studies employ a wide array of AI models, such as ANN and SVM,
which are used for their robustness in handling non-linear relationships in the study [95]
for thermal comfort prediction. LTSM and RNN, suitable for time-series predictions, are
applied in the study [96]. Hybrid models combining multiple techniques are also utilized,
such as CNN-GRU-MLP [97] and FL-BM-ANFIS-BM [98]. Models such as Radial Basis
Function Networks [99] and GNN [100] were applied for different prediction tasks.
Reliability and performance. The studies in Table A5 in Appendix A report a variety of
metrics that are crucial for evaluating AI models. High R2 values, such as 0.976 and 0.981 in
reference [97], demonstrate excellent model performance. Reference [101] reports an MSE
of 0.04, indicating high precision in temperature prediction. RMSE measures prediction
error magnitude, with values like 0.705 in reference [97] indicating reliable performance.
 Energies 2024, 17, 4277 12 of 35

PPD reflects the practical impact of the model, such as a 43.7% cooling load reduction in
reference [98].
Building types. AI models’ effectiveness varies across building types due to varying
environmental conditions and usage patterns. According to Table A5 in Appendix A,
residential buildings focus on thermal comfort and air quality. Reference [95] employs ANN
and SVM for thermal comfort prediction, achieving an MSE of 0.8179, and reference [96]
uses 1D-CNN, RNN, and LSTM for air quality prediction, reporting an RMSE of 10 ppm.
Indoor temperature and air quality are important for occupant productivity in offices;
reference [102] uses CNN-LSTM, achieving an R2 of 0.936 and 50% energy savings, and
GNN models, used in reference [100], improve thermal comfort by up to 81.3%. Models
like FL-BM and ANFIS-BM [98] in educational facilities report significant improvements in
thermal comfort and cooling load reduction. The MLP model [103] also achieves errors as
low as 0.069 for temperature predictions.

4. Discussion
The main aim of this study was to determine what AI models are used in BEMS and
how they contribute to energy savings. Tables A1–A5 show a wide range of different AI
models and combinations of multiple AI models applied to enhance and optimize building
energy efficiency. The reliability metrics of the AI models prove that AI-driven tools play a
significant role in improving building energy management.
Reliability is the model’s ability to predict parameters that control the system. In
Tables A1–A5, different indicators are used to measure the reliability, such as the Root Mean
Square Error (RMSE), R2 , and Mean Square Error (MSE), the performance of the AI models,
and their impact on energy savings, cost reductions, and thermal comfort improvements.
An analysis of papers on AI models used in BEMS shows that the most commonly
applied topics include the following focus areas: error rate, energy savings, accuracy,
performance, and cost reduction. Figure 5 shows the different criteria papers used to
evaluate the reliability of AI models. The predominant metric is the error rate, accounting
for 63.3% of the evaluations. This indicates a strong emphasis on minimizing errors to
enhance the reliability of AI models. Although, at 16.5%, energy savings is the next most
s 2024, 17, x FOR PEER REVIEW significant factor, it is still relatively low, especially when the study’s main purpose 13 ofis 35
to reflect how the model applied contributes to energy savings. Accuracy, at 7.6%, and
performance, at 6.3%, show that different papers use different metrics to evaluate the AI
models, making it slightly difficult to compare the models in the studies. This indicates the
need for a unified dataset for comparison.

6% Error Rate
6%
8% Energy Savings

Accuracy
17%
63%
Performance

Cost Reduction

Figure 5. Presentation of the focus areas of the study results.

Figure 6ofshows
Figure 5. Presentation the number
the focus areas ofofthe
studies
studyconducted
results. in different countries, emphasizing
the top countries contributing to the field. China leads significantly with nearly 35 studies.
South Korea and the USA follow, with about 15 studies each. The UK comes next with
Figure 6 shows the number of studies conducted in different countries, emphasizing
the top countries contributing to the field. China leads significantly with nearly 35 studies.
South Korea and the USA follow, with about 15 studies each. The UK comes next with
around 10 studies, and France has approximately 8. These data highlight China’s domi-
nant role in this research area and the other countries mentioned as key players in advanc-
 Figure 5. Presentation of the focus areas of the study results.

Figure 6 shows the number of studies conducted in different countries, emphasizing

the
Energies 2024, 17,top
4277 countries contributing to the field. China leads significantly with nearly 35 studies. 13 of 35
South Korea and the USA follow, with about 15 studies each. The UK comes next with
around 10 studies, and France has approximately 8. These data highlight China’s domi-
around 10
nant role in this research areastudies, and other
and the Francecountries
has approximately 8. These
mentioned dataplayers
as key highlightinChina’s
advanc-dominant
ing this field of study. This also indicated that the amount of research in Europe needs to this
role in this research area and the other countries mentioned as key players in advancing
field of study. This also indicated that the amount of research in Europe needs to increase
increase to achieve the Paris Agreement goals.
to achieve the Paris Agreement goals.

Figure
Figure 6. The number 6. The number
of studies of studies
conducted conducted per country.
per country.
Figure 7 shows the distribution of different building types for various application
Figure 7 shows theEducational
areas. distribution of different
facilities building
and residential typesarefor
buildings thevarious application
most common, especially for
areas. Educationalforecasting
facilities and
energyresidential
consumption buildings
and HVACarecontrol/optimization.
the most common,Offices especially for
and commercial
forecasting energybuildings
consumption
are also and HVAC
popular in control/optimization.
several application areas.Offices
Public and commercial
buildings, hospitals, and
buildings are alsosports halls in
popular mainly focusapplication
several on energy consumption forecasting
areas. Public and air
buildings, quality. Churches
hospitals, and and
mosques are less frequently studied.
sports halls mainly focus on energy consumption forecasting and air quality. Churches
Figure 8 shows the number of papers dedicated to each application area. Energy
and mosques are less frequently studied.
consumption forecasting is the area researched the most, with 57 papers covering it. HVAC
control/optimization follows with 37 papers, highlighting a significant focus on improving
building efficiency. At 30 papers, load forecasting is also relatively high, followed by
occupancy detection, with 25 papers. Indoor temperature prediction, air quality, and
thermal comfort are less frequently studied. Therefore, these papers were systematically
analyzed to determine how AI models used in BEMS contribute to energy savings, cost
reductions, and thermal improvements. The results of this analysis are presented in Table 2.
As shown in Table 2, the highest energy-savings potential (of up to 37%) can be found
in offices when AI models are used for HVAC control and optimization, as demonstrated
by Wang et al. [104] in their study. The authors developed a DRL-based HVAC control
algorithm that optimized the thermal comfort and energy efficiency of an open-plan office
with a multi-VAV HVAC system. Using AI models for HVAC control in offices can also
reduce costs by up to 14.5%.
Compared to offices, residential buildings can achieve higher cost reductions of up
to 24.29%. However, the energy savings are smaller (up to 23% in residential and 21% in
educational buildings), likely due to the lower level of intelligence of existing building
management systems or installed baseline controllers of the HVAC equipment, as indicated
by An and Chen [79] and Chemingui et al. [68].
 35
30
Energies 2024, 17, x FOR PEER REVIEW 14 of
Energies 2024, 17, 4277
25
Number of studies 14 of 35
20
15
50
10
45
5
40
0
35
30
25
Number of studies

20
15
10 Building Type
5
Thermal Comfort Occupant Detection
0
Load Forecast Indoor Temperature

HVAC Control/Optimization Air Quality

Energy Consumption Forecast

Figure 7. The distribution of building types for each area of application.

Building Type
Figure 8 shows the number of papers dedicated to each application area. Energy con-
sumption forecasting is the area researched
Thermal Comfort
the most, with 57 papers covering it. HVAC
Occupant Detection
control/optimization follows with 37 papers, highlighting a significant focus on improv-
Loadefficiency.
ing building Forecast At 30 papers, load Indoor Temperature
forecasting is also relatively high, followed by
occupancy detection, with
HVAC Control/Optimization25 papers. Indoor temperature
Air Quality prediction, air quality, and ther-
mal comfort are less frequently studied. Therefore, these papers were systematically ana-
Energy Consumption Forecast
lyzed to determine how AI models used in BEMS contribute to energy savings, cost re-
ductions,Figure
and 7.thermal improvements.
The distribution The results
of building types of this
for each area analysis are presented in Table 2.
of application.
Figure 7. The distribution of building types for each area of application.
Demand Response
Figure 8 shows the number 60 of papers dedicated to each application area. Energy co
sumption forecasting is the area
50 researched the most,
Energy with 57 papers covering it. HVA
Consumption
Air Quality
40 37 papers, highlighting
control/optimization follows with Forecast a significant focus on impro
30
ing building efficiency. At 30 papers, load forecasting is also relatively high, followed
20
occupancy detection, with 25 papers. Indoor temperature prediction, air quality, and the
10
mal comfort
Thermal are less frequently0studied. Therefore, theseHVAC
Comfort papers were systematically an
Control/Optimization
lyzed to determine how AI models used in BEMS contribute to energy savings, cost r
ductions, and thermal improvements. The results of this analysis are presented in Table

Indoor Temperature Demand Response Load Forecast

60
50 Energy Consumption
Air Quality
Occupant Detection
40 Forecast
30
Figure 8. The number of papers per application area.
20
10 HVAC
Thermal Comfort 0
Control/Optimization

Indoor Temperature Load Forecast

Energies 2024, 17, 4277 15 of 35

Table 2. The benefits of using AI models to save energy, reduce costs, and improve thermal comfort.

Cost Thermal
AI Models Building
Energy Savings, % Reductions, Comfort
Applications Type
% Increase, %

Energy Office 17.4 - 16.9

consumption median of 57.38% in air
forecasting Commercial - -
conditioning system
Residential 5–23 6.1–24.29 16

HVAC control Office 5–37 14.5 -

and 21
optimization Educational - -
0.6–29 in heating energy
Commercial 10 - -
Residential 30 - -
Occupancy
detection 2.3–8.1
Office - 43–73
10.2 in HVAC and lighting energy

An and Chen [79] used a DRL algorithm to develop a deep-Q network controller that
controls windows, air cleaners, and air conditioners to reduce indoor air pollution and
maintain thermal comfort with relatively low energy consumption in residential buildings.
Similarly, Chemingui et al. [68] applied a DRL framework to control a school building’s
indoor environmental conditions and optimize energy consumption. Their technique,
enhanced with behavioral cloning, was designed to find optimal HVAC control decisions
for different weather conditions throughout the year, minimizing energy consumption,
maintaining thermal comfort, and reducing indoor contaminant levels in a multi-zone
environment.
Each of these studies demonstrated significant improvements in energy efficiency and
environmental conditions by using DRL to control and optimize different HVAC systems
across various building types—offices, residential buildings, and schools. Therefore, DRL-
based methods are well suited for achieving optimal HVAC control strategies and balancing
the trade-offs between building indoor comfort and energy consumption.
The use of AI models to predict energy consumption also shows potential energy
savings of up to 17.4% and improved thermal comfort of up to 16.9% in offices. However,
the results of the papers analyzed show that the integration of AI models for HVAC control
and optimization is more efficient and gives better energy savings potential.
Another promising and efficient application of AI models is occupancy detection,
which can deliver relatively high energy savings of up to 8.1% in offices and improve
thermal comfort by 43% to 73%.
In conclusion, the integration of AI models into BEMS is an effective solution for
ensuring higher energy efficiency in buildings while maintaining a high level of comfort, if
the existing BEMS is open-source and the existing controllers are programmable.

4.1. Estimation of AI Model Reliability

In this review, we aimed to compare the various AI models employed to enhance
energy efficiency in buildings, focusing on their accuracy. To achieve this, we classified the
error metrics as low, medium, and high. However, the comparison posed a challenge due
to the nature of different error metrics. Unlike R2 , which ranges from 0 to 1 and provides
a standardized measure of model accuracy, other metrics like RMSE, MAPE, MAE, and
MSE depend on the data’s scale and distribution, making direct comparison difficult. These
metrics can vary significantly in magnitude and units, thus complicating the process of
establishing a uniform classification.
For instance, RMSE is sensitive to large errors, while MAE provides a linear per-
spective on errors. This makes RMSE and MAE values difficult to interpret uniformly
Energies 2024, 17, 4277 16 of 35

across different datasets and models. In building energy management, the variability
in building types, sizes, and energy usage patterns intensifies these issues, as noted by
Ahmad et al. [105]. Moreover, these error metrics do not consider factors unique to building
energy efficiency implementation, such as retrofit schedules, occupancy patterns, and
renewable energy integration. As a result, the effectiveness of one model in one building or
scenario may not be directly comparable to another, even when using the same error metric,
which Zhao et al. [106] further elaborate on. Therefore, because of its straightforward
and standardized scale, we used R2 as the primary metric for comparing the AI model’s
accuracy in the diverse and complex field of building energy management. In this review
paper, R2 values were classified as follows: high accuracy (1.00–0.66), medium accuracy
(0.65–0.36), and low accuracy (0.00–0.35).
The highest reported accuracy was selected for papers that presented a range of
accuracies. From the studies reporting R2 in Table A6 in Appendix B, the R2 values range
from 0.99911, the highest achieved by a GA model for energy consumption forecasting
in residential buildings, to the lowest value with medium accuracy, according to our
assessment criteria, of 0.4872, achieved by an ANN model for thermal comfort prediction
in residential buildings. The most common area of application reported in Table A6 in
Appendix B is the energy consumption forecast, with various AI models being applied
across different building types. Other high R2 values in this application area include RNN
with R2 values of up to 0.999 and the hybrid DNN-LSTM model with an R2 of 0.9991.
Residential buildings are predominant; several AI models were applied to residential
buildings for energy consumption forecasting and load forecasting, indicating a significant
focus on this building type in these application areas. Followed by offices and educational
buildings, Table A6 in Appendix B highlights that DL, like DNNs, CNNs, and hybrid
models, is highly effective in predicting energy consumption, with R2 values frequently
exceeding 0.9, indicating high accuracy.

4.2. Limitations
Despite the comprehensive approach in conducting this systematic literature review,
it is important to acknowledge certain limitations. The methodology employed may not
have captured all relevant papers in the field of AI models for building energy efficiency.
Though we aimed to be thorough, the effectiveness of our search depended heavily on
the chosen keywords and search strings, meaning some relevant studies may have been
omitted. Differences in terminology and indexing across various databases also added to
this challenge. As a result, our review provides valuable insights but represents only a
portion of the available research in this domain. Future studies should consider expanding
the search criteria and incorporating additional sources to build on this review.

5. Conclusions
This study is designed to contribute and complement existing research in the area of
AI models used in BEMS and their impact on energy savings. The review highlights several
aspects regarding the evaluation and application of AI models in various areas of buildings.
First, the lack of standardized metrics for assessing AI model reliability complicates the
comparison of different studies. This highlights the need for a unified dataset for more
meaningful comparisons. Additionally, many studies do not report savings as error metrics,
which is crucial for understanding the practical impact of these models. While other error
metrics are used, translating these into actual savings is essential for evaluating the models’
effectiveness. Furthermore, the extent of research in Europe is relatively limited compared
to that in the USA and China. To meet the Paris Agreement sustainability goals, Europe
must increase its research efforts in this field. Integrating different AI algorithms in model
design is a popular way to go, which indicates that a combination approach performs
better. Moreover, the review shows that energy consumption and load forecasting are
the most common application areas, whereas air quality receives the least attention. This
distribution highlights the need for a broader focus on diverse application areas to achieve
Energies 2024, 17, 4277 17 of 35

comprehensive advancements in AI-driven sustainability. The main findings of this study

in relation to the research questions are as follows:
1. The use of AI models in BEMS for energy consumption forecasting, HVAC control and
optimization, occupancy detection, and the prediction of indoor climate parameters is
a valuable contribution to building energy efficiency, additional energy savings, cost
reductions, and thermal improvements.
2. The highest energy savings potential of up to 37% can be found in offices, smaller
savings of up to 23% can be found in residential buildings, and savings of 21% can be
found in educational buildings when DRL-based models are used to optimize HVAC
control strategies and balance the trade-offs between indoor comfort and energy
consumption, compared to baseline rule-based methods.
3. AI models, particularly deep learning architectures like DNNs, CNNs, and hybrid
models, are highly effective in predicting energy consumption, with R2 values fre-
quently exceeding 0.9, indicating high accuracy. The most common application area is
energy consumption forecasting, with residential buildings being a predominant focus.

Author Contributions: Conceptualization, D.M.T.E.A. and V.M.; methodology, D.M.T.E.A.; formal

analysis, D.M.T.E.A. and R.D.-T.; resources, D.M.T.E.A. and V.M.; writing—original draft prepara-
tion, D.M.T.E.A., V.M. and R.D.-T.; writing—review and editing, V.M. and R.D.-T.; visualization,
D.M.T.E.A., V.M. and R.D.-T.; supervision, V.M. All authors have read and agreed to the published
version of the manuscript.
Funding: The research was co-funded by the European Union under Horizon Europe program grant
agreement No. 101059903, by the European Union funds for the period 2021–2027, and by the state
budget of the Republic of Lithuania financial agreement Nr. 10-042-P-0001.
Data Availability Statement: No new data were created or analyzed in this study. Data sharing is
not applicable to this article.
Conflicts of Interest: The funders had no role in the design of the study; in the collection, analyses,
or interpretation of the data; in the writing of the manuscript; or in the decision to publish the results.

Abbreviations

AE Autoencoder
AI Artificial Intelligence
AMADRL Asymmetric Multi-Agent Deep Reinforcement Learning
AN Artificial Neural
ANFIS Adaptive Neuro-Fuzzy Inference System
ARIMA AutoRegressive Integrated Moving Average
BBO Biogeography-Based Optimization
BDQ Big Data Query
BEMS Building Energy Management Systems
Bi-GRU Bidirectional Gated Recurrent Unit
Bi-LSTM Bidirectional Long Short-Term Memory
Complete Ensemble Empirical Mode Decomposition with
CEEMDAN
Adaptive Noise
CIFG Coupled Input Forget Gate
CNN Convolutional Neural Network
ConvLSTM2D Convolutional Long Short-Term Memory 2D
CV Cross-Validation
CVRMSE Coefficient of Variation of the Root Mean Squared Error
DBN Deep Belief Network
DF Decision Forest
DFNN Deep Feedforward Neural Network
DRL Deep Reinforcement Learning
Energies 2024, 17, 4277 18 of 35

DNN Deep Neural Network

DQN Deep Q-Network
DRLC Deep Reinforcement Learning Control
DRNN Deep Recurrent Neural Network
DRNN-GRU Deep Recurrent Neural Network-Gated Recurrent Unit
DRL Deep Reinforcement Learning
DUMSL Deep Unsupervised Multilayer Stacking
Deep Unsupervised Multilayer Stacking Learning-Deep
DUMSL-DNN
Neural Network
DT Decision Tree
EPBD Energy Performance of Buildings Directive
EMD Empirical Mode Decomposition
FF Feed-Forward
FFNN Feed-Forward Neural Network
FIS Fuzzy Inference System
FL-BM Fuzzy Logic-Based Model
GA Genetic Algorithm
GAN Generative Adversarial Network
GB Gradient Boosting
GDFA Generalized Dynamic Fuzzy Automata
Generalized Fuzzy Systems with Fuzzy
GFS.FR.MOGUL
Regression-Modified Global Universe of Discourse
GMTCN Gated Memory Time Convolutional Network
GNN Graph Neural Network
GPR Gaussian Process Regression
GRU Gated Recurrent Unit
GRU-RL Gated Recurrent Unit-Reinforcement Learning
HHT Hilbert–Huang Transform
Harris Hawks Optimization-Adaptive Neuro-Fuzzy
HHO-ANFIS
Inference System
HVAC Heating, Ventilation, and Air Conditioning
HyFIS Hybrid Fuzzy Inference System
IEA International Energy Agency
IoT Internet of Things
IPWOA Improved Particle Whale Optimization Algorithm
KNN K-Nearest Neighbors
LR Linear Regression
LSSVR Least Squares Support Vector Regression
LSTM Long Short-Term Memory
MAPE Mean Absolute Percentage Error
MAE Mean Absolute Error
MAQMC Multi-Agent Quantum Monte Carlo
Metaheuristic-based LSTM Metaheuristic-based Long Short-Term Memory
ML Machine Learning
MLR Multiple Linear Regression
MSE Mean Square Error
MR Multiple Regression
Nonlinear Autoregressive with Exogenous
NARX-MLP
Inputs-Multilayer Perceptron
NRMSE Normalized Root Mean Square Error
nMAE Normalized Mean Absolute Error
NZE Net Zero Emissions
OBC Optimal Bayesian Control
PMV Predicted Mean Vote
PPO Proximal Policy Optimization
PPD Predicted Percentage Dissatisfied
PSO Particle Swarm Optimization
Energies 2024, 17, 4277 19 of 35

R Correlation Coefficient
R2 Coefficient of Determination
RBFNN Radial Basis Function Neural Network
RF Random Forest
RL Reinforcement Learning
RMSE Root Mean Square Error
RNN Recurrent Neural Network
SADLA Self-Attentive Deep Learning Algorithm
Seq2Seq Sequence to Sequence
SNNs Spiking Neural Networks
SOS Symbiotic Organisms Search
SPSA Simultaneous Perturbation Stochastic Approximation
ST-GCN Spatio-Temporal Graph Convolutional Network
STLF Short-Term Load Forecasting
SVR Support Vector Regression
SVM Support Vector Machine
TST Temporal Self-Tracking
VSCA Very Short-term Climate Anomaly
WM Wavelet Model
YOLO You Only Look Once

Appendix A. AI Models Used for Different Applications

Table A1. Comparison of AI models used for energy consumption forecasting.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
Linear regression, ANN,
[33] Commercial - Best results with MAPE = 1%
Regression trees
- DNN: R2 = 0.95, RMSE = 1.16
ANN, GB, DNN, RF, Stacking,
[34] Residential - ANN: R2 = 0.94, RMSE = 1.20
KNN, SVM, DT, LR
- GB: R2 = 0.92, RMSE = 1.40
- Daily energy consumption forecast MAPE reduction
compared to ARIMA = 11.2%, Hourly = 16.31%.
[36] LSTM neural network Educational Facility
- Daily energy consumption prediction MAPE reduction
compared to BP = 49%, Hourly = 36.6%
- Single-step forecasting average R2 = 0.964
Asymmetric encoder-decoder
[37] Educational Facility - Three-step ahead multi-step forecasting average
DL algorithm
R2 = 0.915
LSTM, Bidirectional LSTM, Energy savings = 17.4%
[38] CNN, Attention Mechanism, Office Thermal comfort improvement = 16.9%
Soft Actor-Critic, RL - RMSE = 0.07–0.09
[39] DF Commercial R2 = 0.90
- ANFIS-SC: nMSE = 49.16, nMAE = 0.452, R = 58.71%.
- ANFIS-FCM: nMSE = 53.48, nMAE = 0.517,R = 56.44%
[40] ANFIS, GDFA Educational Facility
- AR-ANFIS-GDFA-SC+: nMSE = 7.25,
nMAE = 0.168, R = 95.09%
- MSE = 0.19
[41] Hybrid CNN with LSTM-AE Commercial - MAE = 0.31
- RMSE = 0.47
VSCA, ConvLSTM2D model - MSE = 0.0140
[42] with Conv2D attention Residential - RMSE = 0.1183
mechanism and roll padding - MAE = 0.0875
Energies 2024, 17, 4277 20 of 35

Table A1. Cont.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
[43] LSTM Office - MAPE improvement = 0.54%
Deep learning autoencoder
[44] Educational Facility CV(RMSE) < 9%
coupled with LSTM
- DNN R2 = 0.87
- DNN CV-RMSE = 24.4%
- GB CV-RMSE = 26.5%
MR, RF, ANN-FF, SVR, GB,
[45] Educational Facility - SVR CV-RMSE = 26.5%
DNN
- ANN-FF CV-RMSE = 27.9%
- RF CV-RMSE = 35.3%
- MR CV-RMSE = 39.4%
- MAE = 4.4–21.4 W,
Adaptive decomposition, - MAPE = 4.97–21.97%
[46] Residential
multi-feature fusion RNNs - RMSE = 8.8–37.8 W
- R2 = 0.974–0.999
- Energy savings:
21-layer Fully Connected Median = 57.38%, Maximum = 90%
[47] Commercial
DNN - Energy consumption prediction(test): RMSE = 213 W,
R2 = 0.72, MAPE = 15.1%
-Training dataset accuracy = 89.03%, Standard
error = 0.3
[107] SOM, CNN, GA Public building
- Validation dataset accuracy = 88.91%, Standard
error = 0.33
- Compared to traditional models, DDPG and RDPG
performed better in
[108] A3C, DDPG, RDPG Office
Single-step prediction = 16–14%,
Multi-step prediction = 19–32%.
- Energy consumption prediction accuracy:
DFNN = 92.4%, DRNN = 96.8%
- Air temperature accuracy:
[109] DFNN, DRNN Manufacturing Facility
DFNN = 99.5%, DRNN = 99.4%
- Humidity accuracy:
DFNN = 64.8%, DRNN = 57.6%
- MAPE = 4.5%
[110] CNN Mosque
- R2 = 0.98
PSO, Particle Swarm, Stacking
[111] Educational Facility RMSE = 1.71 lower than that of common ML algorithms.
ensemble model. PFS
[112] SVR Educational Facility -R2 = 0.92
- MAPE = 0.05–0.09
Metaheuristic-based LSTM - MAE = 0.04–0.07
[113] Residential
network - RMSE = 0.13–0.16
- MSE = 0.04–0.05
- Daily model:
RMSE = 0.362, MAE = 19.7%
[114] LSTM Residential
- Monthly model:
RMSE = 0.376, MAE = 17.8%
-SVM MAPE = 7.19%
- WM MAPE = 8.58%
ANN, SVM. HyFIS, WM,
[115] Office - HyFIS MAPE = 8.71%
GFS.FR.MOGUL
- ANN MAPE = 10.23%
- GFS.FR.MOGUL MAPE = 9.87%
Energies 2024, 17, 4277 21 of 35

Table A1. Cont.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
[116] HHT, RegPSO, ANFIS Educational Facility - MAPE = 1.91%
Bidirectional LSTM, stacked
unidirectional LSTM, and - RMSE = 0.0047
[117] Residential
fully connected layers - R2 = 0.998
optimized DTO
LSTM, NARX-MLP, GRU, DT,
[118] Educational Facility - Best model RMSE = 0.23
XGBoost
[119] Adaboost-BP Residential - Average prediction accuracy = 86%
[120] MgHa-LSTM Not Specified - MSE = 0.2821
- RNN MSE = 0.00279
- LSTM MSE = 0.00571
RNN, LSTM, GRU, TST,
[121] Residential - GRU MSE = 0.00483
Ensemble
- TST MSE = 0.00771
- Ensemble MSE = 0.00289
- RMSE = 0.44 kWh
[122] DRNN Residential
- MAE = 0.23 kWh
- Best MAPE = 3.51%
[123] LSTM, GRU, EMD Hospital
- Best RMSE = 55.06kWh
- R2 = 0.9917
[124] GPR Public Building
- CV-RMSE = 0.1035
- Air conditioning prediction:
[125] LSTM Office MSE = 519.77, CV-RMSE = 0.1349,
MAE = 14.52
- LSTM RMSE = 0.0693
[126] LSTM, CNN Residential
- CNN RMSE = 0.0836
[127] SADLA Office SADLA highest R2 = 0.976
LR, SVM, RF, MLP, DNN, - One month ahead prediction: R2 = 88%
[128] Educational Facility
RNN, LSTM, GRU - Three months ahead prediction: R2 = 81%
Proposed eight-layer deep - R2 = 97.5%
[129] Residential
neural network - RMSE = 111 W
- Lowest RMSE = 0.5207
[130] DUMSL-DNN Residential
- Lowest MAE = 0.3325
- Compared to DDPG, the proposed DF-DDPG
method decreased
MAE by 7.15%
[131] DRL, DDPG, DF Office
MAPE by 12.71%
RMSE by 18.33%
Increased R2 by 1.3%
[132] DNN with Stacked Boosters Office NRMSE = 2.35%
- RMSE decreased by 3.06%
A-LSTM, LSTM, RNN, DNN,
[133] Educational Facility - MAE decreased by 6.54%
SVR
- R2 increased by 0.43%
- MAE = 0.015
[134] IILSTM Public Building
- RMSE = 0.109
[135] Vanilla LSTM Residential Best RMSE = 4.4776
Energies 2024, 17, 4277 22 of 35

Table A1. Cont.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
- RMSE = 36.31 kWh
- MAE = 29.45 kWh
[136] LSSVR, RBFNN, SOS Residential
- MAPE = 8.90%
- R2 = 0.93
- R2 = 98.45%
[137] EDA-LSTM Office - RMSE = 4.02
- MAE = 2.87
- IHEPC Dataset:
RMSE = 0.42, MSE = 0.18, MAE = 0.29
[138] CNN, GRU Residential
- AEP Dataset:
RMSE = 0.31, MSE = 0.10, MAE = 0.33
- MAPE = 5.37%
[139] BiGTA Educational Facility
- RMSE 171.3 kWh
- MSE = 0.0095
- RMSE = 0.0974
[140] kCNN-LSTM Educational Facility
- MAE = 0.0711
- MAPE = 0.2697
MAPE: Training = 1.43%, Testing = 4.83%
[141] DNN, GA Office R2 : Training = 0.993, Testing = 0.960
RMSE: Training = 4.33 kW, Testing = 10.29 kW
-RMSE = 0.6170
[142] CNN Residential - MSE = 0.3807
- MAE = 0.4490
[143] DBN, ELM Not Specified Improved accuracy by ~20%
- MAE = 1.30
[144] EWKM, RF, SSA, BiLSTM Public Building - RMSE = 1.63
- MAPE = 0.02
- CNN-GRU daily MAE = 0.151
SVR, LSTM, GRU, - CNN-GRU hourly MAE = 0.229
[145] Residential
CNN-LSTM, CNN-GRU - LSTM daily MAE = 0.183
- LSTM hourly MAE = 0.228
- Improved R2 by 10%
[146] VMD, LSTM Office - Decreased MAE by 48.9%
- Decreased RMSE by 54.7%
- R2 = 0.99911
- RMSE = 0.02410
[147] Hybrid DNN-LSTM Residential
- MAE = 0.01565
- MAPE = 0.01826

Table A2. Comparison of AI models used for load forecasting.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
- RMSE = 2.8735 and 4.7721.
[32] RF, ELM, IPWOA Commercial
- MAPE = 0.2% and 0.45%.
[48] Ensemble, ML, ANN, DT Residential - MAPE = 5.39%
Energies 2024, 17, 4277 23 of 35

Table A2. Cont.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
- LSTM RMSE = 0.0600–0.7527 kW
- Bi-LSTM RMSE = 0.0430–0.3960 kW
- GRU RMSE = 0.0413–1.3805 kW
[49] LSTM, Bi-LSTM, GRU Educational Facility
- LSTM MAE = 0.0003–0.0078 kW
- Bi-LSTM MAE = 0.0005–0.0041 kW
- GRU MAE = 0.0005–0.0144 kW
-RMSE = 0.510
[50] DRNN-GRU Residential - MAE: 0.345
- MAPE: 3.504%
- Minimum MAAPE = 18.70%
[51] BP, XGBoost, LSTM Residential - Maximum MAAPE = 45.95%
- Average MAAPE = 31.20%
- MAPE reduced by 27.48%, 14.05%, and 13.38% for
GMTCN, Bidirectional LSTM. 1-step, 6-step, and 12-step predictions, respectively.
[52] Hotel
SPSA - R2 = 0.971, 0.923, and 0.885 for 1-step, 6-step, and
12-step predictions, respectively.
Best model (CEEMDAN-Bi-LSTM-ARIMA):
CNN, LSTM, Bi-LSTM, GRU, - R2 = 0.983
[53] Educational Facility
CEEMDAN, ARIMA - RMSE = 70.25 kWh
- CV-RMSE = 1.47%
- Heating load training, MAE = 2.15.
[54] BBO Residential
- Cooling load training, MAE = 2.97
- Heating load R2 = 0.94
[55] BBO Residential - Cooling load R2 = 0.99
- Heating and cooling RMSE = 0.148–0.149
Gaussian radial basis function
[56] kernel support vector Residential - Heating and cooling load prediction MAE = 4% less.
regression
[57] LSTM Residential - MAPE = 0.07
Average MAPE reduction of 29.7%, 32.8%, 35.9%, and
[58] CNN Office 25.3% compared to that of GRU, ResNet, LSTM, and
GCNN, respectively.
- RMSE = 0.01
[59] TRN Office - MAE = 0.03
- R2 = 0.98
- RMSE = 44.28 MW
CNN-BiGRU and PSO - MAPE = 3.11%
[60] Residential
optimization - MAE = 29.32 MW
- R2 = 0.9229
- R2 = 98%
[61] HHO-ANFIS Residential
- RMSE =0.08281
- Accuracy improvement = 20–45%
[62] BiLSTM, LSTM, CNN Educational Facility
- RMSLE = 0.03 to 0.3
-MAE = 40.8411
iCEEMDAN-BI-LSTM hybrid - RMSE = 59.6807
[63] Educational Facility
model - MAPE = 2.56%
- R2 = 0.9869
- XGBoost CVRMSE = 21.1% on test set,
[64] XGBoost, LSTM Educational Facility
- LSTM CVRMSE = 20.2%
[65] LSTM, CIFG, GRU, ANN Public Building - Most accurate RMSE = 0.770
Energies 2024, 17, 4277 24 of 35

Table A2. Cont.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
[66] FFNN Hospital - MAPE = 6.6–7.0%
[67] 1D-CNN, Seq2Seq Hotel - MAPE = 10% less
- MAPE = 1.70%, 1.77%, 1.80%, and 1.67% for the
[148] ANFIS, BGA-PCA Residential
summer, fall, winter, and spring seasons, respectively.
- Heating load, R2 = 0.999
[149] 3RF Not Specified
- Cooling load, R2 = 0.997
- Error rate reduction over the IHEPC dataset:
MAE = 15.6
MSE = 8.77%
[150] CNN, LSTM Residential
RMSE = 4.85%
- Error rate reduction over the PJM dataset:
RMSE = 3.4%
[151] DRL, DDPG. TD3 Not Specified - Error = 4.56%
LSTM, Bi-LSTM, GRU,
[152] Railway Station - Best MAPE = 0.2%
Bi-GRU
[153] CNN-LSTM, EMD, Bayesian Residential RMSE = 98.82 for six timestep
- MAE = 35.1 (60 timesteps), 46.5 (120 timesteps), 38.5
(180 timesteps)
- MAPE = 10.93% (60 timesteps), 12.22% (120 timesteps),
[154] Seq2Seq LSTM Residential
13.32% (180 timesteps)
- RMSE: 82.75 (60 timesteps), 86.50 (120 timesteps), 88.65
(180 timesteps)
- Training R = 0.98017
[155] ANFIS Educational Facility - Testing R = 0.9778
- Validation R = 0.97593
Bayesian RNN, Bayesian
[156] Not Specified - MAPE reduction = 15.4%
LSTM, Bayesian GRU

Table A3. Comparison of AI models used for HVAC control and optimization.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
[68] DRL Educational Facility - Energy consumption reduction = 21%
[69] ANN Office - Thermal energy consumption reduction 58.5%
- MSE = 0.0015
[70] LSTM, DRL Not Specified
- Energy savings = 27–30%
[71] FIS Church - Operation time reduction = 5.7%
- Energy consumption reduction = 0.7–4.18%,
[72] AMADRL Office
- Thermal comfort deviation = 64.13–72.08%
[73] YOLOv5 Educational Facility - YOLOv5 model accuracy = 88.1%
- Reduction in natural gas consumption = 22.2%
[74] GPR, ANN, SVM, DT, RF Educational Facility - Reduction in building heating demand = 4.3%
- GPR for heating demand RMSE = 32.1 kW
DNN Bilinear Koopman - CVRMSE: 9.62–19.15%
[75] Office
Predictor - Energy Savings Ratio = 33.71%
- Heating energy consumption reduced by 0.6% to 29.0%
[76] Shallow ANN Educational Facility - Thermal comfort improved by 0% to 58.8%
- Maintained indoor CO2 below 1000 ppm for 89.2%
Energies 2024, 17, 4277 25 of 35

Table A3. Cont.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
- PMV RMSE = 0.2243
- CO2 RMSE = 0.8816
[77] ANN Educational Facility
- PM10 RMSE = 0.4645
- PM2.5 RMSE = 0.6646
[78] DRL Residential - Energy consumption reduction = 5–14%
- PM2.5 healthy period increased by 21%
[79] DRL, DQN Residential - Thermal comfort period increased by 16%
- Energy consumption reduced by 23%
[80] BDQ Office - Cooling energy reduction = 11%
[81] GRU-RL Office - Cost reduction = 14.5%
- Energy reduction = 46%
[82] ANN Sports Hall - Average RMSE = 0.06
- Average R = 0.99
[157] RL Hotel - Estimated energy savings = 21%
[158] DRL Residential - Cost reduction up to 21%
- Energy savings compared to baseline
[159] DRL Office
controller = 5–12%
[160] Double DQN Residential - Energy cost reduction 7.88–8.56%
[161] DRL, PPG Not Specified - Energy consumption reduction 2–14%
[104] DRL Office - HVAC energy consumption reduction = 37%
- Energy savings = 12.24%
[162] MLP, DL Residential
- Cost savings = 12.91%
[163] ANN Commercial - Energy savings = 10%
[164] DDPG Residential - Energy consumption reduction = 65%
[165] AFUCB-DQN Not Specified - Energy savings = 21.4–22.3%
[166] MAQMC Residential - Energy consumption reduction = 6.27%
[167] DDPG Office - Energy savings = 13.71%
[168] RNN, NARX Office - Energy savings = 26%
[169] DDPG Residential - Cost savings compared to DQN = 15%
- Heating energy savings = 36.8%
[170] SNNs Office
- Cooling energy savings = 3.5% to 33.9%
[171] DDPG Residential - Cost savings = 12.79%
- OBC energy savings = 7%
[172] OBC, DRLC Office
- DRLC energy savings = 2.4%
[173] DRL, PPO, DDPG Office - Energy savings = 13.1–14.3%
[174] MARL, DQ Residential - Cost savings = 19%
[175] DDPG Residential - Cost savings = 6.1–10.3%
- Cost savings = 23.63–24.29%
[176] PPO, LSTM Residential
- PMV = 83.3–87.5%
Energies 2024, 17, 4277 26 of 35

Table A4. Comparison of AI models used for occupant detection.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
[83] CNN Office - Accuracy = 80.62%
- Accuracy = 97%
[84] DMFF Residential
- Energy Savings: Up to 30%
- NRMSE = 0.0435
[85] YOLOv5 Office
- Annual HVAC and lighting energy savings = 10.2%
Correlation coefficients for all
[86] GA-LSTM, PSO-LSTM, LSTM Residential
predictions = 99.16–99.97%
[87] 1D CNN, RL Not Specified - Reduction in thermal discomfort = 10.9%
- RMSD = 4.8
[88] MLR Educational Facility
- MAE = 2.5
Faster R-CNN with - Equipment detection accuracy = 78.39%
[89] Office
InceptionV2 - Occupancy activity detection accuracy = 93.60%
[90] Faster R-CNN Office - Average detection accuracy for all activities = 92.2%
- RMSE = 0.883
- NRMSE = 0.141
[91] YOLO v4 Office
- Maintained indoor CO2 < 1000 ppm
- Heating energy savings = 27%
- RMSE = 15.59
[92] LM-BP Office - MAE = 10.16
- MAPE = 6.35
Thermal comfort improved by 43–73%
[93] YOLOv5 Office - Energy savings = 2.3–8.1%
- Occupant detection accuracy = 80–97%
- People counting accuracy = 98.9%
[94] Faster R-CNN Educational Facility
- Activity detection accuracy: 88.5%
- Action recognition accuracy = 87.66%
[177] ST-GCN Educational Facility
- Average thermal comfort prediction accuracy = 82.5%

Table A5. Comparison of AI models used for other areas of application.

Reliability (Accuracy/Savings)
Reference AI/ML Model Building Type
Error (RMSE, MSE, MAPE), Savings (%)
[95] ANN, SVM Thermal Comfort Prediction Residential
[96] 1D-CNN, RNN, LSTM Air Quality Prediction Residential
[97] CNN-GRU, MLP Indoor Temperature Prediction Not Specified
[98] FL-BM, ANFIS-BM Thermal Comfort Prediction Educational Facility
[99] Radial basis function NN Air Quality Prediction Office
[100] GNN Indoor Temperature Prediction Office
[101] ANN Indoor Temperature Prediction Educational Facility
[102] CNN-LSTM Indoor Temperature Prediction Office
[103] MLP Indoor Temperature Prediction Educational Facility
[178] SVR-DNN Thermal Comfort Prediction Residential
[179] MLPNN, GA Thermal Comfort Prediction Public Building
Energies 2024, 17, 4277 27 of 35

Appendix B. Accuracy of AI Models Used for Different Applications

Table A6. Comparison of AI models according to R2 values.

Reference Application Area AI Model Building Type R2 Assessment

DNN: R2 = 0.95
Energy Consumption ANN: R2 = 0.94
[34] ANN, DNN, GB Residential High
Forecast GB: R2 = 0.92
RF: R2 = 0.88
Energy Consumption Asymmetric encoder–decoder Educational
[37] R2 = 0.964 High
Forecast deep learning algorithm Facility
Energy Consumption
[39] DF Commercial R2 = 0.90 High
Forecast
Energy Consumption Educational
[45] DNN R2 = 0.87 High
Forecast Facility
Energy Consumption
[46] RNNs Residential R2 = 0.999 High
Forecast
Energy Consumption
[47] 21-layer Fully Connected DNN Commercial R2 = 0.72 High
Forecast
A3C: R2 = 0.925
Energy Consumption A3C, DDPG,
[108] Office DDPG, RDPG: High
Forecast RDPG
R2 = 0.993
Energy Consumption
[110] CNN Mosque R2 = 0.98 High
Forecast
Energy Consumption Educational
[112] SVR R2 = 0.92 High
Forecast Facility
Optimized deep network model
with bidirectional LSTM,
Energy Consumption
[117] stacked unidirectional LSTM, Residential R2 = 0.998 High
Forecast
and fully connected layers
optimized using DTO
Energy Consumption
[121] Ensemble Residential R2 = 0.92601 High
Forecast
Energy Consumption
[124] GPR Public Building R2 = 0.9917 High
Forecast
Energy Consumption
[127] SADLA Office R2 = 0.967 High
Forecast
Energy Consumption LR, SVM, RF, MLP, DNN, RNN, Educational
[128] R2 = 88% High
Forecast LSTM, GRU Facility
Energy Consumption Proposed eight-layer deep
[129] Residential R2 = 97.5% High
Forecast neural network
Ensemble model combining
Energy Consumption
[136] LSSVR and RBFNN, optimized Residential R2 = 0.93 High
Forecast
by SOS
Energy Consumption
[137] EDA-LSTM Office R2 = 98.45% High
Forecast
Energy Consumption
[141] GA Office R2 = 0.993 High
Forecast
Energy Consumption
[147] Hybrid DNN-LSTM Residential R2 = 0.99911 High
Forecast
Indoor Temperature
[97] Multitask learning Not Specified R2 = 0.981 High
Prediction
Energies 2024, 17, 4277 28 of 35

Table A6. Cont.

Reference Application Area AI Model Building Type R2 Assessment

Indoor Temperature
[102] Transformer NN Office R2 = 0.936 High
Prediction
GMTCN combined with
[52] Load Forecast Hotel R2 = 0.971 High
Bidirectional LSTM with SPSA
Educational
[53] Load Forecast CEEMDAN and ARIMA R2 = 0.983 High
Facility
Multi-layer Perceptron NN
[54] Load Forecast Residential R2 = 0.920 High
optimized with BBO
R2 = 0.94 for
heating load,
[55] Load Forecast BBO-MLP Residential High
R2 = 0.997 for
cooling load
[59] Load Forecast TRN Office R2 = 0.98 High
DL with CNN-BiGRU and PSO
[60] Load Forecast Residential R2 = 0.9229 High
optimization
[61] Load Forecast HHO-ANFIS Residential R2 = 98% High
Educational
[63] Load Forecast iCEEMDAN-BO-LSTM R2 = 0.9869 High
Facility
Respectively,
LSTM: R2 = 0.920,
[65] Load Forecast LSTM, CIFG, GRU Public Building High
CIFG: R2 = 0.914,
GRU: R2 = 0.925
R2 = 0.999 for
heating load,
[149] Load Forecast 3RF Not Specified High
R2 = 0.997 for
cooling load
Thermal Comfort
[95] ANN Residential R2 = 0.4872 Medium
Prediction

References
1. International Energy Agency. Tracking Clean Energy Progress 2023. In Proceedings of the Assessing Critical Energy Technologies
for Global Clean Energy Transitions, Paris, France, 12 July 2023.
2. Bai, Y.; Yu, C.; Pan, W. Systematic Examination of Energy Performance Gap in Low-Energy Buildings. Renew. Sustain. Energy Rev.
2024, 202, 114701. [CrossRef]
3. Publications Office of the European Union. Directive 2024/1275 of the European Parliament and of the Council of 24 April 2024
on the Energy Performance of Buildings (Recast) (Text with EEA Relevance). Off. J. Eur. Union 2024, 8, 1275.
4. International Energy Agency. Digitalization and Energy; International Energy Agency: Paris, France, 2017.
5. Alanne, K.; Sierla, S. An Overview of Machine Learning Applications for Smart Buildings. Sustain. Cities Soc. 2022, 76, 103445.
[CrossRef]
6. Luo, J.X. A Bibliometric Review on Artificial Intelligence for Smart Buildings. Sustainability 2022, 14, 10230. [CrossRef]
7. Broday, E.E.; da Silva, M.C.G. The Role of Internet of Things (IoT) in the Assessment and Communication of Indoor Environmental
Quality (IEQ) in Buildings: A Review. Smart Sustain. Built Environ. 2023, 12, 584–606. [CrossRef]
8. Ghahramani, A.; Galicia, P.; Lehrer, D.; Varghese, Z.; Wang, Z.; Pandit, Y. Artificial Intelligence for Efficient Thermal Comfort
Systems: Requirements, Current Applications and Future Directions. Front. Built Environ. 2020, 6, 49. [CrossRef]
9. Farzaneh, H.; Malehmirchegini, L.; Bejan, A.; Afolabi, T.; Mulumba, A.; Daka, P.P. Artificial Intelligence Evolution in Smart
Buildings for Energy Efficiency. Appl. Sci. 2021, 11, 763. [CrossRef]
10. Heaton, J. Ian Goodfellow, Yoshua Bengio, and Aaron Courville: Deep Learning. Genet. Program. Evolvable Mach. 2018, 19,
305–307. [CrossRef]
11. Morales, E.F.; Escalante, H.J. A Brief Introduction to Supervised, Unsupervised, and Reinforcement Learning. In Biosignal Process-
ing and Classification Using Computational Learning and Intelligence: Principles, Algorithms, and Applications; Elsevier: Amsterdam,
The Netherlands, 2021; pp. 111–129, ISBN 9780128201251.
12. Raj, A. A Review on Machine Learning Algorithms. Int. J. Res. Appl. Sci. Eng. Technol. 2019, 7, 792–796. [CrossRef]
Energies 2024, 17, 4277 29 of 35

13. Guerrisi, A.; Falcone, I.; Valenti, F.; Rao, M.; Gallo, E.; Ungania, S.; Maccallini, M.T.; Fanciulli, M.; Frascione, P.; Morrone, A.; et al.
Artificial Intelligence and Advanced Melanoma: Treatment Management Implications. Cells 2022, 11, 3965. [CrossRef]
14. Wu, J.R.; Nguyen, S.; Alahakoon, D.; De Silva, D.; Mills, N.; Rathnayaka, P.; Moraliyage, H.; Jennings, A. A Comparative Analysis
of Machine Learning-Based Energy Baseline Models across Multiple Building Types. Energies 2024, 17, 1285. [CrossRef]
15. Moghimi, S.M.; Gulliver, T.A.; Chelvan, I.T. Energy Management in Modern Buildings Based on Demand Prediction and Machine
Learning-A Review. Energies 2024, 17, 555. [CrossRef]
16. Merabet, G.H.; Essaaidi, M.; Ben Haddou, M.; Qolomany, B.; Qadir, J.; Anan, M.; Al-Fuqaha, A.; Abid, M.R.; Benhaddou, D.
Intelligent Building Control Systems for Thermal Comfort and Energy-Efficiency: A Systematic Review of Artificial Intelligence-
Assisted Techniques. Renew. Sustain. Energy Rev. 2021, 144, 110969. [CrossRef]
17. Sayed, A.N.; Himeur, Y.; Bensaali, F. Deep and Transfer Learning for Building Occupancy Detection: A Review and Comparative
Analysis. Eng. Appl. Artif. Intell. 2022, 115, 105254. [CrossRef]
18. Bashir, M.B.; Alotaib, A.A. Smart Buildings Cooling and Heating Load Forecasting Models: Review. Int. J. Comput. Sci. Netw.
Secur. 2020, 20, 79–94. [CrossRef]
19. Rodrigues, F.; Cardeira, C.; Calado, J.M.F.; Melicio, R. Short-Term Load Forecasting of Electricity Demand for the Residential
Sector Based on Modelling Techniques: A Systematic Review. Energies 2023, 16, 4098. [CrossRef]
20. Runge, J.; Zmeureanu, R. A Review of Deep Learning Techniques for Forecasting Energy Use in Buildings. Energies 2021, 14, 608.
[CrossRef]
21. Mousavi, S.; Villarreal-Marroquín, M.G.; Hajiaghaei-Keshteli, M.; Smith, N.R. Data-Driven Prediction and Optimization toward
Net-Zero and Positive-Energy Buildings: A Systematic Review. Build. Environ. 2023, 242, 110578. [CrossRef]
22. Ma, P.; Cui, S.H.; Chen, M.S.; Zhou, S.Z.; Wang, K. Review of Family-Level Short-Term Load Forecasting and Its Application in
Household Energy Management System. Energies 2023, 16, 5809. [CrossRef]
23. Shaqour, A.; Hagishima, A. Systematic Review on Deep Reinforcement Learning-Based Energy Management for Different
Building Types. Energies 2022, 15, 8663. [CrossRef]
24. Ardabili, S.; Abdolalizadeh, L.; Mako, C.; Torok, B.; Mosavi, A. Systematic Review of Deep Learning and Machine Learning for
Building Energy. Front. Energy Res. 2022, 10, 786027. [CrossRef]
25. Yan, B.; Hao, F.; Meng, X. When Artificial Intelligence Meets Building Energy Efficiency, a Review Focusing on Zero Energy
Building. Artif. Intell. Rev. 2021, 54, 2193–2220. [CrossRef]
26. Tien, P.W.; Wei, S.Y.; Darkwa, J.; Wood, C.; Calautit, J.K. Machine Learning and Deep Learning Methods for Enhancing Building
Energy Efficiency and Indoor Environmental Quality—A Review. Energy AI 2022, 10, 100198. [CrossRef]
27. Whiting, P.; Savović, J.; Higgins, J.P.T.; Caldwell, D.M.; Reeves, B.C.; Shea, B.; Davies, P.; Kleijnen, J.; Churchill, R. ROBIS: A New
Tool to Assess Risk of Bias in Systematic Reviews Was Developed. J. Clin. Epidemiol. 2016, 69, 225–234. [CrossRef] [PubMed]
28. Moher, D.; Liberati, A.; Tetzlaff, J.; Altman, D.G.; Antes, G.; Atkins, D.; Barbour, V.; Barrowman, N.; Berlin, J.A.; Clark, J.; et al.
Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med. 2009, 6, 889–896.
[CrossRef] [PubMed]
29. Shittu Olumide Root Mean Square Error (RMSE) in AI: What You Need to Know. Available online: https://arize.com/blog-
course/root-mean-square-error-rmse-what-you-need-to-know/ (accessed on 1 July 2024).
30. Noa Azaria a Comprehensive Guide to Mean Absolute Percentage Error (MAPE). Available online: https://www.aporia.com/
learn/a-comprehensive-guide-to-mean-absolute-percentage-error-mape/ (accessed on 1 July 2024).
31. Ejiro Onose R Squared: Understanding the Coefficient of Determination. Available online: https://arize.com/blog-course/
r-squared-understanding-the-coefficient-of-determination/#:~:text=The%20R-squared%20metric%20%E2%80%94%20R%C2
%B2,be%20explained%20by%20your%20model (accessed on 1 July 2024).
32. Gao, Z.K.; Yu, J.Q.; Zhao, A.J.; Hu, Q.; Yang, S.Y. A Hybrid Method of Cooling Load Forecasting for Large Commercial Building
Based on Extreme Learning Machine. Energy 2022, 238, 122073. [CrossRef]
33. Tchagang, A.B.; Ashouri, A. Accurate Energy Forecast in Buildings: A Data Driven Machine Learning Approach. In Proceedings
of the 2019 IEEE Canadian Conference of Electrical and Computer Engineering, CCECE 2019, Edmonton, AB, Canada, 5–8
May 2019.
34. Olu-Ajayi, R.; Alaka, H.; Sulaimon, I.; Sunmola, F.; Ajayi, S. Building Energy Consumption Prediction for Residential Buildings
Using Deep Learning and Other Machine Learning Techniques. J. Build. Eng. 2022, 45, 103406. [CrossRef]
35. Urda, D.; Veredas, F.J.; González-Enrique, J.; Ruiz-Aguilar, J.J.; Jerez, J.M.; Turias, I.J. Deep Neural Networks Architecture Driven
by Problem-Specific Information. Neural Comput. Appl. 2021, 33, 9403–9423. [CrossRef]
36. Zhou, C.G.; Fang, Z.S.; Xu, X.N.; Zhang, X.L.; Ding, Y.F.; Jiang, X.Y.; Ji, Y. Using Long Short-Term Memory Networks to Predict
Energy Consumption of Air-Conditioning Systems. Sustain. Cities Soc. 2020, 55, 102000. [CrossRef]
37. Guo, J.; Lin, P.H.; Zhang, L.M.; Pan, Y.; Xiao, Z.H. Dynamic Adaptive Encoder-Decoder Deep Learning Networks for Multivariate
Time Series Forecasting of Building Energy Consumption. Appl. Energy 2023, 350, 121803. [CrossRef]
38. Zhuang, D.; Gan, V.J.L.; Tekler, Z.D.; Chong, A.D.; Tian, S.; Shi, X. Data-Driven Predictive Control for Smart HVAC System in
IoT-Integrated Buildings with Time-Series Forecasting and Reinforcement Learning. Appl. Energy 2023, 338, 120936. [CrossRef]
39. Zheng, G.F.; Feng, Z.; Jiang, M.K.; Tan, L.; Wang, Z.L. Predicting the Energy Consumption of Commercial Buildings Based on
Deep Forest Model and Its Interpretability. Buildings 2023, 13, 2162. [CrossRef]
Energies 2024, 17, 4277 30 of 35

40. Jallal, M.A.; González-Vidal, A.; Skarmeta, A.F.; Chabaa, S.; Zeroual, A. A Hybrid Neuro-Fuzzy Inference System-Based Algorithm
for Time Series Forecasting Applied to Energy Consumption Prediction. Appl. Energy 2020, 268, 114977. [CrossRef]
41. Khan, Z.A.; Hussain, T.; Ullah, A.; Rho, S.; Lee, M.; Baik, S.W. Towards Efficient Electricity Forecasting in Residential and
Commercial Buildings: A Novel Hybrid CNN with a LSTM-AE Based Framework. Sensors 2020, 20, 1399. [CrossRef] [PubMed]
42. Gonçalves, R.; Ribeiro, V.M.; Pereira, F.L. Variable Split Convolutional Attention: A Novel Deep Learning Model Applied to the
Household Electric Power Consumption. Energy 2023, 274, 127321. [CrossRef]
43. Gao, Y.; Ruan, Y.J. Interpretable Deep Learning Model for Building Energy Consumption Prediction Based on Attention
Mechanism. Energy Build. 2021, 252, 111379. [CrossRef]
44. Runge, J.; Zmeureanu, R. Deep Learning Forecasting for Electric Demand Applications of Cooling Systems in Buildings. Adv.
Eng. Inform. 2022, 53, 101674. [CrossRef]
45. Anand, P.; Deb, C.; Yan, K.; Yang, J.J.; Cheong, D.; Sekhar, C. Occupancy-Based Energy Consumption Modelling Using Machine
Learning Algorithms for Institutional Buildings. Energy Build. 2021, 252, 111478. [CrossRef]
46. Fang, L.; He, B. A Deep Learning Framework Using Multi-Feature Fusion Recurrent Neural Networks for Energy Consumption
Forecasting. Appl. Energy 2023, 348, 121563. [CrossRef]
47. Kannan, R.; Roy, M.S.; Pathuri, S.H. Artificial Intelligence Based Air Conditioner Energy Saving Using a Novel Preference Map.
IEEE Access 2020, 8, 206622–206637. [CrossRef]
48. Tsalikidis, N.; Mystakidis, A.; Tjortjis, C.; Koukaras, P.; Ioannidis, D. Energy Load Forecasting: One-Step Ahead Hybrid Model
Utilizing Ensembling. Computing 2024, 106, 241–273. [CrossRef]
49. Das, A.; Annaqeeb, M.K.; Azar, E.; Novakovic, V.; Kjærgaard, M.B. Occupant-Centric Miscellaneous Electric Loads Prediction in
Buildings Using State-of-the-Art Deep Learning Methods. Appl. Energy 2020, 269, 115135. [CrossRef]
50. Wen, L.L.; Zhou, K.L.; Yang, S.L. Load Demand Forecasting of Residential Buildings Using a Deep Learning Model. Electr. Power
Syst. Res. 2020, 179, 106073. [CrossRef]
51. Yang, W.W.; Shi, J.; Li, S.J.; Song, Z.F.; Zhang, Z.T.; Chen, Z.X. A Combined Deep Learning Load Forecasting Model of Single
Household Resident User Considering Multi-Time Scale Electricity Consumption Behavior. Appl. Energy 2022, 307, 118197.
[CrossRef]
52. Song, C.R.; Yang, H.D.; Meng, X.B.; Yang, P.; Cai, J.Y.; Bao, H.; Xu, K.K. A Novel Deep-Learning Framework for Short-Term
Prediction of Cooling Load in Public Buildings. J. Clean. Prod. 2024, 434, 139796. [CrossRef]
53. Zhou, Y.; Wang, L.Y.; Qian, J.H. Application of Combined Models Based on Empirical Mode Decomposition, Deep Learning,
and Autoregressive Integrated Moving Average Model for Short-Term Heating Load Predictions. Sustainability 2022, 14, 7349.
[CrossRef]
54. Xu, Y.J.; Li, F.; Asgari, A. Prediction and Optimization of Heating and Cooling Loads in a Residential Building Based on
Multi-Layer Perceptron Neural Network and Different Optimization Algorithms. Energy 2022, 240, 122692. [CrossRef]
55. Moayedi, H.; Mosavi, A. Double-Target Based Neural Networks in Predicting Energy Consumption in Residential Buildings.
Energies 2021, 14, 1331. [CrossRef]
56. Li, X.Y.; Yao, R.M. A Machine-Learning-Based Approach to Predict Residential Annual Space Heating and Cooling Loads
Considering Occupant Behaviour. Energy 2020, 212, 118676. [CrossRef]
57. Son, H.; Kim, C. A Deep Learning Approach to Forecasting Monthly Demand for Residential-Sector Electricity. Sustainability
2020, 12, 3103. [CrossRef]
58. Wang, J.S.; Chen, X.H.; Zhang, F.; Chen, F.X.; Xin, Y. Building Load Forecasting Using Deep Neural Network with Efficient Feature
Fusion. J. Mod. Power Syst. Clean Energy 2021, 9, 160–169. [CrossRef]
59. Li, L.; Su, X.Y.; Bi, X.T.; Lu, Y.L.; Sun, X.T. A Novel Transformer-Based Network Forecasting Method for Building Cooling Loads.
Energy Build. 2023, 296, 113409. [CrossRef]
60. Irankhah, A.; Yaghmaee, M.H.; Ershadi-Nasab, S. Optimized Short-Term Load Forecasting in Residential Buildings Based on
Deep Learning Methods for Different Time Horizons. J. Build. Eng. 2024, 84, 113409. [CrossRef]
61. Moayedi, H.; Van, B.L. Feasibility of Harris Hawks Optimization in Combination with Fuzzy Inference System Predicting Heating
Load Energy Inside Buildings. Energies 2022, 15, 9187. [CrossRef]
62. Chitalia, G.; Pipattanasomporn, M.; Garg, V.; Rahman, S. Robust Short-Term Electrical Load Forecasting Framework for
Commercial Buildings Using Deep Recurrent Neural Networks. Appl. Energy 2020, 278, 115410. [CrossRef]
63. He, Y.Q.; Tsang, K.F. Universities Power Energy Management: A Novel Hybrid Model Based on ICEEMDAN and Bayesian
Optimized LSTM. Energy Rep. 2021, 7, 6473–6488. [CrossRef]
64. Wang, Z.; Hong, T.Z.; Piette, M.A. Building Thermal Load Prediction through Shallow Machine Learning and Deep Learning.
Appl. Energy 2020, 263, 114683. [CrossRef]
65. Zhang, C.B.; Li, J.Y.; Zhao, Y.; Li, T.T.; Chen, Q.; Zhang, X.J. A Hybrid Deep Learning-Based Method for Short-Term Building
Energy Load Prediction Combined with an Interpretation Process. Energy Build. 2020, 225, 110301. [CrossRef]
66. Stienecker, M.; Hagemeier, A. Developing Feedforward Neural Networks as Benchmark for Load Forecasting: Methodology
Presentation and Application to Hospital Heat Load Forecasting. Energies 2023, 16, 2026. [CrossRef]
67. Zhou, Y.H.; Liang, Y.M.; Pan, Y.Q.; Yuan, X.L.; Xie, Y.R.; Jia, W.Q. A Deep-Learning-Based Meta-Modeling Workflow for Thermal
Load Forecasting in Buildings: Method and a Case Study. Buildings 2022, 12, 177. [CrossRef]
Energies 2024, 17, 4277 31 of 35

68. Chemingui, Y.; Gastli, A.; Ellabban, O. Reinforcement Learning-Based School Energy Management System. Energies 2020, 13, 6354.
[CrossRef]
69. Yang, S.Y.; Wan, M.P.; Chen, W.Y.; Ng, B.F.; Dubey, S. Model Predictive Control with Adaptive Machine-Learning-Based Model
for Building Energy Efficiency and Comfort Optimization. Appl. Energy 2020, 271, 115147. [CrossRef]
70. Zou, Z.B.; Yu, X.R.; Ergan, S. Towards Optimal Control of Air Handling Units Using Deep Reinforcement Learning and Recurrent
Neural Network. Build. Environ. 2020, 168, 106535. [CrossRef]
71. Martinez-Molina, A.; Alamaniotis, M. Enhancing Historic Building Performance with the Use of Fuzzy Inference System to
Control the Electric Cooling System. Sustainability 2020, 12, 5848. [CrossRef]
72. Yu, L.; Xu, Z.B.; Zhang, T.F.; Guan, X.H.; Yue, D. Energy-Efficient Personalized Thermal Comfort Control in Office Buildings
Based on Multi-Agent Deep Reinforcement Learning. Build. Environ. 2022, 223, 109458. [CrossRef]
73. Lan, H.F.; Hou, H.Y.; Gou, Z.H.; Wong, M.S.; Wang, Z. Computer Vision-Based Smart HVAC Control System for University
Classroom in a Subtropical Climate. Build. Environ. 2023, 242, 110592. [CrossRef]
74. Cotrufo, N.; Saloux, E.; Hardy, J.M.; Candanedo, J.A.; Platon, R. A Practical Artificial Intelligence-Based Approach for Predictive
Control in Commercial and Institutional Buildings. Energy Build. 2020, 206, 109563. [CrossRef]
75. Soleimani, M.; Irani, F.N.; Yadegar, M.; Davoodi, M. Multi-Objective Optimization of Building HVAC Operation: Advanced
Strategy Using Koopman Predictive Control and Deep Learning. Build. Environ. 2024, 248, 111073. [CrossRef]
76. Wang, Z.; Calautit, J.; Tien, P.W.; Wei, S.Y.; Zhang, W.X.; Wu, Y.P.; Xia, L. An Occupant-Centric Control Strategy for Indoor
Thermal Comfort, Air Quality and Energy Management. Energy Build. 2023, 285, 112899. [CrossRef]
77. Cho, J.H.; Moon, J.W. Integrated Artificial Neural Network Prediction Model of Indoor Environmental Quality in a School
Building. J. Clean. Prod. 2022, 344, 131083. [CrossRef]
78. Gupta, A.; Badr, Y.; Negahban, A.; Qiu, R.G. Energy-Efficient Heating Control for Smart Buildings with Deep Reinforcement
Learning. J. Build. Eng. 2021, 34, 101739. [CrossRef]
79. An, Y.T.; Chen, C. Energy-Efficient Control of Indoor PM2.5 and Thermal Comfort in a Real Room Using Deep Reinforcement
Learning. Energy Build. 2023, 295, 113340. [CrossRef]
80. Lei, Y.; Zhan, S.C.; Ono, E.; Peng, Y.Z.; Zhang, Z.; Hasama, T.; Chong, A.D. A Practical Deep Reinforcement Learning Framework
for Multivariate Occupant-Centric Control in Buildings. Appl. Energy 2022, 324, 119742. [CrossRef]
81. Gao, Y.; Shi, S.R.; Miyata, S.; Akashi, Y. Successful Application of Predictive Information in Deep Reinforcement Learning Control:
A Case Study Based on an Office Building HVAC System. Energy 2024, 291, 130344. [CrossRef]
82. Elnour, M.; Himeur, Y.; Fadli, F.; Mohammedsherif, H.; Meskin, N.; Ahmad, A.M.; Petri, I.; Rezgui, Y.; Hodorog, A. Neural
Network-Based Model Predictive Control System for Optimizing Building Automation and Management Systems of Sports
Facilities. Appl. Energy 2022, 318, 119153. [CrossRef]
83. Tien, P.; Wei, S.Y.; Calautit, J.K.; Darkwa, J.; Wood, C. A Vision-Based Deep Learning Approach for the Detection and Prediction
of Occupancy Heat Emissions for Demand-Driven Control Solutions. Energy Build. 2020, 226, 110386. [CrossRef]
84. Sun, K.L. DMFF: Deep Multimodel Feature Fusion for Building Occupancy Detection. Build. Environ. 2024, 253, 111355. [CrossRef]
85. Choi, H.; Um, C.Y.; Kang, K.; Kim, H.; Kim, T. Application of Vision-Based Occupancy Counting Method Using Deep Learning
and Performance Analysis. Energy Build. 2021, 252, 111389. [CrossRef]
86. Mahjoub, S.; Labdai, S.; Chrifi-Alaoui, L.; Marhic, B.; Delahoche, L. Short-Term Occupancy Forecasting for a Smart Home Using
Optimized Weight Updates Based on GA and PSO Algorithms for an LSTM Network. Energies 2023, 16, 1641. [CrossRef]
87. Jung, S.; Jeoung, J.; Hong, T. Occupant-Centered Real-Time Control of Indoor Temperature Using Deep Learning Algorithms.
Build. Environ. 2022, 208, 108633. [CrossRef]
88. Kim, Y.; Park, Y.; Seo, H.; Hwang, J. Load Prediction Algorithm Applied with Indoor Environment Sensing in University Buildings.
Energies 2023, 16, 999. [CrossRef]
89. Wei, S.Y.; Tien, P.W.; Wu, Y.P.; Calautit, J.K. A Coupled Deep Learning-Based Internal Heat Gains Detection and Prediction
Method for Energy-Efficient Office Building Operation. J. Build. Eng. 2022, 47, 103778. [CrossRef]
90. Tien, P.; Wei, S.Y.; Calautit, J.; Darkwa, J.; Wood, C. Occupancy Heat Gain Detection and Prediction Using Deep Learning
Approach for Reducing Building Energy Demand. J. Sustain. Dev. Energy Water Environ. Syst. JSDEWES 2021, 9, 1–31. [CrossRef]
91. Choi, H.; Lee, J.; Yi, Y.; Na, H.; Kang, K.; Kim, T. Deep Vision-Based Occupancy Counting: Experimental Performance Evaluation
and Implementation of Ventilation Control. Build. Environ. 2022, 223, 109496. [CrossRef]
92. Liang, R.; Ding, W.F.; Zandi, Y.; Rahimi, A.; Pourkhorshidi, S.; Khadimallah, M.A. Buildings’ Internal Heat Gains Prediction
Using Artificial Intelligence Methods. Energy Build. 2022, 258, 111794. [CrossRef]
93. Yang, B.; Liu, Y.H.; Liu, P.J.; Wang, F.M.; Cheng, X.G.; Lv, Z.H. A Novel Occupant-Centric Stratum Ventilation System Using
Computer Vision: Occupant Detection, Thermal Comfort, Air Quality, and Energy Savings. Build. Environ. 2023, 237, 110332.
[CrossRef]
94. Wei, S.Y.; Tien, P.W.; Chow, T.W.; Wu, Y.P.; Calautit, J.K. Deep Learning and Computer Vision Based Occupancy CO2 Level
Prediction for Demand-Controlled Ventilation (DCV). J. Build. Eng. 2022, 56, 104715. [CrossRef]
95. Chai, Q.; Wang, H.Q.; Zhai, Y.C.; Yang, L. Using Machine Learning Algorithms to Predict Occupants’ Thermal Comfort in
Naturally Ventilated Residential Buildings. Energy Build. 2020, 217, 109937. [CrossRef]
96. Rescio, G.; Manni, A.; Caroppo, A.; Carluccio, A.M.; Siciliano, P.; Leone, A. Multi-Sensor Platform for Predictive Air Quality
Monitoring. Sensors 2023, 23, 5139. [CrossRef] [PubMed]
Energies 2024, 17, 4277 32 of 35

97. Qiao, D.L.; Shen, B.L.; Dong, X.Y.; Zheng, H.; Song, W.W.; Wu, S. MTL-Deep-STF: A Multitask Learning Based Deep Spa-
tiotemporal Fusion Model for Outdoor Air Temperature Prediction in Building HVAC Systems. J. Build. Eng. 2022, 62, 105364.
[CrossRef]
98. López-Pérez, L.A.; Flores-Prieto, J.J. Adaptive Thermal Comfort Approach to Save Energy in Tropical Climate Educational
Building by Artificial Intelligence. Energy 2023, 263, 125706. [CrossRef]
99. Majdi, A.; Alrubaie, A.J.; Al-Wardy, A.H.; Baili, J.; Panchal, H. A Novel Method for Indoor Air Quality Control of Smart Homes
Using a Machine Learning Model. Adv. Eng. Softw. 2022, 173, 103253. [CrossRef]
100. Zhang, J.; Xiao, F.; Li, A.; Ma, T.Y.; Xu, K.; Zhang, H.B.; Yan, R.; Fang, X.; Li, Y.Y.; Wang, D. Graph Neural Network-Based
Spatio-Temporal Indoor Environment Prediction and Optimal Control for Central Air-Conditioning Systems. Build. Environ.
2023, 242, 110600. [CrossRef]
101. Essa, M.E.M.; El-shafeey, A.M.; Omar, A.H.; Fathi, A.E.; El Maref, A.S.A.; Lotfy, J.V.W.; El-Sayed, M.S. Reliable Integration of
Neural Network and Internet of Things for Forecasting, Controlling, and Monitoring of Experimental Building Management
System. Sustainability 2023, 15, 2168. [CrossRef]
102. Faiz, M.F.; Sajid, M.; Ali, S.; Javed, K.; Ayaz, Y. Energy Modeling and Predictive Control of Environmental Quality for Building
Energy Management Using Machine Learning. Energy Sustain. Dev. 2023, 74, 381–395. [CrossRef]
103. Park, B.K.; Kim, C.J.; Adhikari, R.S. Short-Term Prediction for Indoor Temperature Control Using Artificial Neural Network.
Energies 2023, 16, 7724. [CrossRef]
104. Wang, H.; Chen, X.W.; Vital, N.; Duffy, E.; Razi, A. Energy Optimization for HVAC Systems in Multi-VAV Open Offices: A Deep
Reinforcement Learning Approach. Appl. Energy 2024, 356, 122354. [CrossRef]
105. Ahmad, M.W.; Mourshed, M.; Rezgui, Y. Trees vs Neurons: Comparison between Random Forest and ANN for High-Resolution
Prediction of Building Energy Consumption. Energy Build. 2017, 147, 77–89. [CrossRef]
106. Zhao, H.X.; Magoulès, F. A Review on the Prediction of Building Energy Consumption. Renew. Sustain. Energy Rev. 2012, 16,
3586–3592. [CrossRef]
107. Abdelaziz, A.; Santos, V.; Dias, M.S.; Mahmoud, A.N. A Hybrid Model of Self-Organizing Map and Deep Learning with Genetic
Algorithm for Managing Energy Consumption in Public Buildings. J. Clean. Prod. 2024, 434, 140040. [CrossRef]
108. Liu, T.; Tan, Z.H.; Xu, C.L.; Chen, H.X.; Li, Z.F. Study on Deep Reinforcement Learning Techniques for Building Energy
Consumption Forecasting. Energy Build. 2020, 208, 109675. [CrossRef]
109. Mawson, V.J.; Hughes, B. Deep Learning Techniques for Energy Forecasting and Condition Monitoring in the Manufacturing
Sector. Energy Build. 2020, 217, 109966. [CrossRef]
110. El-Maraghy, M.; Metawie, M.; Safaan, M.; Eldin, A.S.; Hamdy, A.; El Sharkawy, M.; Abdelaty, A.; Azab, S.; Marzouk, M. Predicting
Energy Consumption of Mosque Buildings during the Operation Stage Using Deep Learning Approach. Energy Build. 2024,
303, 113829. [CrossRef]
111. Cao, Y.S.; Liu, G.; Sun, J.; Bavirisetti, D.P.; Xiao, G. PSO-Stacking Improved Ensemble Model for Campus Building Energy
Consumption Forecasting Based on Priority Feature Selection. J. Build. Eng. 2023, 72, 106589. [CrossRef]
112. Huang, J.; Kaewunruen, S. Forecasting Energy Consumption of a Public Building Using Transformer and Support Vector
Regression. Energies 2023, 16, 966. [CrossRef]
113. Alraddadi, G.H.; Ben Othman, M.T. Development of an Efficient Electricity Consumption Prediction Model Using Machine
Learning Techniques. Int. J. Adv. Comput. Sci. Appl. 2022, 13, 376–384. [CrossRef]
114. Hora, S.K.; Poongodan, R.; de Prado, R.P.; Wozniak, M.; Divakarachari, P.B. Long Short-Term Memory Network-Based Meta-
heuristic for Effective Electric Energy Consumption Prediction. Appl. Sci. Basel 2021, 11, 11263. [CrossRef]
115. Jozi, A.; Pinto, T.; Marreiros, G.; Vale, Z. Electricity Consumption Forecasting in Office Buildings: An Artificial Intelligence
Approach. In Proceedings of the 2019 IEEE Milan PowerTech, Milan, Italy, 23–27 June 2019.
116. Eseye, A.T.; Lehtonen, M.; Tukia, T.; Uimonen, S.; Millar, R.J. Short-Term Forecasting of Electricity Consumption in Buildings
for Efficient and Optimal Distributed Energy Management. In Proceedings of the IEEE International Conference on Industrial
Informatics (INDIN), Helsinki-Espoo, Finland, 22–25 July 2019; pp. 1103–1110.
117. Abdelhamid, A.A.; El-Kenawy, E.M.; Alrowais, F.; Ibrahim, A.; Khodadadi, N.; Lim, W.H.; Alruwais, N.; Khafaga, D.S. Deep
Learning with Dipper Throated Optimization Algorithm for Energy Consumption Forecasting in Smart Households. Energies
2022, 15, 9125. [CrossRef]
118. Broujeny, R.S.; Ben Ayed, S.; Matalah, M. Energy Consumption Forecasting in a University Office by Artificial Intelligence
Techniques: An Analysis of the Exogenous Data Effect on the Modeling. Energies 2023, 16, 4065. [CrossRef]
119. Liu, Z.H.; Jiang, G.Y. Optimization of Intelligent Heating Ventilation Air Conditioning System in Urban Building Based on BIM
and Artificial Intelligence Technology. Comput. Sci. Inf. Syst. 2021, 18, 1379–1394. [CrossRef]
120. Wang, H.B.; Li, H.Y.; Gao, S.T.; Zhou, L.K.; Lao, Z.Y. Multigranularity Building Energy Consumption Prediction Method Based on
Convolutional Recurrent Neural Network. Wirel. Commun. Mob. Comput. 2022, 2022, 8524034. [CrossRef]
121. Ramos, P.V.B.; Villela, S.M.; Silva, W.N.; Dias, B.H. Residential Energy Consumption Forecasting Using Deep Learning Models.
Appl. Energy 2023, 350, 121705. [CrossRef]
122. Kiprijanovska, I.; Stankoski, S.; Ilievski, I.; Jovanovski, S.; Gams, M.; Gjoreski, H. HousEEC: Day-Ahead Household Electrical
Energy Consumption Forecasting Using Deep Learning. Energies 2020, 13, 2672. [CrossRef]
Energies 2024, 17, 4277 33 of 35

123. Fernández-Martínez, D.; Jaramillo-Morán, M.A. Multi-Step Hourly Power Consumption Forecasting in a Healthcare Building
with Recurrent Neural Networks and Empirical Mode Decomposition. Sensors 2022, 22, 3664. [CrossRef]
124. Feng, Y.Y.; Huang, Y.X.; Shang, H.F.; Lou, J.W.; Knefaty, A.D.; Yao, J.; Zheng, R.Y. Prediction of Hourly Air-Conditioning Energy
Consumption in Office Buildings Based on Gaussian Process Regression. Energies 2022, 15, 4626. [CrossRef]
125. Lin, X.Y.; Yu, H.; Wang, M.; Li, C.E.; Wang, Z.; Tang, Y. Electricity Consumption Forecast of High-Rise Office Buildings Based on
the Long Short-Term Memory Method. Energies 2021, 14, 4785. [CrossRef]
126. Tomar, D.; Tomar, P.; Bhardwaj, A.; Sinha, G.R. Deep Learning Neural Network Prediction System Enhanced with Best Window
Size in Sliding Window Algorithm for Predicting Domestic Power Consumption in a Residential Building. Comput. Intell. Neurosci.
2022, 2022, 7216959. [CrossRef] [PubMed]
127. Uwiragiye, B.; Duhirwe, P.N.; Seo, H.; Yun, G.Y. Sequential Attention Deep Learning Architecture with Unsupervised Pre-Training
for Interpretable and Accurate Building Energy Prediction with Limited Data. J. Asian Archit. Build. Eng. 2023, 2278460. [CrossRef]
128. Liao, J.M.; Chang, M.J.; Chang, L.M. Prediction of Air-Conditioning Energy Consumption in R&D Building Using Multiple
Machine Learning Techniques. Energies 2020, 13, 1847. [CrossRef]
129. Truong, L.M.; Chow, K.H.K.; Luevisadpaibul, R.; Thirunavukkarasu, G.S.; Seyedmahmoudian, M.; Horan, B.; Mekhilef, S.;
Stojcevski, A. Accurate Prediction of Hourly Energy Consumption in a Residential Building Based on the Occupancy Rate Using
Machine Learning Approaches. Appl. Sci. Basel 2021, 11, 2229. [CrossRef]
130. Jiang, F.F.; Ma, J.; Li, Z.; Ding, Y.X. Prediction of Energy Use Intensity of Urban Buildings Using the Semisupervised Deep
Learning Model. Energy 2022, 249, 123631. [CrossRef]
131. Jin, W.; Fu, Q.M.; Chen, J.P.; Wang, Y.Z.; Liu, L.H.; Lu, Y.; Wu, H.J. A Novel Building Energy Consumption Prediction Method
Using Deep Reinforcement Learning with Consideration of Fluctuation Points. J. Build. Eng. 2023, 63, 105458. [CrossRef]
132. Salmi, T.; Kiljander, J.; Pakkala, D. Stacked Boosters Network Architecture for Short-Term Load Forecasting in Buildings. Energies
2020, 13, 2370. [CrossRef]
133. Xu, Y.; Gao, W.J.; Qian, F.Y.; Li, Y.X. Potential Analysis of the Attention-Based LSTM Model in Ultra-Short-Term Forecasting of
Building HVAC Energy Consumption. Front. Energy Res. 2021, 9, 730640. [CrossRef]
134. Fan, B.Q.; Xing, X.X. Intelligent Prediction Method of Building Energy Consumption Based on Deep Learning. Sci. Program. 2021,
2021, 3323316. [CrossRef]
135. Hyeon, J.; Lee, H.; Ko, B.; Choi, H.J. Deep Learning-Based Household Electric Energy Consumption Forecasting. J. Eng. JOE 2020,
2020, 639–642. [CrossRef]
136. Tran, D.H.; Luong, D.L.; Chou, J.S. Nature-Inspired Metaheuristic Ensemble Model for Forecasting Energy Consumption in
Residential Buildings. Energy 2020, 191, 116552. [CrossRef]
137. Ding, Z.K.; Chen, W.L.; Hu, T.; Xu, X.X. Evolutionary Double Attention-Based Long Short-Term Memory Model for Building
Energy Prediction: Case Study of a Green Building. Appl. Energy 2021, 288, 116660. [CrossRef]
138. Khan, Z.A.; Ullah, A.; Ullah, W.; Rho, S.; Lee, M.; Baik, S.W. Electrical Energy Prediction in Residential Buildings for Short-Term
Horizons Using Hybrid Deep Learning Strategy. Appl. Sci. 2020, 10, 8634. [CrossRef]
139. So, D.; Oh, J.; Jeon, I.; Moon, J.; Lee, M.; Rho, S. BiGTA-Net: A Hybrid Deep Learning-Based Electrical Energy Forecasting Model
for Building Energy Management Systems. Systems 2023, 11, 456. [CrossRef]
140. Somu, N.; Raman, M.R.G.; Ramamritham, K. A Deep Learning Framework for Building Energy Consumption Forecast. Renew.
Sustain. Energy Rev. 2021, 137, 591. [CrossRef]
141. Luo, X.J.; Oyedele, L.O.; Ajayi, A.O.; Akinade, O.O.; Owolabi, H.A.; Ahmed, A. Feature Extraction and Genetic Algorithm
Enhanced Adaptive Deep Neural Network for Energy Consumption Prediction in Buildings. Renew. Sustain. Energy Rev. 2020,
131, 109980. [CrossRef]
142. Aouad, M.; Hajj, H.; Shaban, K.; Jabr, R.A.; El-Hajj, W. A CNN-Sequence-to-Sequence Network with Attention for Residential
Short-Term Load Forecasting. Electr. Power Syst. Res. 2022, 211, 108152. [CrossRef]
143. Zhang, G.Q.; Tian, C.L.; Li, C.D.; Zhang, J.J.; Zuo, W.D. Accurate Forecasting of Building Energy Consumption via a Novel
Ensembled Deep Learning Method Considering the Cyclic Feature. Energy 2020, 201, 117531. [CrossRef]
144. Lei, L.; Shao, S.L.; Liang, L.X. An Evolutionary Deep Learning Model Based on EWKM, Random Forest Algorithm, SSA and
BiLSTM for Building Energy Consumption Prediction. Energy 2024, 288, 129795. [CrossRef]
145. Bhoj, N.; Bhadoria, R.S. Time-Series Based Prediction for Energy Consumption of Smart Home Data Using Hybrid Convolution-
Recurrent Neural Network. Telemat. Inform. 2022, 75, 101907. [CrossRef]
146. Ruan, Y.J.; Wang, G.; Meng, H.; Qian, F.Y. A Hybrid Model for Power Consumption Forecasting Using VMD-Based the Long
Short-Term Memory Neural Network. Front. Energy Res 2022, 9, 772508. [CrossRef]
147. AL-Ghamdi, M.; AL-Ghamdi, A.A.; Ragab, M. A Hybrid DNN Multilayered LSTM Model for Energy Consumption Prediction.
Appl. Sci. 2023, 13, 11408. [CrossRef]
148. Yousaf, A.; Asif, R.M.; Shakir, M.; Rehman, A.U.; Adrees, M.S. An Improved Residential Electricity Load Forecasting Using a
Machine-Learning-Based Feature Selection Approach and a Proposed Integration Strategy. Sustainability 2021, 13, 6199. [CrossRef]
149. Chaganti, R.; Rustam, F.; Daghriri, T.; Diez, I.D.; Mazón, J.L.V.; Rodríguez, C.L.; Ashraf, I. Building Heating and Cooling Load
Prediction Using Ensemble Machine Learning Model. Sensors 2022, 22, 7692. [CrossRef]
150. Khan, Z.A.; Ullah, A.; Ul Haq, I.; Hamdy, M.; Mauro, G.M.; Muhammad, K.; Hijji, M.; Baik, S.W. Efficient Short-Term Electricity
Load Forecasting for Effective Energy Management. Sustain. Energy Technol. Assess. 2022, 53, 102337. [CrossRef]
Energies 2024, 17, 4277 34 of 35

151. Wang, C.D.; Zheng, L.H.; Yuan, J.J.; Huang, K.; Zhou, Z.H. Building Heat Demand Prediction Based on Reinforcement Learning
for Thermal Comfort Management. Energies 2022, 15, 7856. [CrossRef]
152. Lv, R.X.; Yuan, Z.Y.; Lei, B.; Zheng, J.C.; Luo, X.J. Building Thermal Load Prediction Using Deep Learning Method Considering
Time-Shifting Correlation in Feature Variables. J. Build. Eng. 2022, 61, 105316. [CrossRef]
153. Lotfipoor, A.; Patidar, S.; Jenkins, D.P. Deep Neural Network with Empirical Mode Decomposition and Bayesian Optimisation for
Residential Load Forecasting. Expert Syst. Appl. 2024, 237, 121355. [CrossRef]
154. Masood, Z.; Gantassi, R.; Ardiansyah; Choi, Y. A Multi-Step Time-Series Clustering-Based Seq2Seq LSTM Learning for a Single
Household Electricity Load Forecasting. Energies 2022, 15, 2623. [CrossRef]
155. Ghenai, C.; Al-Mufti, O.A.A.; Al-Isawi, O.A.M.; Amirah, L.H.L.; Merabet, A. Short-Term Building Electrical Load Forecasting
Using Adaptive Neuro-Fuzzy Inference System (ANFIS). J. Build. Eng. 2022, 52, 104323. [CrossRef]
156. Xu, L.; Hu, M.M.; Fan, C. Probabilistic Electrical Load Forecasting for Buildings Using Bayesian Deep Neural Networks. J. Build.
Eng. 2022, 46, 103853. [CrossRef]
157. Park, S.; Park, S.; Choi, M.I.; Lee, S.; Lee, T.; Kim, S.; Cho, K.; Park, S. Reinforcement Learning-Based BEMS Architecture for
Energy Usage Optimization. Sensors 2020, 20, 4918. [CrossRef]
158. Kurte, K.; Munk, J.; Kotevska, O.; Amasyali, K.; Smith, R.; McKee, E.; Du, Y.; Cui, B.R.; Kuruganti, T.; Zandi, H. Evaluating the
Adaptability of Reinforcement Learning Based HVAC Control for Residential Houses. Sustainability 2020, 12, 7727. [CrossRef]
159. Brandi, S.; Piscitelli, M.S.; Martellacci, M.; Capozzoli, A. Deep Reinforcement Learning to Optimise Indoor Temperature Control
and Heating Energy Consumption in Buildings. Energy Build. 2020, 224, 110225. [CrossRef]
160. Yang, T.; Zhao, L.Y.; Li, W.; Wu, J.Z.; Zomaya, A.Y. Towards Healthy and Cost-Effective Indoor Environment Management in
Smart Homes: A Deep Reinforcement Learning Approach. Appl. Energy 2021, 300, 117335. [CrossRef]
161. Nguyen, A.T.; Pham, D.H.; Oo, B.L.; Santamouris, M.; Ahn, Y.; Lim, B.T.H. Modelling Building HVAC Control Strategies Using a
Deep Reinforcement Learning Approach. Energy Build. 2024, 310, 114065. [CrossRef]
162. Esrafilian-Najafabadi, M.; Haghighat, F. Occupancy-Based HVAC Control Using Deep Learning Algorithms for Estimating Online
Preconditioning Time in Residential Buildings. Energy Build. 2021, 252, 111377. [CrossRef]
163. Yayla, A.; Swierczewska, K.S.; Kaya, M.; Karaca, B.; Arayici, Y.; Ayözen, Y.E.; Tokdemir, O.B. Artificial Intelligence (AI)-Based
Occupant-Centric Heating Ventilation and Air Conditioning (HVAC) Control System for Multi-Zone Commercial Buildings.
Sustainability 2022, 14, 16107. [CrossRef]
164. Solinas, F.M.; Macii, A.; Patti, E.; Bottaccioli, L. An Online Reinforcement Learning Approach for HVAC Control. Expert Syst.
Appl. 2024, 238, 121749. [CrossRef]
165. Tian, H.; Feng, M.W.; Fan, H.C.; Cao, R.R.; Gao, Q. Optimization Control Strategy for a Central Air Conditioning System Based
on AFUCB-DQN. Processes 2023, 11, 2068. [CrossRef]
166. Ding, Z.K.; Fu, Q.M.; Chen, J.P.; Lu, Y.; Wu, H.J.; Fang, N.W.; Xing, B. MAQMC: Multi-Agent Deep Q-Network for Multi-Zone
Residential HVAC Control. CMES Comput. Model. Eng. Sci. 2023, 136, 2759–2785. [CrossRef]
167. Lee, D.; Jeong, J.; Chae, Y.T. Application of Deep Reinforcement Learning for Proportional-Integral-Derivative Controller Tuning
on Air Handling Unit System in Existing Commercial Building. Buildings 2024, 14, 66. [CrossRef]
168. Yang, S.Y.; Wan, M.P. Machine-Learning-Based Model Predictive Control with Instantaneous Linearization-A Case Study on an
Air-Conditioning and Mechanical Ventilation System. Appl. Energy 2022, 306, 118041. [CrossRef]
169. Du, Y.; Zandi, H.; Kotevska, O.; Kurte, K.; Munk, J.; Amasyali, K.; Mckee, E.; Li, F.X. Intelligent Multi-Zone Residential HVAC
Control Strategy Based on Deep Reinforcement Learning. Appl. Energy 2021, 281, 116117. [CrossRef]
170. Wang, Z.; Calautit, J.; Wei, S.Y.; TIen, P.W.; Xia, L. Real-Time Building Heat Gains Prediction and Optimization of HVAC Setpoint:
An Integrated Framework. J. Build. Eng. 2022, 49, 104103. [CrossRef]
171. Liu, X.F.; Ren, M.F.; Yang, Z.L.; Yan, G.W.; Guo, Y.J.; Cheng, L.; Wu, C.K. A Multi-Step Predictive Deep Reinforcement Learning
Algorithm for HVAC Control Systems in Smart Buildings. Energy 2022, 259, 124857. [CrossRef]
172. Lu, X.; Fu, Y.Y.; O’Neill, Z. Benchmarking High Performance HVAC Rule-Based Controls with Advanced Intelligent Controllers:
A Case Study in a Multi-Zone System in Modelica. Energy Build. 2023, 284, 112854. [CrossRef]
173. Naug, A.; Quinones-Grueiro, M.; Biswas, G. Deep Reinforcement Learning Control for Non-Stationary Building Energy Manage-
ment. Energy Build. 2022, 277, 112584. [CrossRef]
174. Blad, C.; Bogh, S.; Kallesoe, C. A Multi-Agent Reinforcement Learning Approach to Price and Comfort Optimization in HVAC-
Systems. Energies 2021, 14, 7491. [CrossRef]
175. Du, Y.; Li, F.X.; Munk, J.; Kurte, K.; Kotevska, O.; Amasyali, K.; Zandi, H. Multi-Task Deep Reinforcement Learning for Intelligent
Multi-Zone Residential HVAC Control. Electr. Power Syst. Res. 2021, 192, 106959. [CrossRef]
176. Wu, Z.Q.; Mu, Y.F.; Deng, S.; Wang, J.J.; Bai, Y.D.; Xue, J.; Li, Y.; Jiang, Y.T.; Zhang, X.D.; Xu, W.C. Towards Comfortable and
Cost-Effective Indoor Temperature Management in Smart Homes: A Deep Reinforcement Learning Method Combined with
Future Information. Energy Build. 2022, 275, 112491. [CrossRef]
177. Duan, W.J.; Wang, Y.; Li, J.Q.; Zheng, Y.J.; Ning, C.G.; Duan, P.Y. Real-Time Surveillance-Video-Based Personalized Thermal
Comfort Recognition. Energy Build. 2021, 244, 110989. [CrossRef]
Energies 2024, 17, 4277 35 of 35

178. Ding, Z.K.; Fu, Q.M.; Chen, J.P.; Wu, H.J.; Lu, Y.; Hu, F.Y. Energy-Efficient Control of Thermal Comfort in Multi-Zone Residential
HVAC via Reinforcement Learning. Connect. Sci. 2022, 34, 2364–2394. [CrossRef]
179. Wang, P.J.; Hu, J.H.; Chen, W.J. A Hybrid Machine Learning Model to Optimize Thermal Comfort and Carbon Emissions of
Large-Space Public Buildings. J. Clean. Prod. 2023, 400, 136538. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.