See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/328477567

Introduction to Load Forecasting

Article  in  International Journal of Pure and Applied Mathematics · October 2018

CITATION READS

1 1,246

4 authors:

Tahreem Anwar Bhaskar Sharma

Jayoti Vidyapeeth Women's University JK Lakshmipat University
1 PUBLICATION   1 CITATION    7 PUBLICATIONS   8 CITATIONS   

SEE PROFILE SEE PROFILE

Koushik Chakraborty Himanshu Sirohia

Jayoti Vidyapeeth Women's University Jayoti Vidyapeeth Women's University
10 PUBLICATIONS   1 CITATION    12 PUBLICATIONS   1 CITATION   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

PAPAR PROBLEM View project

All content following this page was uploaded by Himanshu Sirohia on 24 October 2018.

The user has requested enhancement of the downloaded file.

International Journal of Pure and Applied Mathematics
Volume 119 No. 15 2018, 1527-1538
ISSN: 1314-3395 (on-line version)
url: http://www.acadpubl.eu/hub/
Special Issue
http://www.acadpubl.eu/hub/

Introduction to Load Forecasting

1
Tahreem Anwar, 2Bhaskar Sharma, 3Koushik Chakraborty and
4
Himanshu Sirohia
1,2,3,4
Jayoti Vidyapeeth Women’s University,
Jaipur.

Abstract
Load forecasting is a technique used by power companies to predict the
power or energy needed to balance the supply and load demand at all the
times. It is mandatory for proper functioning of electrical industry. It can be
classified in terms of time like short-term (a few hours), medium-term (a
few weeks up to a year) or long-term (over a year). In this paper, for
medium and long term forecasting end use and econometric approach is
used. Whereas for short term forecasting various approaches are used like
regression models, time series, neural networks, statistical learning
algorithms and fuzzy logic.
Key Words:Load forecasting, regression models, time series, neural
networks, statistical learning algorithms and fuzzy logic.

1527
International Journal of Pure and Applied Mathematics Special Issue

1. Introduction
Electrical energy cannot be stored. It has to be generated whenever there is a
demand for it. It is, therefore, imperative for the electric power utilities that the
load on their systems should be estimated in advance. This estimation of load
in advance is commonly known as load forecasting. It is necessary for power
system planning. Power system expansion planning starts with a forecast of
anticipated future load requirements. There is a growing tendency towards
unbundling the electricity system. This is continually confronting the different
sectors of the industry (generation, transmission, and distribution) with
increasing demand on planning management and operations of the network. The
operation and planning of a power utility company requires an adequate model
for electric power load forecasting. Load forecasting succor an electric utility to
make conclusions regarding decision on generating and purchase electric power,
load switching, voltage control, network reconfiguration and infrastructure
development. Electric load forecasting is the process used for forecasting future
electric load, given historical load and weather information and current and
forecasted weather information. In the past few decades, several models have
been developed to forecast electric load more accurately. With an introduction
of deregulation of power industry, many new challenges have been encountered
by the participants of the electricity market. Forecasting of wind power, electric
loads and energy price have become a major issue in power systems. Following
needs of the market, various techniques are used to forecast the wind power,
energy price and power demand. The market risk related to trading is
considerable due to extreme volatility of electricity prices. Considering the
uncertain nature of future prices in competitive electricity markets, price
forecasts are used by market participants in their operation planning activities.
In addition, to ensure the secure operation of the power system at some future
time requires the study of its behavior under a variety of postulated contingency
conditions. Demand prediction is an important aspect in the development of any
model for electricity planning, especially in today’s reforming power system
structure [1]. The form of the demand depends on the type of planning and
accuracy that is required. The objective of this paper is to provide a brief
overview of various forecasting problems and techniques in power systems.
Depending on the time zone of planning strategies the load forecasting can be
divided into following three categories namely:
1. Short term load forecasting: this forecasting method is usually has
period ranging from one hour to one week. It can guide us to
approximate load flow and to make decisions that can intercept
overloading. Short term forecasting is used to provide obligatory
information for the system management of daily operations and unit
commitment.
2. Medium term load forecasting: this forecasting method has its period
ranging from one week to one year. The forecasts for different time
horizons are important for different operations within a utility company.

1528
International Journal of Pure and Applied Mathematics Special Issue

Medium term forecasting is used for the purpose of scheduling fuel

supplies and unit management.
3. Long term load forecasting: this forecasting method has its period
which is longer than a year. It is used to supply electric utility company
management with précised prediction of future needs for expansion,
equipment purchase or staff hiring.

To execute different operations within a utility company, the forecast for

different time horizon needed to be employed. The behavior of these forecasts is
different as well. For instance, in a particular region it may be possible to
predict coming day load with an accuracy of approximately 1-3%. Moreover, it
is almost impossible to predict the peak load of next year with the similar
accuracy since the prediction for long term forecasting may lack precision
making this forecast unavailable. But, for the next year peak forecast, it is
possible to produce the probability distribution load based on past weather
observations. According to the industry practice, it is also possible to predict the
so-called weather normalized load, which would take place for average annual
peak weather conditions or worse than average peak weather conditions for a
given area. The load calculated for the normal weather conditions which are the
average of the weather is called weather normalized load. Further, the
techniques employed by most of the forecasting methods are statistical
techniques or artificial intelligence algorithms such as regression, neural
networks, fuzzy logic, and expert systems. The two methods which are broadly
used for medium-term and long-term forecasting are end-use and econometric
approach. And for short-term forecasting, a variety of methods, which include
the so-called similar day approach, various regression models, time series,
neural networks, fuzzy logics and expert systems have been developed. A wide
variety of mathematical methods and ideas have been used for load forecasting.
The development and refinements of appropriate mathematical tools will lead to
development of more accurate techniques for load forecasting. The accuracy is
also the matter of forecasted weather scenarios; it doesn’t solely depend upon
load forecasting techniques. Weather forecasting is the application of science
and technology to predict the conditions if atmosphere for a given location and
time.

2. Basics of Forecasting
Forecasting is a Stochastic Problem

Forecasting, by nature, is a stochastic problem rather than being deterministic in

nature. There is no "certain" in forecasting. Since the forecasters are dealing
with randomness, the output of a forecasting process is supposed to be in a
probabilistic form, such as a forecast under this or that scenario, a probability
density function, a prediction interval, or some quantile of interest. In practice,
probabilistic inputs cannot be taken in many decision making processes, so the
most commonly used forecasting output form is still point forecast, e.g., the
future expected value of a random variable.

1529
International Journal of Pure and Applied Mathematics Special Issue

All Forecasts are Not Accurate

The response variable to forecast is never 100% predictable due to the

stochastic nature of forecasting. Bad data, inappropriate methodologies, and
crappy software, etc are some other factors that may lead us to wrong forecasts.
It's the forecaster's job to apply best practices to avoid these preventable issues.
Some Forecasts are Useful

The two most common aspects of usefulness in the utility industry are accuracy
and defensibility. Accuracy can be calculated based on various peaks (i.e.,
monthly, seasonal or annual peaks), energy or the combination of them.
Defensibility may include interpretability, traceability, and reproducibility. The
points above needs to be prioritized differently depending upon the exact
business need. For example, for regulatory compliance purpose, we would
emphasize defensibility more than the accuracy. As a consequence, statistical
approaches such as multiple linear regressions are usually preferred over black-
box approaches like Artificial Neural Networks.
Forecasts can be Improved

Since all forecasts are not always correct, there is always way for improvement,
at least from the accuracy point of view. Generally speaking, the objective of
forecast improvement is to enhance the utility. For potential improvement there
are some more specific directions which can be adopted:
a) Spread of errors. Nobody wants to have surprisingly big errors.
Reducing the variance or range of the errors means reducing the
uncertainty, which consequently increases the utility of the forecasts.
Sometimes the business may even give up some central tendency of the
error (e.g., MAPE), to gain reduction in the spread (e.g., standard
deviation of APE).
b) Interpretability of errors. For instance, due to uncertainty in long term
weather and economy forecasts in long term load forecasting, the load
forecasts may result in some significant errors from time to time. Then
the forecasters should help the business users to understand how much
of the error is contributed by modeling error, weather forecast error and
economy forecast error. Breaking down the error to its sources increases
the interpretability as well as the utility of forecasts.
c) Requirement of resources. In reality, we always have limited resources
like data, hardware and labor to build a forecast. If we can enhance the
simplicity of the forecasting process by reducing the requirements on
these resources, it can be of use for the business side.
Accuracy is Never Guaranteed

Due to the unpredictable nature of forecasting, the future will never redo the
history in exactly the manner described by our models. Sometimes, the
deviations are large; sometimes, they are small. Even if a forecaster could
maintain a stable accuracy during the past a few years, there is still no guarantee
that the similar accuracy can be attained going forward. Sometimes the

1530
International Journal of Pure and Applied Mathematics Special Issue

consultants and vendors promise impractical accuracy to the clients in order to

vend the services or solutions. This is one of the worst practices, because
eventually the clients will realize that the error is not as low as what's been
promised.
Having the Second Opinion is Preferred

There is never a perfect model. If only one model is being used, the forecaster
will experience ”bad” forecasts from time to time and the predictions and
outcomes will become worse. If multiple models are available, the situation can
be completely different. The forecaster will have good outcomes when the
models will agree to each other. Forecaster will be able to focus on the periods
when these models disagree with each other significantly. Factually, combining
forecasting techniques usually does a better job than each individual, offering
more robust and accurate forecasts. Therefore, one of the best practices is to run
multiple models and combine the forecasts.

3. Important Issues in Load Forecasting

Forecasting Load

Forecasting electricity loads has reached the state of maturity. Short-term (a few
minutes, hours, or days ahead) to the long term (up to 20 years ahead) forecasts
have become increasingly important since the restructuring of power systems.
Many countries have recently follow isolationism and privatized their power
systems, and electricity has been turned into an important commodity available
at market prices. Load forecasting is an arduous task. First, because the load
series is complex and exhibits several levels of seasonality like the load at a
given hour is dependent not only on the previous hour loads, but also on the
load at the same hour on the previous day, and on the same hour load on the day
with the same value in the previous week. Secondly, because there are many
important exogenous variables that must be considered, specially weather
related variables [1] [2]. These issues can be resolved by using various models
and methodologies like autoregressive models, dynamic linear or nonlinear
models, fuzzy inference, fuzzy-neural models, Box and Jenkins transfer
functions ARMAX models, , neural network (NN) etc.
Price Forecasting

Forecasting loads and prices in electricity markets are mutually intertwined

activities, and error in load forecasting will propagate to price forecasting.
Electricity price has its special characteristics. There are at least three main
features that make it so specific. One of them is its non-storability of power,
which means that prices are strongly dependent on the power demand. Another
characteristic is the seasonal behavior of the electricity price at different level
(daily, weekly and annual seasonality) and the third one is its questionable
transportability. In today’s most competitive electricity markets, the hourly
price series have the characteristics such as volatility, non-stationary properties,
multiple seasonality, spikes and high frequency. A price spike can be caused by

1531
International Journal of Pure and Applied Mathematics Special Issue

market power which is a randomized event, and also by unexpected events such
as transmission congestion, transmission contingency and generation
contingencies. It can also be affected by other factors such as fuel prices,
generation unit operation costs, weather conditions, and probably the most
theoretically significant factor, the balance between overall system supply and
demand. Applications of electricity price forecasting fall into different time
horizons which are short-term forecasting, medium-term forecasting and long-
term forecasting. In order to maximize their profits in spot markets, market
participants need to forecast short-term (mainly one day-ahead) prices. And for
successful negotiations of bilateral contracts between suppliers and consumer,
accurate medium term price forecasts are necessary. The decisions on
transmission expansion and enhancement, generation augmentation, distribution
planning and regional energy exchange are influenced by Long-term price
forecasts. The models which are used to resolve these issues are statistical and
non-statistical models. Time-series models, econometric models and intelligent
system methods are the three main categories of statistical methods. Non-
statistical methods include equilibrium analysis and simulation methods.
Forecasting Wind Power

Wind-generated power constitutes a noticeable percentage of the total electrical

power consumed and in some utility areas it even exceeds the base load on the
network. This indicates that wind is becoming a major factor in electricity
supply and in balancing consumer demand with power production. To
integration of wind power into the grid is its variability is its major barrier.
Because of its dependence on the weather, the output cannot be guaranteed at
any particular time. These issues can be solved by techniques with simple
persistence approach, classical linear statistical models such as Moving Average
(MA), Auto-Regressive Moving Average (ARMA) and the Box-Jenkins
approach based on Auto-Regressive Integrated Moving Average (ARIMA) or
seasonally adjusted ARIMA models, also known as SARIMA models.

Fig. 1: Block Diagram of Load Forecasting Issues

1532
International Journal of Pure and Applied Mathematics Special Issue

4. Methodologies for Load Forecasting

Medium/Long-term Load Forecasting Methods

End-use and Econometric approach are two methodology used for medium- and
long-term forecasting.
End-use Approach

The end-use approach directly estimates energy consumption by using extensive

information on end use and end users, such as appliances, the customer use,
their age, sizes of houses, and so on. End-use models focus on the various uses
of electricity in the residential, commercial, and industrial sector. These models
are based on the principle that electricity demand is derived from customer's
demand for light, cooling, heating, refrigeration, etc. Thus end-use models
explain energy demand as a function of the number of appliances in the market.
Econometric Approach

The econometric approach combines economic theory and statistical techniques

for forecasting electricity demand. The approach estimates the relationships
between energy consumption (dependent variables) and factors influencing
consumption.
Short–term Forecasting Methods

Many methods which include the so-called similar day approach are various
regression models, time series, neural networks, expert systems, fuzzy logic,
and statistical learning algorithms, are used for short-term forecasting.
Similar-day Approach

This approach is based on searching historical data for days within one, two, or
three years with similar characteristics to the forecast day. Similar
characteristics include weather, day of the week, and the date.
Regression Methods

Regression is the one of most widely used statistical techniques. For electric
load forecasting regression methods are usually used to model the relationship
of load consumption and other factors such as weather, day type, and customer
class.
Time Series

It has been used for decades in such fields as economics, digital signal
processing, as well as electric load forecasting. In particular, ARMA
(autoregressive moving average), ARIMA (autoregressive integrated moving
average), ARM\AX (autoregressive moving average with exogenous variables),
and ARIMAX (auto regressive integrated moving average with exogenous
variables) are the most often used classical time series methods.

1533
International Journal of Pure and Applied Mathematics Special Issue

Neural Networks

Neural networks are essentially non-linear circuits that have the demonstrated
capability to do non-linear network are some linear or nonlinear mathematical
function of its inputs[20].The inputs may be the outputs of other network
elements as well as actual network inputs. In practice network elements are
arranged in a relatively small number of connected layers of elements between
network inputs and outputs. Feedback paths are sometimes used. In applying a
neural network to electric load forecasting, on must select one of a number of
architectures (e.g. Hopfield, back propagation, Boltzmann machine), the
number and connectivity of layers and elements, use of bi-directional or uni-
directional links, and the number format (e.g. binary or continuous) to be used
by inputs and outputs, and internally. The most popular artificial neural network
architecture for electric load forecasting is back propagation.
Fuzzy Logic

Fuzzy logic is a generalization of the usual Boolean logic used for digital circuit
design. Under fuzzy logic an input has associated with it a certain qualitative
ranges. For instance a transformer load may be "low", "medium" and "high".
Fuzzy logic allows one to deduce outputs from fuzzy inputs. In this sense fuzzy
logic is one of a number of techniques for mapping inputs to outputs. Among
the advantages of fuzzy logic is the absence of a need for a mathematical model
mapping inputs to outputs and the absence of a need for precise or even noise
free inputs [8][9].
Expert Systems

Expert systems are known to apply rules which have been generated by experts
in the field. These procedures and rules are then transformed into software,
which can automatically make forecasts about electricity demand. However, in
order for the software to be efficient, expert forecasters will have to work hand
in hand with software developers to include all expert information into the
software. Here, software developers will be involved in coding all the
information obtained from experts in forming software rules. The advantage of
utilizing this method is that it is fast, accurate and easy to use when forecasting
information about electricity demand. This is because it involves the use of
software applications, which usually run at the click of a button.

5. Advantages & Disadvantages of

Load Forecasting
Advantages
1) It enables the utility company to plan well since they have an
understanding of the future consumption or load demand.
2) Useful to determine the required resources such as fuels required to
operate the generating plants as well as other resources that are required
to ensure uninterrupted and yet economical generation and distribution

1534
International Journal of Pure and Applied Mathematics Special Issue

of the power to the consumers. This is important for all short, medium,
and long term planning.
3) Planning the future in terms of the size, location and type of the future
generating plant are the factors which are determined by the help of load
forecasting.
4) Provides maximum utilization of power generating plants. The
forecasting avoids under generation or over generation.
Disadvantages
1) It is not possible to forecast the future with accuracy. The qualitative
nature of forecasting, a business can come up with different scenarios
depending upon the interpretation of the data.
2) Organizations should never rely 100 percent on any forecasting method.
However, an organization can effectively use forecasting with other
tools of analysis to give the organization the best possible information
about the future.
3) Making a decision based on a bad forecast can result in financial ruin for
the organization, so the decisions of an organization should never base
solely on a forecast.

6. Future Research Directions

In this paper we have discussed several statistical and artificial intelligence
techniques that have been developed for short-term, medium-term, and long-
term electric load forecasting. Several statistical models and algorithms that
have been developed though are operating ad hoc. The accuracy of the forecasts
could be improved, if one would study these statistical models and develop
mathematical theory that explains the convergence of these algorithms.
Researchers should also investigate the boundaries of applicability of the
developed models and algorithms. So far, there is no single model or algorithm
that is superior for all utilities. Selecting the most suitable algorithm by a utility
can be done by testing the algorithms on real data. In fact, some utility
companies use several load forecasting methods in parallel. As far as we know,
nothing is known on a priori conditions that could detect which forecasting
method is more suitable for a given load area. An important question is to
investigate the sensitivity of the load forecasting algorithms and models to the
number of customers, characteristics of the area, energy prices, and other
factors. There is also a clear move towards hybrid methods, which combine two
or more of these techniques we think that the important research and
development directions are:
i. Combining weather and load forecasting and
ii. Incorporating load forecasting into various decision support systems.
7. Conclusion
In this paper we have reviewed some statistical and artificial intelligence
techniques that are used for electric load forecasting. We also discussed basics
and factors that affect the accuracy of the forecasts. Different techniques are

1535
International Journal of Pure and Applied Mathematics Special Issue

applied to load forecasting. It can be inferred that demand forecasting

techniques based on soft computing methods are gaining major advantages for
their effectual use. There is also a clear move towards hybrid methods, which
combine two or more of these techniques. The research has been shifting and
replacing old approaches with newer and more efficient ones.

References
[1] Hippert H.S., Pedreira, C.E., Souza R.C., Neural networks for
short-term load forecasting: A review and evaluation, IEEE
Trans. On Power Syst. 16(1) (2001), 44−55.
[2] Senjyu T., Mandal P., Uezato K., Funabashi T., Next day load
curve forecasting using hybrid correction method, IEEE Trans.
On Power Syst. 20(1) (2005), 102−109.
[3] Peng M., Hubele N.F., Karady G.G., Advancement in the
Application ofNeural Networks for Short-Term Load Forecasting,
IEEE Transactions on Power Systems 7 (1992), 250-257.
[4] Dudhia J., Bresch J.F., A Global Version of the PSUNCAR
Mesoscale Model, Monthly Weather Review 130 (2002).
[5] Willis H.L., Spatial Electric Load Forecasting, Marcel Dekker,
New York (1996).
[6] Chen H., Canizares C.A., Singh A., ANN-Based Short-Term
Load Forecasting in Electricity Markets, Proceedings of the IEEE
Power Engineering Society Transmission and Distribution
Conference 2 (2001), 411-415.
[7] Rahman S., Formulation and Analysis of a Rule-Based Short-
Term Load Forecasting Algorithm, Proceedings of the IEEE 78
(1990), 805-816.
[8] Gellings C.W., Demand Forecasting for Electric Utilities, The
Fairmont Press, Lilburn, GA (1996).
[9] Miranda V., Monteiro C., Fuzzy Inference in Spatial Load
Forecasting, Proceedings of IEEE Power Engineering Winter
Meeting 2 (2000), 1063-1068.
[10] Vapnik V.N., The Nature of Statistical Learning Theory. New
York, Springer Verlag (1995).
[11] Mohandes M., Support Vector Machines for Short-Term
Electrical Load Forecasting, International Journal of Energy
Research 26 (2002), 335-345.
[12] Chen B.J., Chang M.W., Lin C.J., Load Forecasting using
Support Vector Machines: A Study on EUNITE Competition
2001, Technical report, Department of Computer Science and
Information Engineering, National Taiwan University (2002).

1536
International Journal of Pure and Applied Mathematics Special Issue

[13] Li Y., Fang T., Wavelet and Support Vector Machines for Short-
Term Electrical Load Forecasting, Proceedings of International
Conference on Wavelet Analysis and its Applications 1 (2003),
399-404.
[14] Taylor J.W., Buizza R., Using Weather Ensemble Predictions in
Electricity Demand Forecasting, International Journal of
Forecasting 19 (2003), 57-70.
[15] Taylor J.W., Buizza R., Neural Network Load Forecasting With
Weather Ensemble Predictions, IEEE Transactions on Power
Systems 17 (2002), 626-632.

1537
 1538

View publication stats