CHAPTER 1

INTRODUCTION

Precision agriculture is a farming management concept based on observing, measuring

and responding to interand intra-field variability in crops. The goal of precision

agriculture research is to define a decision support system (DSS) for whole farm

management with the goal of optimizing returns on input while preserving resources. The

farmers and researchers ability to locate their precise position in a field allows for the

creationof maps of the spatial variability of as many variables as can be measured(e.g.

crop yield, terrain features/ topography, organic matter content, moisture levels, nitrogen

levels, pH, EC, Mg, K and others)

The Internet of Things is the network of physical objects outfitted with electronics that

enables data collection and aggregation. IoT comes into play with the development of

senor and farm-management software. Farmers can measure phosphorous, potassium, and

nitrogen by using sensors or spectroscopically. Moisture sensor in he soil determines the

best times to remotely water plants.

Linear Regression is a supervised machine learning algorithm where the predicted

continuous and has a constant slope. It is used to predict the values within a continuous

range, rather than trying to classify them into categories. Linear Regression uses

traditional slope-intercept form. Electrical Conductivity (EC) can be an indirect indicator

of soil nutrient concentration, and strong relationships between EC and nutrients may

help fertility management zones.

1
Deep Learning is class of Machine Learning algorithms that use multiple layers to

progressively extract higher level features from raw input. Deep learning is part of a

border family of machine learning methods based on artificial neural networks. Learning

can be supervised, semi-supervised or unsupervised. In deep learning a computer model

learns to perform classification tasks directly from images, text, or sound. Deep learning

models can achieve state-of-the-art accuracy, sometimes exceeding human-level

performance. Models are trained by using a large set of labelled data and neural network

architectures that contain many layers.

ANN is a layered approach to processing information and making decisions that ANNs

are trying to simulate. In its simplest form, an ANN can have only three layers of

neurons: the input layer (where the data enters the system), the hidden layer (where the

information is processed) and the output layer (where the system decides what to do

based on the data). But ANNs can get much more complex than that, and include multiple

hidden layers.

Deep learning represents the very cutting edge of artificial intelligence. Instead of

teaching computers to process and learn from data (which is how machine learning

works), with deep learning, the computer trains itself to process and learn from data. A

deep learning system is self-teaching, learning as it goes by filtering information through

multiple hidden layers, in a similar way to humans. Massively parallel computation

through the use of GPU scalable for large volume of data.

2
 CHAPTER 2

SOFTWARE PROJECT PLAN

S.No Phases Description Time Bound Deliverables Remarks

1. Project Identification of July 2019 Abstract Identification

Initialization Project activity of project
and responsible mentor, chief
person architect and
technology
familiarization
2. Hardware Hardware July 2019 Resource Got the
Requirement Requirements are estimation hardware
documented resources.
Estimation

3. Software Software August 2019 Sensor values IoT is treated as

Requirement requirements are are measured baseline
documented
Estimation

4. Software Convert analysis August 2019 Phosphorous Activities

Development model into design estimation carried out by
model using Deep
Learning all team
members

5. Software Software is September Product Development

Deployment implemented 2019 installation carried out

3
6. Prediction Time series September Phosphorous Further
Forecasting is 2019 content is developed for
used for measured for future
Prediction future enhancement

7. Software Risk management October 2019 Risk Identification of

configuration associated potential risk
and Risk
management

8. Software Formation of October 2019 Solution for Update is done

Maintenance maintenance team the problems
and plans identified

Table 2.1 software project plan

4
 CHAPTER 3

PROBLEM STATEMENT

Precision agriculture is an emerging farm management strategy that is changing the way

people farm. The key factor of soil is Phosphorous. Phosphorous is an essential nutrient

both as a part of several key plant structure compounds and as a catalysis in the

conversion of numerous key biochemical reactions in plants. Phosphorous deficiency is

more difficult to diagnose than a deficiency of nitrogen or potassium. A wide range of IoT

and Deep Learning technique is used to build a more accurate system. To measure the

phosphorus, methods available only based on lab assisted techniques and time

consuming.So we need to develop a handy method for assess the phosphorus value based

other soil parameters such as,Soil temperature, Soil humidity, Soil moisture. The deep

learning platform reads the data and predict the phosphorous content.

5
 CHAPTER 4

SOFTWARE REQUIRMENTS SPECIFICATION

4.1 INTRODUCTION

4.1.1 PURPOSE

To estimate the phosphorous content in the soil

4.1.2 DOCUMENT CONVENTIONS

This document is written in the following style

Font style : Times New Roman

Chapter Headings : 14 pts

Sub-Heading : 12 pts

Sub-Section Heading : 12 pts

Description : 12 pts

4.1.3 INTENDED AUDIENCE AND READING SUGGESTION

This document is intended for the developers. This SRS contains the purpose of the

proposed method and the diagrams explained for reference. The table also included to

clearly explain the use cases and sequence of the proposed method. And also it explains

about the merits and future enhancement of the propose method

6
4.1.4 PROJECT SCOPE

The scope of the proposed system method is to estimate the phosphorous content present

in the soil using Deep Learning

4.2 OVERALL DESCRIPTION

4.2.1 PRODUCT PERSPECTIVE

A method is proposed that uses deep learning based technique to measure the

phosphorous content in the soil. Linear Regression is used to measure Electrical

Conductivity. Deep Learning (Back Propagation) is used for training and testing purpose.

Using the state of the art considerable efficiency upto 90 percentage can be achieved.

4.2.2 PRODUCT FUNCTIONS

 NodeMCU Development board

 DHT11 Sensor
 Soil Moisture Sensor

4.2.3 OPERATING ENVIRONMENT

It is created with the help of GOOGLE COLAB . This product can be changed with this

IDE and updated version using windows OS 7, 8, 10 by any personal computer, and has

been tested for compliance with higher versions.

4.3 OTHER FUNCTIONAL REQUIREMENTS

4.3.1 Software Quality Attributes

Maintainability

 It does not require maintenance.

7
Usability

 Goals are easy to accomplish quickly and no user error.

Availability

 There is no downtime acceptable in this proposed

method

Hardware Requirements:

 •
Processor - Intel Pentium

Speed - 2.1 GHz

RAM - 4GB(min)

Hard Disk - 40 GB

Key Board - Standard Windows

Keyboard

Mouse - Two or Three

Button Mouse

Monitor - SVGA

4.3.2 HARDWARE REQUIREMENTS

4.3.2.1NodeMCU Development Board

NodeMCU is an open source IoT based firmware developed for ESP8266 Wi-Fi chip. By

exploring functionality with ESP8266 chip, NodeMCU firmware comes with ESP8266

8
Development board/kiti.e. NodeMCU Development board. Since NodeMCU is open

source platform, their hardware design is open for edit / modify / build. NodeMCU Dev

Kit/board consist of ESP8266 Wi-Fi enabled chip. The ESP8266 is a low-cost Wi-Fi chip

developed by Espressif Systems with TCP/IP protocol. For more information about

ESP8266, you can refer ESP8266 Wi-Fi Module. There is Version2 (V2) available for

NodeMCU Dev Kit i.e. NodeMCU Development Board .

Fig 4.1 NodeMCU Development Board

4.3.2.2 DHT11 Sensor

The digital temperature and humidity sensor DHT11 is a composite sensor that contains a

calibrated digital signal output of temperature and humidity. The technology of a

dedicated digital modules collection and thetemperature and humidity sensing technology

9
are applied to ensure that the product has high reliability and excellent long-term stability.

The sensor includes a resistive sense of wet component and an NTC

temperaturemeasurement device, and is connected with a high-performance 8-bit

microcontroller. DHT11 is a low-costdigital sensor. This sensor can be easily interfaced

with any micro-controller such as Arduino, Raspberry Pietc.,The temperature range of

DHT11 is from 0 to 50 degree Celsius with a 2-degree accuracy.Humidity range of this

sensor is from 20 to 80% with 5% accuracy.

Fig 4.2 DHT11 Sensor

4.3.2.3 Soil Moisture Sensor

Soil moisture is basically the content of water present in the soil. This can be measured

using a soilmoisturesensor which consists of two conducting probes that act as a probe. It

can measure the moisture content in the soil based on the change in resistance between

10
the two conducting plates.When there is more water,the soil will conduct more electricity

which means that there will be less resistance. Therefore, the moisture levelwill be higher.

Dry soil conducts electricity poorly, so when there will be less water, then the soil will

conductless electricity which means that there will be more resistance. Therefore, the

moisture level will be lower. This sensor reminds the user to water their plants and also

monitors the moisture content of soil. It has been widely used in agriculture, land

irrigation and botanical gardening.

Fig 4.3 Soil Moisture Sensor

11
 CHAPTER 5

SYSTEM ANALYSIS

5.1 DATA FLOW

Fig 5.1 Data Flow Diagram

12
5.2 USECASE DIAGRAM

Fig 5.2 Usecase diagram.

13
5.3 ARCHITECTURE DIAGRAM

Fig 5.3 Architecture diagram

14
 CHAPTER 6

CODING

6.1 ALGORITHM

//LINEAR REGRESSION

Step 1: Start

Step 2: Declare arrays x, y

Step3: Initialize array xx1, x2, x3,….,xn

Initializearray yy1, y2, y3,…, yn

Step 4: Plot x and y values as scatter plot

Step 5: Calculate intercept

step 6: calculate co-efficient of slope

mmean(y)-b*mean(x)

step7: calculate predicted y values

ypredmx+b

step 8: calculate error between actual y values Vs predicted y values

15
Step 9: adjust the slope coefficient to minimize RMSE error

Step 10: again find the new y predicted values

Step 11: plot the line with new predicted y values against x values

Step 12: Stop

16
6.2 PROGRAMS

6.2.1 ARDUINO

#include <dht.h>

#include <ESP8266WiFi.h>

#include <ESP8266WiFiAP.h>

#include <ESP8266WiFiType.h>

#include <WiFiClient.h>

#include <WiFiClientSecureBearSSL.h>

#include <WiFiServer.h>

#include <WiFiUdp.h>

#include "ThingSpeak.h"

#include "secrets.h"

#include <ESP8266WiFi.h>

Initialize your network SSID (name)

Initialize your network password

Your network key Index number (needed only for WEP)

WiFiClient client;

unsigned long myChannelNumber = SECRET_CH_ID;

const char * myWriteAPIKey = SECRET_WRITE_APIKEY;

dht DHT;

17
#define DHT11PIN 5

Initialize our values

void setup() {

Initialize serial

WiFi.mode(WIFI_STA);

Initialize ThingSpeak

void loop() {

Connect or reconnect to WiFi

If Wi-Fi is connected

Print the SSID

If it is not connected

Connect to WPA/WPA2 network. Change this line if using open or WEP network

delay(5000);

Print connected

intchk = DHT.read11(DHT11PIN);

readsoil_moisture frompin A0

read the temperature value

readthe.humidity value

18
Print the temperature

Print the humidity

Print the soil moisture

delay(1000);

set the fields with the values

write to the ThingSpeak channel

if (x == 200) {

print channel is updated successfully

else {

print there is an error

Wait 20 seconds to update the channel again

SECRET FILE

replaceMySSID with your WiFi network name

replaceMyPassword with your WiFi password

replace 0000000 with your channel number

replace XYZ with your channel write API Key

19
6.2.2 LINEAR REGRESSION

fromsklearn.linear_model import LinearRegression

fromsklearn.model_selection import train_test_split

import pandas as pd

importmatplotlib.pyplot as plt

read moisture.csv file

print the values

sample=df['SAMPLES'].values

sample=sample.reshape(-1,1)

print sample

digital=df['DIGI'].values

digital=digital.reshape(-1,1)

print digital

spilt the data into train and test data

call the regression function

model = regression.fit(sample,digital)

print the model.intercept

print the model.coef

y_pred = regression.predict(X_test)

find the absolute of the predicted value

plot the values in graph

plt.show()

20
Fig 6.1 Output of Linear Regression

21
6.2.3 DEEP LEARNING

importmatplotlib.pyplot as plt

importnumpy as np

import pandas as pd

fromsklearn.model_selection import train_test_split

fromsklearn.preprocessing import MinMaxScaler

fromtensorflow.python.keras.models import Sequential

fromtensorflow.python.keras.layers import Dense

read the'P_EST.csv an initialize it to data variable

x=dataset.iloc[:,:-1].values

y=dataset.iloc[:,-1].values

print the dataset

print(x)

y=y.reshape(-1,1)

scaler_x = MinMaxScaler()

scaler_y = MinMaxScaler()

print(scaler_x.fit(x))

xscale=scaler_x.transform(x)

print(scaler_y.fit(y))

yscale=scaler_y.transform(y)

X_train, X_test, y_train, y_test = train_test_split(xscale, yscale)

model = Sequential()

model.add(Dense(12, input_dim=3, kernel_initializer='normal', activation='relu'))

22
model.add(Dense(8, activation='relu'))

model.add(Dense(1, activation='linear'))

model.summary()

model.compile(loss='mse', optimizer='adam', metrics=['mse','mae'])

history = model.fit(X_train, y_train, epochs=500, batch_size=50,

verbose=1,validation_split=0.25)

print(history.history.keys())

# "Loss"

plt.plot(history.history['loss'])

plt.plot(history.history['val_loss'])

plt.title('model loss')

plt.ylabel('loss')

plt.xlabel('epoch')

plt.legend(['train', 'validation'], loc='upper left')

plt.show()

Xnew =

np.array([[0.3000,16.0000,12.0000],[0.3181,16.3289,12.5570],[0.3362,16.6577,13.1141],

[0.3544,16.9866,13.6711],[0.3725,17.3154,14.2282],[0.3906,17.6443,14.7852],[0.4087,1

7.9732,15.3423],[0.4268,18.3020,15.8993],[0.4450,18.6309,16.4564],[0.4631,18.9597,17

.0134],[2.2752,51.8456,72.7181], [3,65,95]])

ynew = (model.predict(Xnew))

ynew=ynew/2.0

print(ynew)

23
fig 6.2 Output of Deep Learning

24
6.2.4 TIME SERIES FORECASTING

import pandas as pd

importseaborn as sns

importmatplotlib.pyplot as plt

fromstatsmodels.tsa.arima_model import ARIMA

fore = pd.read_csv('forcast.csv',

parse_dates=['Month'],

index_col='Month',

date_parser=lambda d: \

pd.datetime.strptime(d, '%Y-%m'))

print(fore)

fromstatsmodels.tsa.seasonal import seasonal_decompose

fore_decomp = seasonal_decompose(fore, freq=12)

fore_trend = fore_decomp.trend

fore_seasonal = fore_decomp.seasonal

fore_residual = fore_decomp.resid

ts = fore-fore_seasonal

tsdiff = ts.diff(1)

model = ARIMA(ts, order=(1, 1, 0))

mares = model.fit()

mares.plot_predict(start='2011-12-01', end='2050-01-01',alpha=0.10)

plt.legend(loc='upper left')

sns.despine(ax=plt.gca())

plt.xlabel('Year')print(mares.aic, mares.bic)

25
Fig 6.3 Output of Time Series Casting

26
 CHAPTER 7

TESTING

7.1 UNIT TESTING

Unit testing involves the design of test cases that validate that the internal program logic

is functioning properly, and that program inputs produce valid outputs. All decision

branches and internal code flow should be validated. It is the testing of individual

software units of the application .it is done after the completion of an individual unit

before integration. This is a structural testing, that relies on knowledge of its construction

and is invasive. Unit tests perform basic tests at component level and test a specific

business process, application, and/or system configuration. Unit tests ensure that each

unique path of all business process

performs accurately to the documented specifications and contains clearly defined inputs

and expected results.

TEST ID UTST-01
UNIT TESTED Input values (EC, temperature, humidity)
PURPOSE To identify the Phosphorous content
PRE-REQUIREMENTS Should be in .csv format
TEST DATA Set of sensor values and EC
TEST STATUS Pass
TEST RESULT Phosphorous content has successfully
measured

Table 7.1 Unit Testing

27
7.2 INTEGRATION TESTING

Integration tests are designed to test integrated software components to determine if

theyactually run as one program. Testing is event driven and is more concerned with the

basic outcomeof screens or fields. Integration tests demonstrate that although the

components were individuallysatisfaction, as shown by successfully unit testing, the

combination of components is correct andconsistent. Integration testing is specifically

aimed at exposing the problems that arise from thecombination of components.

7.3 FUNCTIONAL TESTING

Functional tests provide systematic demonstrations that functions tested are available as

specified by the business and technical requirements, system documentation, and user

manuals.

Functional testing is centred on the following items:

Valid Input: identified classes of valid input must be accepted.

Invalid Input: identified classes of invalid input must be rejected.

Functions: identified functions must be exercised.

Output: identified classes of application outputs must be exercised.

Systems/Procedures: interfacing systems or procedures must be invoked.

Organization and preparation of functional tests is focused on requirements, key

functions, orspecial test cases. In addition, systematic coverage pertaining to identify

Business process flows;data fields, predefined processes, and successive processes must

be considered for testing. Beforefunctional testing is complete, additional tests are

identified and the effective value of current tests is determined.

28
 CHAPTER 8

CONCLUSION AND FUTURE WORK

The proposed method uses Deep Learning techniques which gives improved efficiency

and accuracy.The efficiency and accuracy in predicting the PHOSPHOROUS CONTENT

is very much effective in this architecture.The technology growth paves a major role in

this area. This strategy considers that soil respiration rates during the day are affected by

temperature, humidity and conductivity changes, and are therefore linked to P

sorption/desorption, in addition to other biological activities. Using time series

forecasting method the feature phosphorus also planned to predict. Cost wise low priced

sensors are used to sense the soil parameters.

29
 CHAPTER 9

REFERENCES

1. P. McLoud, R. Gronwald, and H. Kuykendall, “Precision Agriculture:NRCS Support

for Emerging Technologies,” Natural Resources Conservation Service (NCRS),

Greensboro, NC, Tech. Rep. Agronomy Tech. Note 1, Jun. 2007.

2. R. ongiovanni and J. Lowenberg-Deboer, “Precision Agriculture and Sustainability,”

Precision Agriculture, vol. 5, no. 4, pp. 358–387, Aug. 2004.

3. J. I. Huircán, C. Muñoz, H. Young, L. Von Dossow, J. Bustos, G. Vivallo, and M.

Toneatti, “Zigbee-based wireless sensor network localization for cattle monitoring in

grazing fields,” Computers and Electronics in Agriculture, vol. 74, no. 2, pp. 258–264,

2010.

4. R. W. Coates, M. J. Delwiche, A. Broad, and M. Holler, “Wireless Sensor Network

with Irrigation Valve Control,” Computers and electronics in agriculture, vol. 96, pp. 13–

22, 2013.

5. J. Gutiérrez, J. F. Villa-Medina, A. Nieto-Garibay, and M. A. PortaGándara,

“Automated irrigation system using a wireless sensor network and GPRS module,” IEEE

Trans. Instrum. Meas., vol. 63, no. 1, pp.166–176, Jan. 2014.

6. F. Viani, M. Bertolli, M. Salucci, and A. Polo, “Low-cost wireless monitoring and

decision support for water saving in agriculture,” IEEE Sensors J., vol. 17, no. 13, pp.

4299–4309, Jul. 2017.

30
7. R. G. Vieira, A. M. da Cunha, L. B. Ruiz, and A. P. de Camargo, “On the design of a

long range WSN for precision irrigation,” IEEE Sensors J,, vol. 18, no. 2, pp. 773–780,

Jan. 2018.

8. D. García-Lesta, D. Cabello, E. Ferro, P. López, and V. M. Brea,

“Wireless sensor network with perpetual motes for terrestrial snail activity monitoring,”

IEEE Sensors J., vol. 17, no. 15, pp. 5008–5015, Aug. 2017.

9. F. Viani, M. Bertolli, and A. Polo, “Low-cost wireless system for agrochemical dosage

reduction in precision farming,” IEEE Sensors J., vol.17, no. 1, pp. 5–6, Jan. 2017.

10. T. Ruan, Z. J. Chew, and M. Zhu, “Energy-aware approaches for energy harvesting

powered wireless sensor nodes,” IEEE Sensors Journal,vol. 17, no. 7, pp. 2165–2173,

April 2017.

31
32