SDP MID-TERM EVALUATION

[Forecasting Hydro Power Energy Generation In India

Using Machine Learning Models]

Group No. N8: Supervised By: DR. PRAVAT KUMAR JENA

ANKIT NAVRANG MESHRAM [2041013065]
KHITISH ADHIKARI [2041001018] Department of Computer Sc. and Engineering
Faculty of Engineering & Technology (ITER)
AMANDEEP [2041001047] Siksha ‘O’ Anusandhan (Deemed to be) University
Bhubaneswar, Odisha
SUBRA PRATIK ROUT [2041018124]
1
PRESENTATION OUTLINE
 Introduction
 Motivations
 Uniqueness of the work
 Literature Survey
 Existing System
 Problem Identification
 Schematic Layout OR Model Diagram
 Methods OR Tools OR Algorithms used
 Experimentation and Results
 System Specifications
 Datasets Description
 Parameters used (if any)
 Experimental outcomes (Tables and Figures)
 Summary (Key findings)
 Bibliography 2
INTRODUCTION
 Overview
The purpose of this project is to understand how the amount of rainfall in India relates to the
energy produced by Hydro power stations in the country. To do this, we have collected data
spanning several years and analyse it. we have used a machine learning algorithm called linear
regression to figure out the connection between these two factors. India, like many other
countries, experiences varying amounts of rainfall each year. Rainfall is crucial because it affects
many aspects of life, including agriculture, water resources, and energy production. This project
specifically focuses on how rainfall influences the amount of energy generated by Hydro power
stations in India. To carry out this study, we have gathered data from government’s publicly
open databases, which included records of annual/monthly rainfall in various regions of India
and data on energy production from different Hydro power stations across the country. We have
collected information over several years to get a comprehensive understanding of the patterns
and trends.

3
INTRODUCTION CONTINUED…
 Motivations
This project is motivated by the critical need to understand the relationship between rainfall and
hydro energy production, offering insights for planning and managing energy resources in India.
The significance of this study lies in its potential to predict energy generation based on expected
rainfall, which is essential for ensuring reliable and sustainable energy production. By uncovering
how changes in rainfall impact energy generation, we can inform decision-making processes and
develop strategies to mitigate risks associated with fluctuating weather patterns. Ultimately, this
project aims to contribute to the optimization of energy resources and resilience to
environmental changes in India.
LITERATURE SURVEY:
we conducted a comprehensive review of several research papers focusing on the prediction of
rainfall and the application of linear regression, as well as exploring the concepts of time series
analysis and the ARIMA model. Our examination revealed key insights from the following selected
works:
 "Machine Learning Techniques for Rainfall Prediction: A Review": This paper provides an in-depth
overview of various machine learning techniques employed for rainfall prediction, offering
valuable insights into the advancements and challenges in this field.
 “A Review on Linear Regression Comprehensive in Machine Learning": This review delves into the
fundamental principles of linear regression within the context of machine learning, elucidating its
significance, methodologies, and applications across diverse domains.

5
 “Prediction of Hydropower Generation via Machine Learning Algorithms at Three Gorges Dam,

China": Focusing on a specific application, this study investigates the use of machine learning
algorithms to predict hydropower generation at the Three Gorges Dam in China, highlighting
the practical implications and challenges associated with such predictive modeling in the
energy sector.
LITERATURE SURVEY CONTINUED….
 Problem Identification

As we surveyed through these research papers and articles, we found that none had mentioned a
direct relation between rainfall and hydropower energy generation and sought to predict future
energy generation based solely on rainfall data. Therefore, the problem statement of our project
is to establish a direct relationship between rainfall and actual hydropower generation, aiming to
predict the hydro power generation based on rainfall only.
SCHEMATIC LAYOUT OR MODEL DIAGRAM

8
 METHODS OR TOOLS OR ALGORITHMS USED:
 ARIMA Model:
ARIMA stands for Autoregressive Integrated Moving Average. It's a popular statistical method
used for analysing and forecasting time series data.
 Autoregressive (AR): This component refers to the regression of the series against its past values.
It assumes that the current value of the series depends linearly on its previous values, weighted
by coefficients. The "order" of the AR component (denoted by p) specifies how many past values
are included in the model.
 Integrated (I): This component refers to differencing the raw time series data to make it
stationary. Stationarity means that the statistical properties of the series, such as mean and
variance, remain constant over time. Differencing involves subtracting the current value from the
previous one. The order of differencing (denoted by d) is determined by the number of
differencing steps needed to achieve stationarity.

12
 Moving Average (MA): This component represents the regression of the series against past
forecast errors. It assumes that the current value of the series depends on the errors made in
previous predictions. The "order" of the MA component (denoted by q) specifies how many past
errors are included in the model.
Together, the ARIMA model is described by three parameters: (p, d, q). The ARIMA(p, d, q) model is
often written as ARIMA(p, d, q), where:
p is the order of the autoregressive part.
d is the degree of differencing.
q is the order of the moving average part.
In summary, ARIMA models capture the temporal dependencies in time series data by incorporating
past values, past forecast errors, and differencing to achieve stationarity.
 Liner Regression:
Linear regression is a statistical method used to model the relationship between a dependent
variable (target) and one or more independent variables (predictors). It assumes that there is a
linear relationship between the independent variables and the dependent variable.
 Dependent Variable (Y): This is the variable we are trying to predict or explain. It is often referred
to as the target variable.
 Independent Variables (X): These are the variables used to predict the dependent variable. They
are also known as predictor variables or features.
Linear Relationship: Linear regression assumes that the relationship between the independent
variables and the dependent variable can be represented by a straight line. Mathematically, this can
be expressed as: Y = β0 + β1X1 + β2X2 + ... + ε
Where:
Y is the dependent variable. X1, X2, ... are the independent variables.
β0, β1, β2, ... are the coefficients representing the intercept and slopes of the linear relationship.
ε is the error term, representing the difference between the observed and predicted values of the
dependent variable.
 Fitting the Model: The goal of linear regression is to find the "best-fitting" line that minimizes
the difference between the actual values of the dependent variable and the values predicted
by the model. This is typically done by minimizing the sum of squared differences between the
observed and predicted values, a method known as Ordinary Least Squares (OLS).
Linear regression is widely used in various fields for prediction, inference, and understanding the
relationship between variables. It serves as a fundamental tool in statistical analysis and machine
learning.
 EXPERIMENTATION AND RESULTS
 System Specifications
 CPU: Intel Core i5

 RAM: 8GB

 Operating System: Windows 10

 Software: ANACONDA.NAVIGATOR (Jupyter Notebook)

EXPERIMENTATION AND RESULTS CONTD..
 Datasets Description
EXPERIMENTATION AND RESULTS CONTD..

 Parameters used :
 Actual Power Generation
 Rainfall
EXPERIMENTATION AND RESULTS CONTD..

APPLYING ARIMA MODEL ON RAINFALL SERIES

 Checking Stationarity Of The Series
 Calculating Hyper-Parameters (p, d, q) :
 p is the order of the autoregressive part.
 d is the degree of differencing.
 q is the order of the moving average part .
 Building ARIMA Model
 Predicting Test Dataset
 Evaluating Model
 Predicting Future values
 CHECKING STATIONARITY OF THE SERIES
 Visually
 AUGMENTED DICKEY-FULLER TEST (ADF TEST)
 Calculating hyperparameters p,d,q
To Calculate these parameters we have 3 methods
 ACF and PACF plots
 AUTO ARIMA method
 Custom FOR loop method
p=6
q=0
d=7
WHAT’S REMAINING ?

 Applying Liner Regression

 Predicting Future Power Generation
 Exploring use cases of this Project
SUMMARY(KEY FINDINGS)

In the beginning of our project, we got rainfall data from government websites and made sure it was
accurate by cleaning it up. Then, we used a special model called ARIMA to predict future rainfall
based on this data. Next, we'll work on figuring out how rainfall is connected to actual hydro power
generation. We'll use this relationship to make a model that can predict how much hydro power will
be generated in the future. This will help us understand more about renewable energy and make
better decisions about managing energy.

29
 BIBLIOGRAPHY
 Google. Retrieved from https://www.google.com
 Wikipedia. Retrieved from https://www.wikipedia.org
 YouTube. Retrieved from https://www.youtube.com
Articles and Research Papers:
Parmar Aakash , Mistree K., & Sompura, M. (2017). Machine Learning Techniques for Rainfall
Prediction: A Review. In International Conference on Innovation in Information Embedded and
Communication (ICIIECS).
Maulud D. H., & Abdulazeez, A. M. (Year). A Review on Linear Regression Comprehensive in Machine
Learning. [Full citation details if available].
Huang Y. F., & Kumar, P. (Year). Prediction of Hydropower Generation via Machine Learning
Algorithms at Three Gorges Dam, China.
31