FEASIBILITY OF RENEWABLE ENERGY GENERATION USING MACHINE LEARNING BASED PREDICTION MODEL OF WIND VERSUS SOLAR POWER GENERATION

Power output of solar and wind energy depend upon the factors such as humidity, temperature, wind and other. The purpose of the project is to develop an application which would determine which renewable source will be used and which source will be more efficient. Main motivation behind this project

Project Title

FEASIBILITY OF RENEWABLE ENERGY GENERATION USING MACHINE LEARNING BASED PREDICTION MODEL OF WIND VERSUS SOLAR POWER GENERATION

Project Area of Specialization

Artificial Intelligence

Project Summary

Power output of solar and wind energy depend upon the factors such as humidity, temperature, wind and other. The purpose of the project is to develop an application which would determine which renewable source will be used and which source will be more efficient. Main motivation behind this project is to present such a model that will predict the power produced by solar energy and wind energy and on the basis of those power outputs the model will decide which energy is most feasible. First, the model will be trained by the datasets which are available on Kaggle websites which includes the weather details of different countries and on the basis of these datasets model will predict the power produced by both solar and wind energy. The datasets consist of time step of 1 hour for at least 35 years from 1/1/1985 to 12/31/19. The models trained requires 8 features in order to predict the output of the source. These 8 features would not be available for future prediction. For this purpose, variables are also predicted by using 8 different models for sequential regression of single variables. Then the predicted features would be fed to the most accurate model for regression of source output. After the prediction of output power, model will also determine the cost which is required to produce either from solar or wind energy. Proposed project has wide range of applications such as model can be used to predict the power output that can be produce in a particular area to setup an industry. Model can also predict the power that can be produced in next 5 to 10 years.

Project Objectives

The basic purpose of this project is to predict the power produced through solar and wind energy. In modern era, prediction of something which brings comfort to the living of an investor is very rare. So, when an investor knows about the power produced in certain environment with specific cost and feasibility, his misconceptions are reduced to great extent. Hence investor will respond accordingly. Main motivation behind this project is to present such a model that will predict the power produced by solar energy and wind energy and on the basis of those power outputs the model will decide which energy is most feasible. The basic purpose of this project is to predict the power produced through solar and wind energy. In modern era, prediction of something which brings comfort to the living of an investor is very rare. So, when an investor knows about the power produced in certain environment with specific cost and feasibility, his misconceptions are reduced to great extent. Hence investor will respond accordingly.

Project Implementation Method

This study proposes six prediction models to solve a real-life problem in the renewable energy sector by accurately estimating the amount of wind and solar energy by applying machine learning and deep learning techniques using historical wind and power solar power generation data and weather forecasting reports.

The models would then be compared as to determine which model is better suited for the stated problem in terms of accuracy, efficiency and budget. The dataset was divided into two parts, training part and testing part, the training part for solar consists of entries from 1/1/1985 to 8/7/2010, which makes at least entries for 25 years, with the time step of 1 hour total consisting of 224400 entries. The rest was separated for testing, the trained models were to regress the value of output for the next 9 years from 9/7/2010 to 21/31/2019. For wind data, the training data contains 40424 entries, from 1/1/2018 to 18/10/2018 with a time step of 10 minutes making the data of at least 10 months. The rest was separated for testing, the trained models were to regress the value of output for the next 2 months from 19/10/2018 to 31/12/2018. The regressed output was then compared with the original output by calculating RMSE and then the results were compiled. Root mean square error is the error or difference which is between the values predicted by the model and the values which are being observed. 

Benefits of the Project

Prediction of power generation systems under varying environmental conditions is a difficult task and still an open research topic. Feasibility of a generation technology for a specific environment is therefore a challenging problem. Renewable energy sources are clean source of power production such as sunlight, heat, wind and etc. These sources are highly helpful in producing power which can run industries with very low maintenance cost. This project is regarding the power production from renewable energy sources which are solar and wind energy. On the basis of machine learning algorithms, the models are trained and on the basis of trained models the model will predict the amount of power which is being produced by solar and wind energy and after predicting the power the model will also predict the cost which will tell about the feasibility. Applications would include;

• Solar Power production forecast.
• Wind Power production forecast
• Comparative feasibility of renewable energy sources
• Estimation of efficient renewable energy production plan
• Comparative efficiency of multiple regression models

Technical Details of Final Deliverable

The project will involve training of different machine learning and neural networks models. And these models will be trained through the datasets containing features of solar and wind energy. After training, the models will predict the power as an output. On the basis of the output power the project will determine which renewable source is more feasible in terms of installation cost, maintenance cost and area. Final deliverable would be an application capable of constructing and simulating a renewable energy plan for a specific load while outlining the feasibility analysis of Wind and Solar renewable energy source.

Final Deliverable of the Project

Software System

Core Industry

Energy

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Clean Tech

Sustainable Development Goals

Affordable and Clean Energy, Industry, Innovation and Infrastructure, Sustainable Cities and Communities, Climate Action

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com