Comparison of Artificial Neural Network, Linear Regression and Support Vector Machine for Prediction of Solar PV Power
Ans Maria Kuriakose, Denny Philip Kariyalil, Marymol Augusthy, S Sarath, Joffie Jacob, Neenu Rose Antony
Abstract
Solar Photo voltaic (PV) system's usage is increasing day by day as a substitute of energy considering about the environmental factors. But at the same time its performance have a huge impact. Due to the uncertainty of solar Photo voltaic (PV) power outputs, an accurate method of predicting the output power is required. Thus, a proper estimation of the solar power must be done. Predicting the solar power will help in optimal planning of PV units in generating or transmission, scheduling of other generators by considering the predicted values etc. This paper deals with a comparison between Machine Learning algorithms on Day-Ahead forecasting of the Solar Photo-voltaic output considering the weather parameters (which include wind speed, humidity, radiance and temperature). Artificial Neural Network, Linear Regression and Support Vector Machine have been considered and a conclusion is drawn based on the results obtained. The objective of the paper is to predict the Solar Power at Amal Jyothi College of Engineering, Kanjirapally, Kottayam. To provide real time values weather station has been used. The daily forecast results will help to improve the forecast accuracy which will eventually help in its proper utilization.