Data analysis to generate models based on neural network and regression for solar power generation forecasting

Abstract

This paper presents methods for forecasting solar power generation by a solar plant. Solar power generation depends primarily on relative position of sun and some extrinsic as well as intrinsic factors. Extrinsic factors such as cloud cover, temperature, wind speed, rainfall and humidity have been used with intrinsic ones such as degradation of solar panels as inputs for proposed techniques for generation forecasting. The authors have used multiple linear regression, logarithmic regression, polynomial regression and artificial neural network method on the data of past one year (January 2014-December 2014) for creation of forecasting models. These forecasting models are then compared on the basis of their accuracy to forecast the solar generation.