Photovoltaic sizing using machine learning

Abstract

Renewable energy is the future of energy in the world. Solar energy is a major renewable energy source which addresses the energy deficiency problem in various countries. This endless energy source is captured using photovoltaic systems. However, in order to make the best out of photovoltaic systems, the size of the system must be optimized. Sizing has always been a critical issue that leads to huge losses once not properly designed. Existing methods consider sizing problem as a case dependent optimization problem that requires data from the place where the photovoltaic system is to be installed. This study introduces machine learning algorithms into the photovoltaic systems’ sizing problem for the first time. The proposed case-independent model automates the sizing process and is generalized to be applied at any location. Linear regression, polynomial regression, neural network, random forest, and decision tree were implemented and compared in terms of accuracy. Results show that random forest is the best suited algorithm for this application.