Neural Network – based Sensitivity Analysis of the Factors affecting the Solar Photovoltaic Power Output

Abstract

Technological advancements and modernization of different industries and disciplines contributed to more consumption of oil and electricity which powers these industries. Aligned with the United Nations (UN) Sustainable Development Goals (SDG), the use of alternative and renewable energy (RE) sources is encouraged as it allows the utilization of clean energy resources and access of populations in developing countries to electricity and energy. Forecasting and maximizing the harvest for renewable energy requires an understanding of the mechanics behind the variables that impact solar photovoltaic production. 755 datasets were created from 150 days of recorded data and used in the model building and sensitivity analysis. The approach used in this study to identify the variable importance of each meteorological variable to the solar photovoltaic (PV) production was the Garson’s algorithm (GA). In this study, an artificial neural network (ANN)-based sensitivity analysis (SA) using Garson’s algorithm (GA) was implemented to identify the relative importance (RI) of the factors influencing the solar PV output including the solar irradiance (SI), rainfall, maximum temperature (MaT), minimum temperature (MiT), relative humidity (RH), and wind speed (WS). The model also considers the relative significance of these parameters to the solar PV output. Results indicate that, with a relative value of 29.48% and 5.01%, respectively, solar irradiance and wind speed are the most and least important factors.