The Estimation of Hourly Solar Radiation on tilted Surfaces using Artificial Neural Network: A Case Study of Surabaya

Abstract

This study aims to generate an accurate model for estimating the radiation of solar panels on different inclination angles. The output of this model is useful for determining the optimal installation angle of the solar panel either on land or on the ships. Furthermore, the amount of the hourly direct and diffuse radiation on the horizontal surface is estimated using Artificial Neural Networks (ANN), which were trained with the monthly radiation data of Surabaya from 2018 to 2019. Subsequently, the radiation on the inclined surface is estimated using a mathematical model. Also, the ANN accuracy was validated with a regression value higher than 99% for either direct or diffuse radiation estimate. A full-year evaluation based on the proposed model suggests an inclination angle of 25° for the solar panel installed in Surabaya. Meanwhile, the evaluation gives different angles for each month with the advantage compared with the fixed angle installation.