Application of Artificial Neural Network (ANN) for Solar Photovoltaic Power Forecasting

Abstract

The integration of renewable energy sources, particularly solar photovoltaic (PV) systems, is crucial in the global fight against climate change and the reduction of dependence on fossil fuels. However, the variability in environmental conditions makes accurate forecasting of PV system power output challenging. This project investigates the application of Artificial Neural Networks (ANN) to predict PV system power output using historical and real-time environmental data. Data were collected over 20 days at Lorong Haji Idris, Parit Raja, Batu Pahat, Johor, with measurements taken five times daily. The ANN model effectively addresses the nonlinear relationships and complexities of PV performance. Statistical metrics such as Mean Squared Error (MSE) were used to validate the model's accuracy. The findings show that the ANN model achieved high accuracy with R = 0.98661, MSE = 5.37 W², and RMSE = 2.32 W, with most predictions within ±5% of measured values. In comparison, the calculation method produced higher errors wih MSE = 26.89 W² and RMSE = 5.19 W, confirming that ANN is more reliable for PV power forecasting.