Short-Term Photovoltaic Power Forecasting Using Artificial Neural Network Considering Multiple Time Lags

This paper presents the artificial neural network (ANN) used to perform the short-term photovoltaic power forecasting (STPPF) for the next 24 hours. The input data of ANN is comprising with the multiple time lags of hourly data of power, current, temperature, solar irradiance and hour that have been denoised by using the wavelet decomposition. The multiple time lags are used to improve the input data while wavelet decomposition removes the noises available in the data which will then significantly improve the STPPF results. The data of a solar photovoltaic (PV) system in the year 2015 and 2016 obtained from the Green Energy Research (GERC), UiTM, Shah Alam, Malaysia. The results have shown that the ANN provides relatively accurate results of STPPF and able to forecast the PV output power for the next 24 hours with less error.