Applying wavelets to predict solar PV output power using generalized regression neural network

Abstract

This paper presents a hybrid intelligent approach to forecast short-term output power of a PV system. The proposed hybrid method is composed of a data filtering technique based on wavelet transform (WT) and generalized regression neural network (GRNN). In order to validate the prediction capability of the proposed WT+GRNN model, test results are compared with other soft computing models (SCMs). This paper uses a PV system data derived from Ashland, Oregon. Simulation results demonstrate the greater ability of GRNN model to handle nonlinear solar PV time-series data, and when it is combined with the WT, the forecasting accuracy is greatly enhanced.