Short-Term Solar PV Forecasting Using Gated Recurrent Unit with a Cascade Model

摘要

The fluctuation in solar photovoltaic (PV) generation system causes inefficiency in PV power management. Thus, predicting solar PV power is essential to assist PV system in improving the overall performance of a solar plant operation. In this paper, solar PV forecasting model with multiple Gated Recurrent Unit (GRU) networks is proposed to effectively improve the prediction accuracy and the training time compared to the typical GRU network. In addition, other popular prediction machine learning algorithms, namely Feed-forward Artificial Neural Network (ANN), Support Vector Regression (SVR) and K Nearest Neighbors (KNN), were implemented for comparison with the proposed model. Each model was evaluated with Normalized Root Mean Squared Error (NRMSE). The proposed model, GRU, Feed-forward ANN, SVR, and KNN has NRMSE of 9.64%, 10.53%, 11.62%, 11.45%, and 11.89%, respectively. Hence, the proposed model provides enhanced prediction accuracy with improved speed compared with a GRU network.