PV Power Forecasting for Operation of BESS Integrated with a PV Generation Plant

Enhancing the control performance of a battery energy storage system (BESS) integrated with a photovoltaic (PV) generation plant necessitates accurate PV output forecasting, a crucial component in the operation of energy management system (EMS) of the integrated system. This paper introduces a comprehensive multiple hybrid model designed for intraday hourly PV power output forecasting. A genetic algorithm (GA) is employed to optimize the parameters of each gated recurrent unit (GRU) model based on forecasted weather data from weather research forecasting (WRF) platforms. The primary objective is to provide multiple forecasts, thereby enhancing the accuracy of PV power output predictions and thus increasing the effectiveness of the EMS in the integrated PV and BESS system. Numerical results demonstrate superior prediction accuracy compared to traditional single-model and non-rolling one-day ahead forecasting approaches. The prediction accuracy contributes to a decrease in curtailed PV energy and a significant boost in total revenue, regardless of prevailing weather conditions. These results affirm the values of the proposed approach in practical applications.