Parameter Optimization Using PSO for Neural Network-Based Short-Term PV Power Forecasting in Indian Electricity Market

Because of developing concern to environmental changes, renewable energy is being looked as a key alternative to the conventional sources. Photovoltaic power is a green and an abundant renewable energy source. Photovoltaic power is dependent upon the solar irradiation which is highly intermittent in nature. So the precise forecasting of PV power is necessary to improve the operation of an electrical grid with the distributed energy resources. This paper proposes a novel hybrid model by combining particle swarm optimization (PSO) and the feed-forward neural network (FFNN) together. Proposed hybrid model is applied to forecast the PV power in Indian electricity market. One-year data consisting of hourly PV power generation, direct radiation, diffused radiation, and ambient temperature from the Indian energy market has been used for PV power forecasting. The developed model is applied for one-week ahead PV power forecasting for winter, summer, rainy, and autumn season, respectively. The performance of the proposed hybrid model outmatches and compared with some recently reported model.