Short-term photovoltaic solar power forecasting using a hybrid Wavelet-PSO-SVM model based on SCADA and Meteorological information

Photovoltaic (PV) solar power generation is always associated with uncertainties due to solar irradiance and other weather parameters intermittency. This creates a huge barrier in integrating solar power into the grid and biases power industries against deploying PV systems. Thus accurate short-term forecasts are important to efficiently integrate PV systems into the grid. This paper proposes a hybrid forecasting model combining wavelet transform, particle swarm optimization and support vector machine (Hybrid WT-PSO-SVM) for short-term (one-day-ahead) generation power forecasting of a real microgrid PV system. The model is developed by incorporating the interactions of the PV system Supervisory Control and Data Acquisition (SCADA) actual power record with Numerical Weather Prediction (NWP) meteorological data for one year with a time-step of 1 h. In the proposed model, the wavelet is employed to have a considerable impact on ill-behaved meteorological and SCADA data, and SVM techniques map the NWP meteorological variables and SCADA solar power nonlinear relationship in a better way. The PSO is used to optimize the parameters of the SVM to achieve a higher forecasting accuracy. The forecasting accuracy of the proposed model has been compared with other seven forecasting strategies and reveals outperformed performance with respect to forecasting accuracy improvement.