Gaussian Kernel Based SVR Model for Short-Term Photovoltaic MPP Power Prediction

Predicting the power obtained at output of photovoltaic (PV) system is fundamental for optimum use PV system. However, it varies different times day depending on intermittent and nonlinear environmental conditions including solar irradiation, temperature wind speed, Short-term prediction vital in systems to reconcile generation demand terms cost capacity reserve. In this study, a Gaussian kernel based Support Vector Regression (SVR) model using multiple input variables proposed estimating maximum from perturb observation method irradiation temperatures short-term DC-DC boost converter The performance kernel-based depends availability suitable function that matches learning objective, since an unsuitable or hyper parameter tuning results significantly poor performance. study first time literature both point estimation made. While evaluating suggested model, data simulated variable irradiations one MATLAB were used. irradiance was 852.6 W. accuracy evaluation forecasting identified utilizing computing error statistics such as root mean square (RMSE) (MSE) values. MSE RMSE rates which 4.5566 * 10−04 0.0213 ANN model. 13.0000 0.0362 SWD-FFNN Using SVR 1.1548 10−05 0.0034 obtained. prediction, gave higher according SWD-FFNN.