The use of SVM and AR Methods for allotment of Load Scheduling and Energy Management in SPS

Integration of photovoltaic (PV) systems into energy grids to meet the solar energy consumption requires high-performance predictive algorithms Although genetic process (GP) based additive regression (AR) models so provide answers to nonlinear prediction problems For the information Another hybrid model is proposed that combines the SR with a support vector machine (SVM) architecture. The hybrid model seeks to improve accuracy and robustness by using real-world climate data including solar radiation and historical PV power estimates. Using the Elastic Net and Extreme Boosting methods, the hybrid AR-SVM algorithm makes fast feature selection by eliminating redundant inputs. Stability and reliability are ensured by maintaining the hyper-criteria during the training and testing phases. The model reduces the computational complexity and improves accuracy by using fewer layers and neurons. The improved prediction accuracy is demonstrated by simulation results, which show a significant reduction in the absolute error (MAE) and root mean squared error (RMSE) The improvement in the $\mathbf{R} 2$ evaluation measure confirms the power of the model has to recognize the underlying pattern emphasis. By preventing the AR from reaching a local minimum with the contribution of the SVM branch, the hybrid model exhibits high robustness to forecast errors, and shows superiority in PV power forecasting applications.