A Hybrid Support Vector Machine with Grasshopper Optimization Algorithm based Feature Selection for Load Forecasting

Electrical Load forecasting plays a significant part for power systems planning, operation and control for efficacy companies as well as policy makers to develop the reliable as well as stable energy infrastructure. Various approaches like conservative, Artificial Intelligence (AI) as well as hybrid approaches had been introduced to analyze the short-term load forecasting. However, these approaches had faced several problems like low convergence speed, high computational complexity and minimum prediction accuracy. To overcome these challenges, this research proposes the hybrid method of improved Support Vector Machine (SVM) and Grasshopper Optimization Algorithm (GOA) called SVM-GOA based feature section and hyperparameter optimization for electrical load forecasting. In pre-processing step, the min-max normalization technique is used for the scaling the feature data. Furthermore, the proposed SVM-GOA is trained and tested by simulations on the Singapore dataset. The effectiveness of the proposed SVM-GOA is estimated by various performance metrics like Root Mean Square Error (RMSE), Mean Absolute Percent Error (MAPE), Mean Absolute Error (MAE), R-Square (R2) and it achieves the values of 0.6547, 2.13, 0.69 respectively when compared to the previous methods like Artificial Neural Network (ANN) and Federated Learning (FL).