Development of an Optimized Support Vector Regression Model Using Hyper-Parameters Optimization for Electrical Load Prediction

Abstract Electrical load prediction is important to the effective operation, management and control of electric power systems. Several machine learning forecasting models have been developed for electrical load prediction. However, inappropriate selection of model hyperparameters, could result in low prediction accuracy of machine learning models. Hence, the development of an optimized Support Vector Regression model for electrical load prediction was presented in this study. Historical daily data of temperature, rainfall, relative humidity and windspeed for Osogbo, Nigeria was obtained from the National Aeronautics and Space Administration; while electrical load data for the same location was collected from the Transmission Company of Nigeria. The data captured a period of five years (2017 to 2021). The SVR models were developed with MATLAB, and two hyper-parameters, epsilon and box constraint, were optimized. The models were evaluated using mean absolute error (MAE) and root mean square error (RMSE). The MAE and RMSE for the non-optimized SVR model were 6.1977 and 8.0926 respectively, meanwhile, for the optimized SVR model, the MAE and RMSE were 5.8031; 7.0571 respectively. The obtained results show that the optimized SVR performed better than the non-optimized SVR models. Electrical power utility providers could adopt the method developed in this research.