Short-term power load forecasting method by radial-basis-function neural network with support vector machine model

Energy demand forecasting is crucial in the energy industry. Various methods have been proposed for load forecasting, including artificial neural networks - enhanced prediction algorithms based on structures and functions of biological neurons. According to the chaotic and nonlinear characteristics of power load data, the support vector machine (SVM) model has been used to predict power load. This paper proposes a new radial-basis-function neural network with a support vector machine (RBFNN-SVM) model to enable high accuracy in short-term power load forecasting. This approach applies RBFNN to predict the monthly increase before training the hourly data through SVM to obtain a final prediction. Using power load data from France for testing and verification, RBFNN-SVM performed well in predicting future load, showing that the proposed method is accurate and effective.