Comparison of the LS-SVM based load forecasting models

Load forecasting plays an important role in the planning and management of electric power system. For the load forecasting model based on Least squares support vector machine (LS-SVM), the selection of learning parameters of the LS-SVM has significant impact on the forecasting accuracy. In this paper, a research on the comparison of two the LS-SVM load forecasting models, grid search based LS-SVM model and bayesian framework based LS-SVM model, is conducted, and the learning parameter selection of LS-SVM is discussed. In the experiments, these two models are employed to forecast the daily maximum load demands in one month. Results show that both of the two models have a high forecasting accuracy and great generalization ability, while bayesian framework based LS-SVM load forecasting model requires much less computation time for parameter learning.