Research On Support Vector Machines In Power System Short-term Load Forecasting

The daily operation and planning activities of an electric utility requires the prediction of the electrical demand of its customers. In general,the required load forecasts can be categorized into short-term,mid-term,and long-term forecasts. The quality of short-term load forecasts has a significant impact on the economic operation of the electric utility since many decisions based on these forecasts have significant economic consequences. The importance of accurate load forecasts will increase in the future because of the dramatic changes occurring in the structure of the utility industry due to deregulation and competition. This environment compels the utilities to operate at the highest possible efficiency,which as indicated above requires accurate load forecasts. Generally, there are two kinds of methodologies for load forecasting. One is in a traditional way, represented by time series, another one is termed as new artificial intelligence method, represented by the artificial neural network.This paper analyses the basic theories of SVM. SVM have the remarkable advantages of non-linear regression,high forecasting accuracy and small time complexity. According the non-linear relationship between the forecasting load and its influence factors, This paper proposes to use its advantages of non-linear processing and generating ability to accomplish short-term load forecasting of power system,so as to improve forecasting precision and executed speed, and proposes a short-term load forecasting model based on SVM. Compared with the forecasting method of artificial neural networks(ANN), the simulation results of the practical application show that the SVM method is much better than ANN.In regards to forecasting on the base of SVM, feature selection has significant influence as it may lower the complexity of learning, enhance the generalization ability and simplify the learning model. F-score feature selection finds the main feature component from the mutli-dimension data in order to eliminate the collinearity and chaos among the variables and compress the dimension in the space.. This article proposes to combine the F-score and SVM to realize the short-term load forecasting. After the feature dimensions are reduced by F-score selcetion, the principle component containing the information of sample data are sent to support vector machine for training. This proposal combined the feature selcetion ability of principle component analysis together with the excellent ability of support vector machine for non-linear function approaching. The empirical results show that the precision and the generalization ability of the load forecasting model are improved by this method.