Short-term Load Forecasting Based On Support Vector Mahcines

Power system short-term load forecasting has important significance for the safe andeffective operation of power grid. At the same time the reforming of power market has made ahigher requirements on the accuracy of short-term load forecasting of power system. So studyof short-term load forecasting has important significance for the safe operation of the powergrid and improving the economic efficiency of power grid.In this paper, support vector machine method is used for power system short-termload forecasting research. Firstly, historical load data is processed. Longitudinal processingmethod of data is improved, which is used to repair the missing data and bad data. Fuzzy Cmeans algorithm is used to analyze the historical load data to get the fuzzy membershipdegree matrix. The Jffreys&Matusita distance is used to select the similar data after historicalload data is clustered. Historical data that are similar to the data of prediction day is chosenbased on the Jffreys&Matusita distance. Short-term load is predicted by support vectormachine model. The simulation results show that the prediction results of the error are in theallowable range. The method of fuzzy C means clustering combined with Jffreys&Matusitadistance and support vector machines have the best forecast effect, which is superior to themethod of Jffreys&Matusita distance combined with support vector machines. It’s alsosuperior to the method of only using support vector machine.