Middle and Long-Term Daily Load Curve Forecasting Based on Support Vector Machine

A new method to forecast middle- and long-term load curve is proposed. By means of historical typical daily load data, an annual time series of typical daily load is constituted; then adopting support vector machine(SVM) method, the values of loads at any point of time in forecasted day is forecasted and typical daily load curve is obtained. As for the proposed method, the forecasting of daily load feature, maximum load and electrical energy demand is not needed, so the possible error accumulation can be avoided. Taking a certain power network for example, the proposed method is tested. Test results show that the accuracy of daily load curve forecasted by the proposed method can be improved.