Skip to Main Content
In order to forecast the integrated load model of substation with a certain random time variation character and increase the accuracy of forecasting, this article put forward a forecasting method of electrical consumption proportion of different industries in substation based on the daily load curve. First of all, load sequence is decomposed into a number of different frequency stationary components by using EMD, according to the variation of the components, select the appropriate SVM parameter and support vector machine with different kernel function construction to forecast separately, and get the load curve forecasting value combined from each forecasted value by SVM. Then, classify the attributive for each industry and combine the industry equivalent daily load curve by using the fuzzy C means clustering principle. Finally, structure the related relation with the forecasted load curve, namely that obtain the final forecasted industry electrical consumption proportion in substation industry through the normalized projection of forecasted load curve calculated by industry typical feature vector. Refer to the simulation result, there are strong generalization ability and high precision for this method.