Load Forecasting Method Research Based on Improved K-means Algorithm and SVM

Abstract

A short-term load forecasting method combining k-means clustering algorithm and SVM is proposed. Euclidean distance and waveform similarity clustering of double standards is used in improved k-means clustering algorithm. The different load curves is accurately classified and their typical load curve is extracted, realized the classification function of different types of user. Then according to the classification results, select the same type of load curves and load factors with the predicted load as input of support vector machine prediction model. This method is used to classify and predict the actual daily load curve of shanghai. It shows that the method can greatly improve the prediction accuracy and is practical.