Short-term load forecasting: Learning in the feature space based on local temperature sensitive information

Abstract

A novel hybrid method based on feature extraction and neural network for short-term load forecasting was presented. It is well known that temperature information is very important for load forecasting, but the local structure of temperature sensitive information is not adopted in the literature. The proposed model adopts an integrated architecture to handle the local temperature sensitive information. Firstly, the input load data set is clustered into several temperature similar days subsets by the k-means algorithm in an unsupervised manner, Then compute max temperature factor in each subsets and split the time point (5 minutes, 288/day) into several time range, in each time range, we extract the features (coefficients) from load data using flourier basis system, and then learn the function in the feature space using artificial neural network. Finally, we smooth the whole forecasted load curve using linear programming. The empirical results indicate that our hybrid method results in better forecasting performance than the original generic support vector regression.