Hybrid intelligent method of relevant vector machine and regression tree for probabilistic load forecasting

This paper proposes a new hybrid intelligent method for probabilistic short-term load forecasting (STLF) in power systems. It consists of Relevance Vector Machine (RVM) of the statistical learning method called Kernel Machine and regression tree (RT) of data mining. As the preconditioned technique of data, RT is used to classify learning data into some clusters with the data similarity. After classifying data into some clusters, RVM is constructed to predict one-step ahead loads at each cluster. RVM is one of efficient Kernel Machines that extend Support Vector Machine (SVM) to deal with continuous variables. It has advantage to narrow the lower and upper bounds of predicted values with high accuracy. The proposed method is successfully applied to real data of Japanese utilities.