Daily load forecasting using support vector machine and case-based reasoning

Regarding to the daily load forecasting, the sample Selection and data preprocessing are crucial to its' precision. In this paper, case-based reasoning (CBR) is adopted to search the historical data whose features are the same as the predict day. CBR is realized through the steps of case representation, indexing, retrieval, and adaptation, and the key idea in CBR involves the use of already existing knowledge about objects or situations to predict aspects of similar objects. This method uses not only case specific knowledge of past problems, but also uses additional knowledge derived from the clusters of cases. After the data pretreated the sample set becomes more relational with the predict day. Meanwhile the training sample set for support vector machine (SVM) for daily load forecasting (DLF) becomes smaller. With the prediction precision increasing, the time for calculating and predicting decreased. At last, the testing results on a real power system show that the proposed model is feasible and effective for load forecasting.