Applied Study On Short-Term Load Forecasting Method Based On Knowledge Base

Short-term load forecasting (STLF) is very important to run safely, steadily, economically for the power system. The key to improve predication accuracy is properly handing the problem in abnormal days. In this paper, a novel approach based on knowledge base is put forward. Data preprocessing is first of all, then predicates with genetic-fuzzy algorithm from whose result we can get the correction value. Using C4.5 algorithm of the Data Mining Technology finds the knowledge relation between load correction value and holiday, weather factors by the premise of comprehensive consideration of holiday, day-type, weather. The Keeping method is applied to estimate the correct rate of classification and get the classification tree and result. Finally, the knowledge base is built by use of Visual Prolog software, in which the knowledge is represented by production rules. The testing results of STLF in Baoding area show that the proposed method has higher forecasting accuracy.