Short-term Electric Load Forecasting Using Neural Network For Dalian District

Shot-Term Load Forecasting (STLF) has played an important role in the running and dispatching of electric power system. It is a key basis for guaranteeing the safe and economical operation of power system and achieving the scientific management and dispatch of power grid. It is also a basic content for the commercial operation of power grid. Therefore, how to establish a load forecasting model with better performance and prediction accuracy has become the focus problem of power system.In the context of short-term electric load forecasting of Dalian district, we first analyzed the research status and development trend of load forecasting techniques, and discussed the characteristics of load forecasting, especially the grid status and changing rule of load curves in Dalian district. Then, according to the actual load characteristics and load forecasting skills of Dalian district, and taking into account the factors such as temperature, date types and weather conditions which affect the accuracy of load forecasting, we established two short-term electric load forecasting models using the BP neural and RBF neural network, separately. With the established models, different kinds of short-term load forecasting work was carried out using the historical load data of Dalian district, the forecasting results of the chosen model are compared. Finally, with the well-performed RBF neural network as the basic operation structure of the whole forecasting model, and the combination of the fuzzy control theory and its operation method, we constructed an hybrid short-term load forecasting model. The coarse load forecast was first obtained via the RBF network, and then we adopted the fuzzy controller with online self-turning correction factor to adjust the forecast error intelligently and to improve the forecasting accuracy. With the hybrid model, a24-hour short-term load forecasting work was also carried out using the historical load of Dalian district, and the forecasting results demonstrated the effectiveness and practicability of the proposed approach.The research work in this paper will provide an important basis for technology management department on the power-system scheduling, planning, marketing, et al. in Dalian district. It will also help to realize the modern management of power system, possessing not only the very important theoretical significance, but also the very prominent engineering value.