Short-term Load Forecasting Of Power System Based On RBF Neural Network And Fuzzy Control Theory

Short-term load forecasting is the basis of the power system optimization running. It has significant influence for safety, reliability and economical of power system. At present, intelligent network technology becomes the new direction of national power grid, which has more and more demand for load forecasting, therefore, the application of intelligent algorithm for short-term load forecasting, which improving the precision and stability of the load forecast, has the extremely vital significance.According to the characteristics of electric power load variation, considering the date type, temperature, weather conditions of load forecast factors, the paper puts forward a short-term load forecasting method which is the combination of radial Basis Function (RBF) neural network and fuzzy theory. Firstly, the paper adopts the horizontal and vertical data smoothing method to process different load data, for the missing data, it uses similar selection methods for addenda, which avoids the impact for load forecasting accuracy by error data or missing data. Secondly, based on the actual area of power load characteristics analysis, it determines the main influence factors of the short-term load curve. According to the results of analysis, it choices the biggest impact factors of short-term load forecasting as input data, the paper selects the characteristic parameters of the day type, the highest temperature, the lowest temperature and weather conditions. Finally, the fuzzy control theory is introduced to the RBF neural network, based on RBF neural network forecasting model, it uses the relative error and relative error as input for fuzzy controller, and the output is power modifying factor, after that, it puts the algebra of predicted RBF network values and the fuzzy adjustment quantity as the final prediction, at the end, according to the actual area of the historical load data analysis, the RBF neural network combine fuzzy theory used in load forecasting compares with the RBF neural network method.Practical examples show that the combination forecast method of RBF neural network with fuzzy controller accelerate the learning rate and improve the load forecasting accuracy, and it has great application prospects.