Research of the method for shortterm wind speed forecasting of adaptive  index smoothing based on gradient optimizion

ZHANG Jun, WUE Li, WANG Bo, GUAN Yongchang, ZHAO Xin

(State Grid Bin Zhou Power Supply Company, Binzhou 256610, China)

Abstract: Aiming at the problem of traditional three exponential smoothing forecast without high accuracy, and the selfadaptive cubic exponential smoothing method was poor in computational efficiency, this thesis indroduced gradient optimization algorithm based on equaldimensional new information prediction, a method for winds from the shortterm forecasting of wind farm based on gradient optimization of the adaptive exponential smoothing was proposed, which the speed of tracking optimum smooth coefficient accelerated, not only the computing efficiency improved, but also got forecasting statistics with high accuracy. By comparing with the traditional cubic exponential smoothing method、gray model and adaptive index smoothing based on gradient optimizion forecasting method, the results show that the computational efficiency of this method increases by 80% than selfadaptive traversal cubic exponential smoothing method, the accuracy of this method increases by 27% than traditional exponential smoothing method and increases by 32% than grey prediction method, the prediction model has high accuracy when wind speed changes gently and fluctuations smooth. These results verify the accuracy and validity of the method.

Key words: wind speed forecasting; selfadaptive; smooth parameter; gradient optimized