The Reserch Of Short-term Wind-Speed Predication Based On PSO-GRNN Algorithm

Wind is a kind of new energy presented by nature and it has been widely favored for its high quality and no pollution.At present,the method for developing and utilizing wind resources is large-scale grid-connection.But due to its intermittence,randomness and uncertainty,it has caused a lot of difficulties to the stability of connection grid and transmission grid,and the dispatching of power load,which limits the large-scale application of wind power generation.The accurate prediction of wind speed is helpful for dispatchers to dispatch the grid normally and is one of the effective ways to solve the problems.This paper adopts the neural network method for short-term wind speed forecasting,mainly including BP and generalized regression neural network.During the small training,it is found that the forecasting precision of BP network is not high,so the generalized regression neural network is adopted due to its characteristics of simple structure,easy implementation and undemanding quantity of training data.The experiment shows that the forecasting precision and accuracies are improved compared with the BP neural network.Then the smoothing parameters in generalized regression neural networks are selected through PSO,so that its values are no longer blind and the forecasting precision is further improved.