Short-term Intelligent Combination Forecast Technology Of Wind Power

With the rapid growth of wind power installed capacity in China, theproportion of wind power in power grid is bigger and bigger; However,because of the volatility and intermittent of wind power which impacted bywind speed and wind direction, large-scale grid-connected wind power willinevitably bring serious problem to the operation of power system. While aneffective wind power prediction can provide a reference for grid scheduling,therefore, to study how to improve the veracity and reliability of wind powerprediction research has important practical and theoretical significance.This paper puts forward a new wind power intelligent combination forecastmodel by analyzing the advantages and disadvantages of existing windpower prediction method. The specific research content is as follows.First, in view of the nonlinear and non-stationary of power series andwind speed sequence, Elman neural network wind power prediction modelbased on phase space reconstruction of Kalman filtering is proposed, whichbuild foundation of wind power prediction.Second, in view of selecting Kernel functions and penalty coefficientparameters of support vector machine optimal model,a two dimensionaloptimization algorithm——adaptive colony algorithm is proposed whichcan be used to get the optimal model parameters,and take the forecast windspeed value as input and predicted wind power as output, on basis of this tobuild a short-term wind power prediction model optimized by adaptive beecolony algorithm,the validity and reliability of this model is verified by casesimulation.Third, two kind of combination model (Intelligent CombinationForecast Model Based on Entropy Criterion Colony Algorithm and Combination Forecast Model Based on Theil Coefficient and IOWAoperator) are proposed by using short-term wind power prediction modeloptimized by adaptive Bee Colony algorithm、Elman Feedback NeuralNetworks and BP Neural Networks. The former puts forwardMulti-dimensional optimization algorithm-entropy criterion colonyalgorithm to solve the weight definition problem for combination model,and from the perspective of relevance, the later adopts IOWA operator tomake the weight coefficient of each time point is only related to theprediction accuracy.Last， the validity and reliability of the two methods is verified by themeasured data from two wind farms, and compared with other conventionalmethods, this method has a higher accuracy. On basis of this, take theintelligent combination forecast model based on entropy criterion colonyalgorithm as main prediction model, and the combination forecast model ofTheil coefficient and IOWA operator as compensation model to conductwind power prediction, which get a good effect and show that it has a certainpractical value by the actual wind power prediction system testing.