Study Of Wind Power Forecasting Based On Nonparametric Estimation Interval

Wind power prediction for maintaining power system stability and development of the power dispatching plan is important,however,wind power is intermittent,volatility and uncertainty will seriously affect the prediction accuracy,to solve this problem,proposed wind power forecasting method based on nonparametric estimation interval.At present,wind power forecasting focus on the value of power curve prediction,but its prediction accuracy is just passable,will seriously affect the safe operation of power system when the prediction error is large.The prediction of wind power interval is predicted from the perspective of probability interval,and it can provide probabilistic prediction results for the relevant power workers,and the prediction results are more real and effective.In the wind power point prediction,BP neural network as a kind of nonlinear prediction methods,when the known data species large quantity,prediction results can be obtained with high accuracy.In the prediction of the probability interval,compared to the parameter interval estimation method requires prior probability distribution hypothesis,non parametric interval estimation is a requirement of prior knowledge is relatively low,the overall probability density distribution method can use only part of the data can be inferred,especially suitable for the distribution results of asymmetry or estimation under the condition of multi peak.Therefore,based on the forecast result value power curve by using BP neural network method,combined with the non parameter interval estimation method can get the prediction results of response prediction error of interval distribution of real power and higher accuracy of the.In this paper,the main factors affecting the prediction accuracy of wind power are studied,and the classification of groups is completed according to the influence factors.Analysis of the existing wind power forecasting methods,selects BP neural network complete the value of wind power prediction,and the convergence speed of the problems such as the particle swarm algorithm is optimized.The statistical distribution of errors is studied for all kinds of groups.A nonparametric interval estimation method for wind power is put forward,and the power interval prediction of each point to be measured is completed by this method.Through a large number of experimental analysis,it is shown that the method is effective and accurate.