An Optimization Algorithm Based On Fitting and Center Approximation Principle For Wind Power Prediction

Neural network attracts more and more attention in many research fields. However, neural network prediction is scarce in determining the hidden layer. In this paper an algorithm for rapidly discovering hidden layer nodes in neural networks is proposed. Establish a neural network to predict future wind power. Weather forecast information is used as an input data set for neural networks. Then two test sites in the hidden layer are identified by traditional methods. The fitting degree of the two test points is compared through the fitting judgment. BP NNW is founded base on the weather-report data, and the prediction of future wind power is finally completed