Research On Short-term Prediction Of Wind Power Based On Optimized Neural Network

With the popularization of the concept of clean energy,wind power has gradually become one of the main power generation methods due to its abundant resources,mature technology and high utilization efficiency.However,due to the intermittent,uncertain wind,and the increasing safety of large-scale wind power grids,the precise of short-term power forecasting technology plays an extremely important role.The power forecasting enables the grid staff to arrange a more reasonable dispatch plan,accurately determine the start and stop,enable more wind power to be absorbed.To this end,the research in this paper focuses on how to improve the accuracy of short-term wind power forecasting.Firstly,the research and classification methods of wind short-term power forecasting are summarized.The process and theoretical basis of wind power forecasting are introduced.And operating parameters of wind power are analyzed.The data collected by the wind field was fitted by a polynomial of six times,and the results were in accordance with the wind speed-power curve.Secondly,BP neural network,Elman neural network and wavelet neural network are used to model and simulate short-term power forecasting of wind farms.The prediction effects and errors of three neural networks are compared.The grey wolf optimization algorithm is used to optimize the parameters of the three models.The single shallow prediction model is compared with the optimized model.The prediction results and error analysis are used to predict the shallow neural network based on grey wolf optimization.The predicted value of the model is closer to the true value,which confirms that the gray wolf optimization algorithm can improve the accuracy of power prediction.Thirdly,based on the shallow neural network for wind power short-term power prediction,a new prediction method of deep neural network combined model is proposed.Based on the convolutional neural network,the method uses the grey wolf optimization algorithm to initialize and optimize the hyperparameters,then fine-tunes the whole network through BP neural network,introduces the second correction of prediction bias,and establishes GWO-CNN-ec.Through simulation analysis,the prediction curve based on the GWO-CNN-ec combination model is closer to the actual value,especially in the high wind speed segment.It is proved that the deep convolutional neural network combination model has a significant effect on the improvement of short-term power prediction accuracy.Finally,the deep and shallow network prediction effects optimized by the grey wolf algorithm are compared,and it is confirmed that the deep neural network combination model has more advantages in wind power short-term power prediction.