Wind Power Prediction Model Based On Deep Neural Network

Power prediction is an important research topic in wind power field.Due to the randomness,volatility and intermittently of wind power,it poses a severe challenge to the safety and stability of large-scale wind power grid connection.Therefore,the prediction of wind power has important reference meaning for power system dispatching and economic operation.In recent years,with the rapid development of deep neural networks,wind power prediction by combining numerical weather forecast data in wide area and space with deep learning algorithm has become a promising research direction.The main contents and research results of this paper are as follows:(1)Combining the characteristics of convolutional neural network CNN and gated Recurrent unit network GRU,a CGRU deep neural network is proposed.CGRU network combines CNN and GRU,making this network possess excellent data feature extraction and dimensionality reduction ability and excellent ability to process time series data.(2)A new grasshopper optimization algorithm(GOA)was studied in detail,and the concept of quantum rotation was introduced into the standard grasshopper optimization algorithm to improve it.A hybrid grasshopper optimization algorithm(HGOA)was proposed.Then,HGOA is used to optimize the network parameters of CGRU model,and a HGOA-CGRU wind power prediction model is established.(3)Format cleaning and normalization of the original wind power data are carried out.The pre-processed data are input into HGOA-CGRU,CGRU and ordinary GRU models for case prediction analysis,and the prediction results are compared horizontally.The prediction accuracy isevaluated by mean error and root mean square error.The results show that the HGOA-CGRU model has better prediction effect than the CGRU model,and the CGRU model has better prediction effect than the ordinary GRU model.The results show that CNN can improve the prediction performance of the GRU model,and the HGOA optimization algorithm can further improve the prediction accuracy of the CGRU model.(4)In order to compare the wind power prediction performance of BP and SVM models and HGOA-CGRU models.Wind power time series data from the same data set are input into three different models.By comparing the prediction results of different network models,it is found that the prediction performance of HGOA-CGRU model is better than that of the other two models,and the conclusions can be used as a reference in the future research and application of wind power prediction.