Research On Wind Energy Prediction And Performance Evaluation Of Wind Turbine Based On Deep Learning

As a kind of renewable and clean energy,wind energy has attracted wide attention because of its huge development potential.As the main application of wind energy,the installed capacity of wind power generation is expanding at home and abroad.However,due to the uncertainty of wind energy,there are a series of problems in the operation and integration of wind power generation.With the development of big data and artificial intelligence technology,more and more industrial problems have been solved.Therefore,based on the measured data of wind farms and deep learning theory,wind speed prediction,wind power prediction and performance evaluation of wind turbine are studied in this paper.The main contents are as follows:The characteristics of energy conversion and operation are analyzed based on the principle of wind power generation.The abnormal wind power data is detected and removed based on Isolation Forest(IF)algorithm,and the feasibility and effectiveness of the method are verified by the results.The correlation between variables is analyzed by Maximum Information Coefficient(MIC),which serves as the basis for feature selection in the follow-up research.A wind speed prediction method based on Convolutional Recurrent Neural Network(CRNN)and Generalized Autoregressive Conditional Heteroscedasticity(GARCH)model is proposed.Firstly,a wind speed prediction model based on wind power data and CRNN model is established.Secondly,the heteroscedasticity of the prediction error is tested and the preliminary prediction error is processed by GARCH model based on the test results.The case study shows that the proposed method can improve the accuracy of wind speed prediction.A wind power prediction method based on Bidirectional Gated Recurrent Unit(Bi GRU)network and Random Forest(RF)algorithm is proposed.Firstly,a wind power point prediction model based on the wind power data and Bi GRU model is established.Then,an error prediction model based on the RF algorithm is established to correct the preliminary prediction results.Finally,the probability density of the corrected prediction error is estimated by kernel density estimation and the confidence interval is calculated.The case study shows that the proposed method has some advantages both in point prediction accuracy and interval prediction accuracy.A performance evaluation method of wind turbine based on Deep Belief Network(DBN)and Student’s t-test is proposed.Firstly,a power model based on wind power data and DBN model is established.Secondly,the significant difference is tested by Student’s t-test and the performance improvement is calculated based on the test results.The case study shows that the proposed method can effectively evaluate the performance improvement of wind turbine which is optimized by installing vortex generators.