Research On Intelligent Fault Diagnosis Algorithm Of Wind Turbine Based On SCADA Data

With the rapid development of wind power industry in China,it is of great significance to carry out intelligent fault diagnosis research on wind turbines to improve the reliability and economy of wind turbines.Aiming at the problem that it is difficult to establish accurate mathematical models for wind turbine blade icing faults and cog belt breakage faults,the research on the intelligent fault diagnosis algorithm of wind turbine based on SCADA data is carried out.The main research work is as follows:(1)Aiming at the problem that original SCADA data features are directly used for classification algorithm training and can not obtain ideal prediction results,a feature preprocessing method is proposed.The method firstly performs feature construction based on partial domain knowledge and statistical knowledge.Secondly,feature extraction is completed based on mutual information algorithm.The research results show that feature preprocessing effectively improves the performance of fault diagnosis models.(2)Aiming at the problem that SCADA data class imbalance is easy to cause the prediction performance degradation of classification algorithm,the SMOTE oversampling algorithm is introduced.This algorithm uses the linear interpolation method to achieve the balance between the number of normal instances and fault instances according to the ratio between normal instances and fault instances.The research results show that SMOTE oversampling algorithm effectively improves the performance of fault diagnosis models.(3)Aiming at the problem that it is difficult to establish accurate mathematical models for blade icing faults,an intelligent fault diagnosis method based on SCADA data is proposed.This method uses traditional machine learning algorithms and ensemble learning algorithms to establish six fault diagnosis models for blade icing.The research results show that XGBoost model,AdaBoost model and LR model have good diagnostic performance and generalization performance and this method can effectively diagnose blade icing faults.(4)Aiming at the problem that it is difficult to establish accurate mathematical models for cog belt breakage faults,an intelligent fault diagnosis method based on SCADA data is proposed.This method uses traditional machine learning algorithms and ensemble learning algorithms to establish six fault diagnosis models for cog belt breakage.The research results show that SVM model,AdaBoost model and XGBoost model have good diagnostic performance and generalization performance and this method can effectively diagnose cog belt breakage faults.The research results of this paper can provide a certain reference for the intelligent fault diagnosis of wind turbine.