Wind Power Bearing Operating Status Based On SCADA Data Evaluation And Prediction

This paper takes the main component bearings of wind turbines as the research object,based on the temperature characteristic parameters,and through the analysis of the correlation of SCADA data,comprehensively considers the complex and changeable conditions of the wind turbines,and establishes a bearing operating state evaluation model based on the identification of operating conditions.,Can detect abnormalities in time before failures,early warning,and then establish a bearing operating state prediction model,predict its degradation indicators,and realize the tracking of its operating state trend.Real samples verify the effectiveness of the model.The research content of this article mainly includes the following aspects.(1)Data preprocessing of the SCADA system.Firstly,the reasons for the abnormal data of the wind turbine are introduced,and then the characteristics of the wind speed-power curve are analyzed,and the distribution of different types of abnormal data is summarized,and the identification and cleaning are performed respectively to obtain the normal operation of the wind turbine.Finally,the healthy samples are standardized.(2)Research on the evaluation model of wind turbine bearing operating state,based on the data samples during normal operation of the unit,and fully considering the complexity of the wind turbine operating conditions,a bearing operating state evaluation model based on identification of operating conditions is proposed.First select the appropriate working condition parameters and state parameters from the SCADA data,and then use the working condition parameters to use the clustering algorithm to divide the bearing’s working conditions,and use the state parameters to establish the bearing health benchmark model according to the different working conditions.After condition identification,the deviation degree between the current state parameters and the reference model is calculated.In order to eliminate the influence of random factors,a sliding time window is introduced to construct the bearing degradation index,and the running state of the bearing can be accurately evaluated through verification of examples.Identify its degradation process and realize early failure warning.(3)Research on the trend prediction of bearing operation status.Aiming at some of the shortcomings of the Elman neural network,a prediction method based on the genetic algorithm optimization of the Elman neural network is proposed.Finally,the optimized Elman neural network is used to predict the running state trend of the bearing,and compared with the prediction results of other methods,it is verified through examples that the optimized Elman neural network has a better prediction effect and can better track the bearing running degradation trend.It is better than several other methods and can provide a reference for bearing operation and maintenance.