Research Of Vibration Monitoring And Fault Diagnosis System For Wind Turbine Driveline

Wind turbine is a rotary machine with the capability of converting wind energy into electrical energy.Their operating conditions are very bad as they usually operate in remote areas or offshore,and are readily affected by climate and geographical environment.Broken-down of wind turbine is likely to cause significant economic losses and even casualities,so that the wind power equipment manufacturers and the wind farm management operators pay more and more attention to the safety of wind turbine’s operation,thus the condition monitoring and fault diagnosis of wind turbine are put forward.The driveline of wind burbine is the subsystem with high rate of failure and contributes most of the down time to wind turbine.The condition monitoring and fault diagnosis for the driveline can reduce the failure rate,maintenance cost and down time,so it has practical engineering meaning and economic benefit.Here,a three-layer structure of the remote condition monitoring and diagnosis system platform is designed,and an overall scheme design of the vibration monitoring and fault diagnosis system is also designed.Then some faulty signal processing algorithms for the driveline are discussed in some detail,and the fault feature extraction approach based on the max-relevance&min-redundancy principle is proposed to forecast the signal evolution trend of wind turbine generator.Finally,the program interface code writing is accomplished.The main research work is as follows:1.Based on the working principle,basic structure and operating condition of the wind turbine,the features and causes of the common types of faults are analyzed.Then,a three-layer structure of the wind turbine monitoring system platform is put forward.2.The characteristic of vibration signal of wind turbine with different types of faults is analyzed,and some actual operating and experimental signals collected from wind turbines are used to validate the signal processing and analysis algorithm adopted in this work from three aspects:time domain,frequency domain and time-frequency curves.3.The max-relevance&min-redundancy method based on mutual information is proposed to extract the fault signal feature from wind turbine.The effectiveness of the method is verified by using the DCS temperature signals of an operating generator and the BP neural network.4.The proposed vibration monitoring and fault diagnosis system is developed,including the designs of the database management system and the functions of the relevant softwares.