Research On Wheel Set Bearing Self-powered Sensing And Weak Fault Diagnosis Methods Of Rail Transit Vehicles

The safety is the precondition and the key element for the rail transit development.The safety of rail vehicles is the essential requirement to ensure the stable and efficient operation of the rail transit system.As the critical component of the rail vehicle,the wheel set bearing plays an important role in the train operation and its health state has a close relationship with the train operation safety.Therefore,the online monitoring of the wheel set health state is crucial to the rail transit train operation.The timely detection of the wheel set bearing fault will provide a solid and reliable foundation to the rail vehicle maintenance.However,the existed wheel set monitoring technology still can not meet this critical demand.In order to ensure the online monitoring of the wheel set bearing,this paper designed the research idea and and framework from the points of wheel set bearing vibration data sensing and health state identification.First,starting from the beginnings,this paper researched the self-powered sensing technology of the wheel set bearing vibration data.Then,considering the the environmental noise influence on the bearing vibration data under complex working conditions,this paper investigated the bearing weak fault diagnosis methods,including the fault feature extraction and fault diagnosis.Finally,to meet the demand of the filed maintenance,this paper designed and developed the prototype system of the wheel set bearing fault diagnosis.This paper made following research work:(1)Focusing on the problem of power supply during the process of data acquisition device,including the wheel set bearing monitoring sensor and the data transmission module,this paper proposed a method based on the triboelectric nanogenerator to collect the low-frequency vibration energy of rail vehicles based on the spring-mass model.This designed device can provide the intermittent power supply of the data acquisition device,including the wheel set bearing monitoring sensor and the data transmission module.Thus,it can ensure the timely data acquisition of the wheel set bearing data.Besides,the simulated platform of the rail vehicle vibration energy collection and data transmission were constructed.The simulated experiments proved the effectiveness of the vibration energy collection method.(2)Focusing on the problem of sensor installation limitation,sole signal channel and low signal to noise ratio,this paper proposed a blind source extraction of the bearing fault signal feature based on the signal decomposition and kernelized correlation.Thus,the bearing fault can be detected by analyzing the extracted signal.The proposed method was compared with the blind source signal feature extraction method based on the signal decomposition and second order correlation.The comparison results proved that the advancement and robustness of the proposed method.(3)Focusing on the problem of complex train operation conditions and signal contamination by impulsive noise,this paper proposed the cyclic correntropy method based on the cyclostationary theory for the signal feature extraction.This method could identify the bearing fault frequencies contaminated by the impulsive noise.To validate the effectiveness and advantage of the proposed method,the algorithm performance of kurtogram method was compared with the proposed method.(4)Focusing on the computation burden of the cyclic correntropy algorithm,combining the narrow band filtering theory,this paper proposes the fault feature extraction method based on the cyclic correntropy and narrow band filtering.Thus,the intelligent bearing fault diagnosis was achieved combining the least square support vector machine.To validate the effectiveness and advantage of the proposed method,the analysis results of traditional fault feature extraction methods were conducted.(5)In order to meet the demand of field applications of the bearing fault diagnosis,this paper designed and developed the wheel set bearing fault diagnosis software based on the App Designer of MATLAB.The software contains three proposed methods of this paper and several classical fault diagnosis methods.The software can provide a reliable support for the wheel set bearing state monitoring and maintenance.