Research On Motor Bearing Monitoring And Fault Diagnosis System

Traction motor is the core device of the train drive system,equivalent to the "heart" of the train,realizing the conversion of electrical energy to mechanical energy and providing efficient driving force.Bearings are the basic components of the traction motor,which plays the role of supporting the rotation of the shaft,and about 40% of the failures in the motor are caused by bearings.The train running time in China is long,the temperature and humidity differences in different regions are obvious,and the severe working environment puts forward higher requirements for the monitoring and daily maintenance of motor bearings.In this paper,a motor bearing monitoring and fault diagnosis system is designed.The system is designed around three functions: data acquisition and receiving,data access and data analysis.The main work contents are as follows:(1)The data acquisition and reception function is divided into two parts: acquisition and reception.Although the existing data acquisition system is rich in functions,it is not designed for the characteristics of bearing vibration signals;This system designs a simple and efficient data acquisition unit and the corresponding upper computer software.The data acquisition unit achieves accurate high-speed signal acquisition,which is accurate in two aspects: one is accurate acquisition,and the other is low noise.The upper computer software communicates with the acquisition unit through TCP/IP through the producer,consumer,Queued Message Handler(QMH)and other structures,and realizes the high-speed receiving and storage of multi-sensor data,and solves the problem of data blockage and loss in the real-time transmission of a large number of data.(2)In the data access function,the acquisition system usually stores the data on the local U disk,which is easy to cause data leakage and loss.The upper computer software of this system uses Microsoft data connection file to call Access database,and realizes the functions of saving,reading,downloading and deleting vibration signals.(3)The data analysis function is the core of the motor bearing monitoring and fault diagnosis system,and some upper computer software realizes the vibration signal acquisition,reception,storage and data analysis functions in the two systems,separating the function of the diagnosis system.The data analysis function of the software takes the bearing vibration signal as the research object to judge the bearing state;Aiming at the problem of bearing fault classification,two kinds of intelligent fault diagnosis methods are proposed through Lab VIEW and MATLAB mixed programming in the upper computer software.The first fault diagnosis method is that of wavelet packet permutation entropy and SOM neural network.The method calculates the Permutation Entropy(PE)of the Wavelet Packet Decomposition(WPD)signal of the bearing vibration signal.It was constructed as the Feature vector of Self-Organizing Feature Map(SOM).After data training,SOM neural network realized 4classification of rolling bearing faults,and the recognition accuracy was 100%.However,the diagnosis result of this method is not ideal for the 10 classification problems of bearings.On this basis,a second fault diagnosis method is proposed: wavelet packet spread entropy-MRMR feature selection and HHO-KELM rolling bearing fault diagnosis method.The method uses Fast Independent Component Analysis(Fast ICA)to filter the bearing base noise signal,and applies Max-Relevance and Min-Redundancy algorithms.m RMR)selected14 sensitive features in wavelet packet Dispersion Entropy(DE),and selected Kernel based Extreme Learning Machine(KELM)with higher performance to replace SOM neural network.The Harris Hawks Optimization(HHO)algorithm is used to find the best penalty factor and kernel parameters for KELM.The experimental results show that the recognition accuracy reaches 100% for 10 classification problems with different fault degrees,different loads and different fault types.This paper completed the hardware selection,software design and algorithm research of the motor bearing monitoring and fault diagnosis system.During the experimental verification,TCP assistant was first used to verify the data reception and data access functions of the upper computer software,and then the validated upper computer software was used to test the acquisition accuracy and speed of the lower computer.Finally,the gearbox fault test bench was taken as the test object.The data acquisition unit of the system is used to collect four kinds of fault signals of the inner and outer rings,gear broken teeth and wear of the test bench at different speeds,and the data is transmitted to the upper computer software for data analysis.The three functions of the system’s data acquisition and reception,data access and data analysis are tested as a whole.The experimental results show that: The data acquisition unit of the system runs stably,and the upper computer software can identify the fault accurately,which has certain engineering application value.