Research On The Application Of The Multi-Sensor Data_Fusion Technology In Fault Diagnosis Of Bearing

Multi-sensor data fusion technology applied to the military domain at first and developed gradually to the civil domain afterwards. Rolling bearing is a important part of revolving machine and sometimes in order to guarantee the machine's security and reliability, we need to monitor the rolling bearing and determine the flault type of bearing. The precision will be reduced since surrounding temperature and other disturbing facts affect sensor when diagnose fault of rolling bearing by single sensor. This paper intends to study new way of fault diagnosis of bearing by use of single and multi-sensor.Recently, the multi-sensor data fusion technology is a studying hotspot. Information fusion is a new technology of multi-source information processing synthetically and is to synthesize multi-source information coming from a certain object intellectually and gets more accurate and more complete estimate and judgment. Multi-sensor information fusion technology's prominent advantage is that information is redundant, permissibility of error, complementary, real-time and cheap.This paper has analyzed the bearing's common fault and vibration mechanism;chosen acceleration sensor and sound sensor to acquire liberation signal and sound signal after analyzing the token of bearing's fault;designed modularized program to control common DAQ card to acquire data based on LabVIEW;acquired and processed multi-source information by LabVIEW excellent hardware matching and strong and convenient signal processing ability;programmed by use of transferring MATLAB's wavelet analysis toolbox function to realize decomposing bearing's signal into multi- frequency bands signal;chosen Hilbert transition to demodulate the first layer's minutia signal analyzed by wavelet to get bearing's fault signal;analysed the fault signal through FFT transformation and diagnosed the fault of bearing by comparing the bearing's fault frequencies. That's not all, this paper explain how to get the bearing's eigenvalues through calculating spectrograph's a certain band frequency's energy and compose latent vector and transfer neutral network toolbox function in MATLAB to realize BP neutral network and fuse the feature information to diagnose the fault of rolling bearing and contrast the technique of single sensor fault diagnosis of rolling bearing to analysis the virtue and characteristic of the way ofusing the multi-sensor data fusion technique to diagnose the fault of bearing.