Research On Rolling Bearing Fault Diagnosis In Nonstationary Condition Based On Order Analysis

Rolling bearing is one of the most widely used and easily broken part in rotating machinery,especially when the load is large and the speed change is dramatic in start-stop stage.Fourier analysis must meet the signal stability requirements in fault diagnosis.It will produce serious frequency smearing phenomenon in dealing with non-stationary signals.But order analysis can convert non-stationary signals to stationary signals for further analysis.So it has an important significance to research on the diagnosis method based on order analysis for bearing fault diagnosis in non-stationary condition.Firstly,according to the models for rolling bearing fault impulse signal in non-stationary condition,this paper provides an improved order envelope spectrum algorithm based on the theory of determining the order by low-pass filter.The envelope signal obtained from the original vibration signal is low pass filtered,then determine the re-sampling frequency in COT(Computed Order Tracking)and re-sample the filtered envelope signal.Finally the improved order envelope spectrum is obtained as the absolute value of the discrete Fourier transform of the re-sampled envelope signal.The results show that the improved algorithm reduces the re-sampling frequency and avoids the order aliasing compared with the traditional order envelope spectrum algorithm.Secondly,for the non-tachometer order tracking,this paper studies the instantaneous frequency estimation method based on Viterbi algorithm.In order to improve the efficiency of the algorithm,the paper improves its iterative process.And the advantage of this method is verified by linear and sinusoidal sweep signals,compared with the method of IFE based on STFT peak search.The method has good adaptability and antinoise capability.Thirdly,The paper proposes a novel order tracking method for bearing diagnosis based on tacholess and Vold-Kalman order tracking filtering technique,which overcomes the shortcomings of the traditional method depending on tachometer andother keyphase hardwares.In addition,studies the influence of weighting factor on the tracking performance of the filter.This method can separate order wave in time domain without phase bias and energy leakage.And the filter can separate the close and crossing orders in multiple shafts systems.Finally,through the test platform,the algorithm is applied to fault diagnosis for rolling bearing in non-stationary condition.Through the experimental analysis,the algorithm can diagnose the faults of rolling bearings in non-stationary condition effectively.