Research On Fault Diagnosis And Condition Monitoring Technology For Rolling Bearing Of High Speed Train

The operation condition of high-speed trains is of great importance.Axle-box bearing is the core component of high-speed train,and its operation condition directly determines the overall performance of high-speed train,and affects the train's stability,safety and comfort.At present,the monitoring means for high-speed train are relatively limited,the condition assessment methods are relatively scarce,and the fault diagnosis technologies are not mature enough to meet the sustained development of high-speed train.For this reason,based on the systematic review of the condition monitoring thchnologies of rolling element bearing,this thesis has carried out the research of axle-box bearing fault diagnosis and performance degradation assessment technologies,and mainly completed the following research contents:Firstly,to overcome the shortcomings of MED,OMEDA,MCKD and MOMEDA in solving inverse filter,a novel inverse filter solution strategy is proposed based on particle swarm optimization(PSO)algorithm and generalized spherical coordinate transformation.The deconvolution method that uses the PSO algorithm to solve the inverse filter by maximizing the kurtosis(correlated kurtosis/D-norm/Multi D-norm)of the filtered signal can be referred to as PSO-MED(PSO-MCKD /PSO-OMEDA / PSO-MOMEDA).The simulation and experimental results show that PSO-MED and PSO-OMEDA can effectively overcome the influence of strong random impulses in comparison with MED,MEDA and OMEDA,and PSO-MCKD and PSO-MOMODA can effectively overcome the influence of the deviation of the fault period in comparison with MCKD and MOMEDA.Secondly,to overcome the shortcomings of kurtosis,correlation kurtosis,D-norm and Multi D-norm,a new fault characteristic parameter,impulse norm,is proposed based on the sequence statistics.A new deconvolution algorithm,PSO-MIND,is proposed based on impulse norm and PSO algorithm.PSO-MIND takes the impulse norm of the filter signal as the optimization objective,and uses PSO algorithm to search the global optimal solution of the inverse filter coefficients.The simulation and experimental results show that compared with MED and PSO-MED,PSP-MIND can effectively overcome the influence of random impulse.Compared with MCKD,PSO-MCKD,MOMEDA and PSO-MOMEDA,PSO-MIND has similar effect and can get rid of the influence of fault period.Then,a new signal decomposition algorithm,complementary complete EEMD with adaptive noise(CCEEMDAN),is proposed.A time–frequency analysis method based on CCEEMDAN and PSO-MIND is proposed for fault detection of rolling element bearings.First,a raw signal is decomposed into a series of intrinsic mode functions(IMFs)by using the CCEEMDAN method.Then a sensitive parameter(SP)based on adjusted kurtosis and Pearson's correlation coefficient is applied to select a sensitive mode that contains the most fault-related information.Finally,the PSO-MIND is applied to enhance the fault-related impulses in the selected IMF.High-speed train bearing outer race fault and rolling element fault signals are used to demonstrate the effectiveness of our proposed method.The comparisons demonstrated the superiority of the proposed method over the individual CCEEMDAN and PSO-MIND methods for rolling-element-bearing fault diagnosis.The comparative analysis also demonstrated that the proposed SP had an advantage over kurtosis in selecting the sensitive mode from the resulting CCEEMDAN signal.Also,some comparisons with variational mode decomposition(VMD)and the fast kurtogram are conducted to show that the proposed time-frequency method is more effective and robust in revealing the fault information.Finally,based on hypothesis testing technologies and early fault diagnosis method,a new condition degradation assessment scheme for rolling bearings is proposed.The new scheme uses F/KS/U test statistics to quantify the performance degradation trend of service bearings,uses hypothesis test results to identify the moment when the bearing condition changes,and uses early fault diagnosis method to identify the bearing's fault type.The experimental results show that the proposed method can realize the quantitative evaluation of the state of the bearing,the tracking of the degradation trend and the accurate identification of the fault during the full life cycle of the bearing,and can provide theoretical and technical support for the online monitoring of the axle box bearing.