Multiple Fault Diagnosis Of Rolling Element Bearing Based On Signal Decomposition

Rolling element bearing is a key component of train running part,its reliability and stability are very important to the safety of train.Due to long-term operation under harsh working environment such as high temperature,high pressure and heavy load,rolling bearing may simultaneously occur two or more compound faults which are coupled with each other.However,compound faults are not a simple superposition of single fault,but various fault features add together and interfere with each other,thereby making the fault feature extraction and identification more difficult.Most of the methods concentrate on the research of single fault,and don’t fully take the situation when different faults excite multiple different resonance frequency bands under multi-fault into consideration.What’s more,the interferences in the collected vibration signal could be diverse,which affect the accurate detection of resonance frequency bands of bearing.In view of the deficiencies of the existing diagnostic methods,the paper concentrates on the compound fault diagnosis algorithm of rolling bearing under random impulse interference and gear meshing harmonics interference by using signal decomposition,the specific contents and results are as follows.(1)To solve the problem of multiple bearing faults,the vibration simulation models of commonly occurred bearing compound faults of roller,outer race and inner race are established.Then,the time domain,frequency domain and envelope spectrum characteristics of compound fault vibration signals are studied.Finally,the major problems existed in compound fault diagnosis are elaborated.(2)Aiming at the problem of wrong location of the optimal resonance frequency bands of the rolling bearing under random impulse interference,a compound fault diagnosis algorithm for rolling bearing based on the squared envelope spectrum sparsogram is established.The improved redundant second generation wavelet packet transform(IRSGWPT)that can best match characteristics of the vibration signal is used to decompose the multiple fault signal to achieve the separation of different fault signatures.Then the squared envelope spectrum sparsity index which is more sensitive to fault impulses and robust to interferences is proposed,thus the squared envelope spectrum sparsogram is constructed to identify multiple resonance frequency bands excited by different faults under large extraneous shocks.Simulation and experimental results show that the method can accurately identify the bearing demodulation frequency bands and effectively extract multi-fault features under random impulse interferences.(3)Aiming at the problem that the gear meshing harmonics mask the transient impulse characteristics of the bearing and affect the identification of bearing high-frequency resonance bands,a novel rolling bearing fault diagnosis method based on complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)and the intrinsic mode function(IMF)multi-index fusion is proposed.The compound fault vibration signal is decomposed into multiple IMF components by using the CEEMDAN,then the IMF evaluation index with multi-index fusion is presented to reconstruct the fault IMF and weaken the gear meshing interference.Subsequently,the squared envelope spectrum sparsogram is applied on reconstructed signal to extract high frequency fault characteristic information of bearing and effectively diagnose the fault type.Simulation results show that the method can realize the compound fault diagnosis of rolling bearing under gear meshing signal interference without the use of tachometer.