Font Size: a A A

Fault Diagnosis Of Rolling Element Bearing Weak Fault Based On Sparse Decomposition And Figure Sparse Representation

Posted on:2016-07-16Degree:DoctorType:Dissertation
Country:ChinaCandidate:H C WangFull Text:PDF
GTID:1222330503993829Subject:Mechanical design and theory
Abstract/Summary:PDF Full Text Request
There are increasing requirements of harsh work and operating environment with the large-scale, complex, systematic, efficient, high-speed and heavy-duty of modern machinery equipment. The safe operation of machinery equipment not only involves the economic benefits of the enterprise, but also may cause serious injuries and adverse social impaction. It becomes more and more important to diagnose the machinery equipment’s weak fault and the compound faults opportunely so that corresponding effective measure can be taken to ensure the safe and efficient operation of the machinery equipmentRolling element bearing is one of the most widely used machinery parts of a mechanical device, and meantime it is also one of the easiest damaged components. The safe operation of rolling element bearing often determines whether the entire machinery device can operate safely or not. There are an important economy and safe implication in extracting the fault feature of rolling bearing opportunely prior to its complete failure. However, the fault feature of rolling bearing is very weak under the following three situations: The sensor collecting signal is installed far away from the virbration source; the early stage weak fault and the feature signal is interrupted strongly; compound faults. Besides, the traditional feature extraction methods basing on stationary and gauss signal theory could not be fit for handling the non-stationary and non-gauss signals occurring when fault arises in rolling element bearing. Sparse decomposition is a new and effective signal processing method which can match the impulsion characteristic signal when fault arises in rolling element bearing. The Sparse decomposition has been used in fault diagnosis of rolling element bearing to some extent. Basing on the above stated, the paper commences on the depth study on the weak fault diagnosis methods of rolling element bearing with the help of sparse decomposition. Inspired by the ideas of figure handling common used method-Non-negative Matrix Factor(NMF) and sparse decomposition, an intelligent fault diagnosis method of rolling element bearing compound fault is proposed by the improved NMF: Sparse NMF(SNMF).The main contents of the paper include the following aspects:(1) The backdrop and meaning of the selected topic of the dissertation are discussed from the perspective of theoretical analysis and engineering application. The research status of modern non-linear signal processing methods, feature extraction methods, fault diagnosis system and figure sparse representation are discussed roughly in the paper. Besides, the current research problems to be solved are summarized and the study contents of this article are established.(2) The thory of sparse decomposition and its main mathematics are disscussed. Besides, the common used sparse solving algorithms and the measurement mthods of sparseness are given. The construction of redundancy dictionary is also disscussed. The development history of figure sparse representation is stated simply, and one of figure sparse representation methods-the multisacale geometric analysis method is stated detailedly. The contents of the chapter lays a solid theoretical support for the following chapters.(3) Due to the interference of strong background noise and the severe signal attenuation phenomenon between the fault source and the sensor collecting the signal, the fault feature buried in strong background noise is very hard to extract using traditional signal processing methods. The minimum entropy de-convolution(MED) and sparse decomposition are combined to extract the feature of rolling bearing’ weak fault. The attenuation effect of transmission path is deconvolved effectively and the fault transient impulses are clarified using the MED technique. Sparse decomposition algorithm can match the transient impulses effectively using the best selected atomic compositions. The virtues of the two methods are combined to extract the weak fault feature of rolling bearing and better extraction results are obtained through simulation and experimental signal. Furthermore, the advantage of the proposed method over wavelet transform method, ensemble empirical mode decomposition(EEMD) method, frequency slice wavelet transform(FSWT), spectral kurtosis(SK) based method are verified through the rolling bearing’ inner race weak fault signal.(4) The feature of rolling bearing’s early stage weak fault is very hard to extract using traditional methods. The resonance sparse decomposition method-tunable Q-factor wavelet transform(TQWT) is the improvement of traditional one signal Q-factor wavelet transform and it is very fit for separating the low Q-factor transient impact component from the high Q-factor sustained oscillation components(rotating frequency and its harmonics) when fault emerges in rolling bearing. However, it is hard to extract the rolling bearing’s early weak fault feature perfectly using the TQWT directly. EEMD is the improvement of empirical mode decomposition(EMD) which not only has all the virtue of of EMD but also overcomes the mode mixing problem of EMD. The original signal of rolling bearing’ early weak fault is decomposed by EEMD and several intrinsic mode functions(IMFs) are obtained. Then the IMF with biggest kurtosis index is selected and handled by TQWT subsequently. At last, the envelope demodulation method is applied on the low Q-factor transient impact component and satisfactory extraction result is obtained. The effectiveness of the proposed method is verified through the early stage weak fault of rolling bearing accelerated life test.(5) The result of signal handling based fault diagnosis of rolling element bearing compound faults is not satisfactory usually because the compound faults signal is very complex due to the interferences of the single fault signals and the coupling phenomenon between them. Combine the figure handling common used method-NMF with the idea of sparse decomposition, an intelligent fault diagnosis method of rolling element bearing compound faults based on SNMF of bispectrum is proposed. Abundant fault information is buried in the bispectrum spectrum and it is decomposed by the SNMF method firstly, then the sparse coefficient matrix of the bispectrum is obtained which is used as the trian and test input of SVDD. Through the verification of three kinds of rolling element bearing compound faults(inner race outer race compound faults、outer race rolling element compound faults and inner race outer race rolling element compound faults) satisfactory fault diagnosis results are obtained at last. Besides, the fault diagnosis results based on NMF-SVDD is also given to verify the advantage of the proposed method.
Keywords/Search Tags:Rolling element bearing, Weak fault, Sparse decomposition, Resonance sparse decomposition, No-negative matrix factor, Sparse no-negative matrix factor, Figure sparse representation, Compound faults, Fault diagnosis
PDF Full Text Request
Related items