Research On Composite Fault Diagnosis Method Of Rolling Bearing Based On Single Channel Blind Source Separation

Rolling bearings are the most common and important components in mechanical equipment,and rolling bearings tend to be in high-speed operation while the equipment is in operation,which is more prone to malfunction than other components in the equipment.In the actual engineering environment,failures of rolling bearings often occur in the form of composite faults,or multiple parts fail at the same time.In this case,multiple fault signals overlap each other,and there is also noise influence.The effective feature information cannot be directly extracted from composite fault signal.In addition,considering the limitation of installing the sensor in the actual engineering environment,it is difficult to satisfy the condition that the number of observed signals is greater than or equal to the number of source signals.The thesis takes the composite fault diagnosis of rolling bearing as the research content,and uses the single-channel blind source separation method as the tool to study the problems such as the source signal separation,feature extraction and state recognition in the composite fault diagnosis process of the rolling bearing combined the pattern recognition method.The main contents of the thesis are summarized as follows(1)The vibration mechanism of rolling bearing is analyzed.The method of calculating the characteristic frequency of each component of rolling bearing is summarized.The various failure modes of rolling bearing are summarized to select the faults with high frequency and high damage as the research object of this thesis.(2)The mathematical model of blind source separation and the basic principle of blind source separation are analyzed,and the extreme case of underdetermined blind source separation,that is,single channel blind source separation is analyzed.When the channel expansion is carried out,the pseudo component may be generated in the Intrinsic Mode Function(IMF)component when the virtual multi-channel observation signal is constructed with the Empirical Mode Decomposition(EMD)and Ensemble Empirical Mode Decomposition(EEMD)methods.For which a channel expansion method based on Complementary Ensemble Empirical Mode Decomposition(CEEMD)is proposed.To reduce noise interference,Total-Least Squares is combined with JADE to minimize the error function between the observed and estimated signals.(3)The plan of composite fault diagnosis id designed.Composite fault of rolling bearings is diagnosed using Support Vector Machines(SVM).The estimated signal of the source signal separated by the composite fault signal is decomposed by CEEMD.The noise is distributed into the IMF component to reduce the influence of noise.The sample entropy of several IMF components are calculated as features which are used to train the SVM model.At the same time.Particle swarm optimization is used to optimize the model to obtain the optimal parameters,which improving the accuracy of state recognition.Finally,the composite fault diagnosis of rolling bearings under single-channel conditions is realized.