Research On Fault Diagnosis Method Of Rolling Element Bearing Based On Vibration Signal Analysis

With the improvement of industrial modernization,mechanical equipment is gradually developing in the direction of large-scale,complex,integrated,and the components are tightly coupled.The fault occured at one component will cause the entire system failure.Rotary machine is one of the most widely used mechanical equipment in industrial production.Rolling element bearings are one of the critical components in rotary machine and rolling element bearing failures are the most frequent faults in rotating machinery.Therefor,it is of great significance for rolling element bearing fault diagnosis.The vibration signal of rolling element bearing contains rich fault information and is easy to obtain.The fault diagnosis method based on vibration analysis has become the most widely used technology in rolling element bearing fault diagnosis.However,the fault vibration signal has the characteristics of non-stationary,non-linear,complicated frequency components and high noise level.Thus,it is key to extract the fault features from the vibration signal and identify the rolling element bearing state.The dissertations are organized as follows:(1)Incipient fault detection of rolling element bearings based on fuzzy entropy and mathematical morphology fractal dimensionFor the problem that the incipient fault features in rolling element bearing is not obvious,a feature extraction method based on fuzzy entropy and mathematical morphology fractal dimension is proposed.Due to great generalization performance,the support vector machine(SVM)is utilized for state identification.With historical vibration signal data under normal operation and fault conditions,the incipient fault features are extracted and utilized to train the incipient fault detection model.The effectiveness of this method is validated by experiments.(2)A parameter optimized variational mode decomposition method based on Salp Swarm AlgorithmVariational mode decomposition(VMD)is an effective signal decomposition method with good anti-noise and anti mode mixing ability.But it is difficult to select parameters without any prior knowledge.A parameter optimized VMD method based on SSA is proposed,denoted as SSAVMD.A novel frequency domain mode mixing density is proposed to quantitatively measure the mode mixing severity between intrinsic mode functions(IMFs).The correlation coefficient between reconstructed signal and original signal is embedded in frequency domain mode mixing density to constructed objective function.Using SSA search the fitness of objective function to obtain the optimal parameter.(3)Rolling element bearing fault diagnosis based on SSA-VMD,mathematical morphology fractal dimension and ABC-ELMSince the characteristics of complicated frequency components and high noise level,the signal decomposition method is needed for rolling element bearing fault vibration signal.A feature extraction method is proposed using SSA-VMD decomposed the vibration signal into IMFs,then the mathematical morphology fractal dimension is utilized to extracted fault features.Finally,using extreme learning machine(ELM)optimized by artificial bee colony algorithm(ABC)identify the operation state.Using historical data to construct fault diagnosis model.The experiment validates that the scheme has effectiveness in rolling element fault diagnosis.