Research On Fault Diagnosis Of Motor Bearing Based On Variational Mode Decomposition

In most electromechanical rotating devices,bearings are a common rotating part and are the most widely used.The main load in the electromechanical rotating equipment is to bear and transmit through the bearing,and the bearing has much worse working environment than other mechanical equipment or components,so the rolling bearing in the electromechanical rotating equipment is most vulnerable to damage.The early diagnosis of rolling bearing faults can effectively avoid serious accidents and has important economic value and practical significance.In this paper,rolling bearing is taken as the research object,and a series of research work on bearing fault diagnosis is carried out.The main contents are as follows:First,in order to break through the limitations of wavelet basis and decomposition layer selection in wavelet decomposition method and overcome the defects of end effect and spurious component in the empirical mode decomposition method,it is possible to extract faults that can be characterized from bearing vibration signals with weak fault information.The effective information of the feature,this paper proposes the use of matching pursuit algorithm to optimize the variational modal decomposition method and effectively decompose the bearing vibration signal.The paper first analyzes the effect of the variational modal decomposition method on the vibration signal of rolling bearing by constructing the simulation signal.The simulation results show that the variational modal decomposition algorithm can complete the frequency well for non-stationary signals such as rolling bearing vibration signals.Decompose to overcome the problems of end-effects and spurious components.Then the matching trajectory algorithm is used to optimize the variational modal decomposition,and the vibration signal of the measured bearing is decomposed.The decomposition result analysis shows that the optimized decomposition method greatly improves the decomposition effect of the bearing vibration signal.Secondly,in non-stationary signals such as vibration signals of rolling bearings,some characteristic parameters form a one-to-one correspondence with the type of fault and the degree of fault.Using these characteristic parameters,the type of fault or the degree of fault can be characterized,and how to extract directly.These parameters are the key to fault diagnosis.The average value can reflect the overall trend and reliability of non-stationary signals such as rolling bearing vibration signals;multi-scale entropy can analyze the complexity of multi-scale data such as vibration signals.Based on the characteristics of the two parameters of mean value and multi-scale entropy,this paper proposes a multi-scale entropy mean deviation value feature extraction method based on variational modal decomposition.The feature parameter value of the method can achieve a one-to-one correspondence with bearing failure type.with the advantages of overall trends and high reliability.Finally,fault type identification is accomplished by using the fault feature parameters extracted earlier in this paper.The traditional BP neural network algorithm has the disadvantages of slow network convergence,unstable results,etc.This is due to the fact that the initial weights and initial thresholds of the traditional BP neural network algorithm are generated randomly.Particle swarm algorithm has the problems of parameter selection,premature convergence and stability.Combining the above problems,this paper uses the improved particle swarm algorithm to optimize the initial weights and initial thresholds of the BP neural network.The variation of the multi-scale entropy average deviation is used as a network input for classification testing to complete the bearing fault diagnosis.