Research On Fault Diagnosis Of Coal Mine Fan Bearing Based On IVMD And MSSSA-ELM

Mine ventilator is the key equipment of mine ventilation system,which is of great significance to ensure the safety of coal production.Bearing is the core component of coal mine ventilator.Once failure occurs in work,it will not only seriously affect the normal operation of coal mine ventilator,but also may cause coal mine safety accidents.Therefore,this paper takes the bearing as the research object,decomposes its vibration signal and extracts characteristic information to realize the fault diagnosis of coal mine fan bearing.In the research part of signal decomposition:firstly,based on the advantages of variational mode decomposition(VMD)in solving mode aliasing problem,VMD is selected as the basic decomposition algorithm.Then,aiming at the problem of the unreasonable setting of key parameters in VMD algorithm affecting the decomposition effect,an improved variational mode decomposition(IVMD)was proposed.The energy difference ratio method and correlation coefficient were used to complete the adaptive optimization of decomposition scale K,and the minimum index of average sample entropy was constructed to optimize the punishment factor α.Finally,the actual signal decomposition experiments are compared with EMD and VMD to verify that the proposed IVMD algorithm can get the original signal frequency components better.In the feature extraction research part:firstly,based on the advantages of multiscale weighted permutation entropy(MWPE)in scale research,MWPE is chosen as the basic feature extraction method;then,the influence of the time series length N,component dimension m,scale factor s and time delay τ of MWPE on its entropy calculation is analyzed,and the optimization of the four calculation parameters is carried out separately,and the calculation parameters of MWPE under different fault Finally,the experimental comparison with the alignment entropy(PE),weighted alignment entropy(WPE)and empirical MWPE for practical signal feature extraction verifies that the difference of the parameter-optimized MWPE proposed in this thesis is more obvious in different faults,which lays a solid foundation for fault diagnosis,a solid foundation for fault diagnosis.In the fault diagnosis model building and system design study section:firstly,ELM is selected as the base classification algorithm based on the advantages of less training parameters,fast learning speed and strong generalization ability of extreme learning machine(ELM).To address the problem that randomly generated input weights and implied layer thresholds affect the classification accuracy of ELM,Tent chaotic mapping,uniformly distributed dynamic adaptive weight factors and the CorsiGaussian variational multi-strategy enhanced sparrow search algorithm(MSSSA)are introduced to find the optimization of ELM parameters;then,the MSSSA-ELM fault diagnosis model is established and compared with SVM,ELM,Then,the MSSSAELM fault diagnosis model was developed and compared with SVM,ELM,PSO-ELM and SSA-ELM models for fault classification experiments,and it was verified that the MSSSA-ELM fault diagnosis model with IVMD decomposition and MWPE features extraction had higher classification accuracy;finally,a coal mine ventilation fan bearing fault diagnosis system was designed using MATLAB GUI tool,which could diagnose the bearing fault types more concisely and efficiently.This thesis includes 70 figures,12 tables and 108 references.