Research On Fault Diagnosis Of Coal Mine Ventilator Bearing

With the development of coal economy,underground safety is becoming more and more important,and the working performance of ventilators has also put forward high requirements.The rolling bearing,as a precision part of the ventilator,is also a high-fault part,which has a direct impact on the working performance of the ventilator.Therefore,in this paper,we use the Adaptive local iterative filtering algorithm to process the vibration signals of ventilation fan bearings and construct a subset of features with obvious differentiation to classify and identify the bearing faults,taking into account the harsh environment of underground.Firstly,the current status of domestic and foreign research on ventilation fan fault diagnosis technology is analyzed and studied,and the structural characteristics of ventilation fans,bearing fault vibration mechanism and fault characteristic frequency are elaborated to provide the theoretical basis for subsequent fault diagnosis.Secondly,the parameters that affect the effect of ALIF decomposition are analyzed using the instantaneous frequency averaging method.And the combination of parameters of ALIF is optimized by using Particle Swarm Optimization algorithm.Three different types of simulated signals are compared with Empirical Mode Decomposition to verify that ALIF outperforms EMD in terms of modal confounding suppression.The improved ALIF is verified to be effective in improving the efficiency of bearing fault diagnosis by combining Hilbert with the improved ALIF.Again,the PSO-ALIF decomposition is performed on the bearing vibration signal,and the optimal modal components are selected to construct a hybrid domain feature set.The dimensionality reduction method combining PLS and Relief F is used to compare and analyze the clustering effect with that of PLS after dimensionality reduction,and it is verified that PLSRelief F can obtain feature subsets with more obvious differentiation.Finally,the improved ALIF is combined with Support Vector Machine for bearing fault diagnosis,and the parameters that affect the performance of SVM are optimized by using PSO.The PSO algorithm is also improved by combining the GS method with the SA algorithm to address the problem that PSO is prone to premature convergence.By analyzing the iterative search process and the fault identification effect,and comparing the analysis with SA-PSOSVM,PSO-SVM and SVM,the optimized classification model is verified to have a higher correct diagnosis rate.