Research On Fault Diagnosis Method Of Subway Plug Door Based On Improved EMD

The fault diagnosis of subway plug doors is an indispensable part in guaranteeing the smooth operation of urban subway system and also an important guarantee of passengers’ safe travel.How to recognize the fault type accurately and establish corresponding maintenance strategies when the plug doors in subway break down has been a research hotspot in the subway maintenance field.Based on the motor current data of subway plug doors collected from the simulation test platform of a depot in Wuxi Subway,this thesis combined the signal analysis method with the intelligent fault diagnosis method to explore more effective diagnosis strategies for the faults of subway plug doors.At first,the structure and working principle of subway plug doors were analyzed,and the occurrence mechanism of common faults was studied.The corresponding fault diagnosis strategies were proposed with four typical faults as the object of study and based on the signal analysis method for motor current.Then,the Empirical Mode Decomposition(EMD)theory was introduced.Aiming at the end effect and modal aliasing problem of EMD method,mirror image continuation and Ensemble Empirical Mode Decomposition were introduced for improvement,then simulation signals and measured signals were applied for comparative validation.Moreover,an adaptive optimal fault feature extraction model was established.The model decomposed the fault signal with the improved EMD to obtain the Intrinsic Mode Function(IMF)representing different frequency bands,and the false components were eliminated according to the Spearman correlation coefficient so that the dimensionless parameters and energy parameters in the effective components were taken as the original fault features,and the preferred features were further selected based on the sensitive indexes.According to experimental analysis,the fault features obtained by the feature extraction method herein was superior to the traditional method,as better classification effects could be obtained.Afterwards,based on the classification model of Multi-class Support Vector Machine(MSVM)optimized by Whale Optimization Algorithm(WOA),aiming at the WOA defect of loss of population diversity at the initial stage of optimization and iteration and the defect of easily falling into local optimum at the later stage,the chaos opposition-based learning and nonlinear convergence strategy were introduced to improve the algorithm,and then the improved WOA was tested for performance by trial functions.Furthermore,combining the feature extraction work,the fault diagnosis model of improved EMD and IWOA-MSVM was constructed,and different classification algorithms made comparative validation.The results showed that the fault diagnosis algorithm constructed in this thesis had an accuracy of 93.75%,and better comprehensive diagnosis performance was shown compared to traditional methods.Finally,an offline fault diagnosis host computer interface for subway plug doors was developed through the MATLAB GUI design platform.The interface was characterized by simple operation and high diagnosis accuracy,and the diagnosis system had good stability.It could provide a certain reference value for the visual analysis process of subway plug door fault diagnosis.