Research On The Detection Method Of Wheel Irregularities In Urban Rail

With the continuous advancement of society and the rapid development of the city,urban rail transit plays an increasingly important role.Therefore,it is necessary to perform online detection on the train wheels to detect faults in time.Based on summarizing domestic and foreign research,this paper has completed the following work:Firstly,based on SIMPACK,the vehicle-rail coupling dynamics model of urban rail train is established.The model includes the urban rail train vehicle model and the elastic line model as a source of data for subsequent algorithms.Secondly,analyzing the fault form and formation mechanism of the wheel not rounding;especially the mathematical expression of the wheel periodic polygon and flat wheel fault that can be accurately simulated by SIMPACK,and as the fault input of the vehicle-rail coupled dynamic model of urban rail train,get the corresponding orbital vibration response.Thirdly,the representation of wheel state characteristics based on variational mode decomposition and high-order statistics is studied.From the perspective of time-frequency analysis,the wheel state feature representation based on variational VMD and high-order statistic is proposed.By selecting the high-order cumulant of the corresponding mode component with wheel state discrimination as the feature dimension,the normal wheel is not realized.The state of the rounded abnormal wheel is distinguished.Fourthly,the diagnosis algorithm of wheel non-circular anomaly based on multivariate Gaussian distribution is studied.Aiming at the actual situation and characteristics of the train wheel detection field,the multivariate Gaussian distribution model based on the anomaly detection theory is used to judge the non-circular state of the wheel.Finally,the practical engineering application of wheel non-circular detection is studied.The function modules and design requirements were determined according to the requirements of the system,and the schemes were determined for communication,signal acquisition and data splitting from the perspective of the overall architecture.Finally,the wheel non-circular anomaly detection algorithm was applied by using the measured signals.Actual data verification was performed.