Urban Rail Transit Vehicle Key System Fault Classification Algorithm Research And Diagnosis System Development

The operational safety of urban rail transit has been well concerned bygovernment and people. MTR is the direct carrier for the carriage of passengers,which safe or not, directly and closely related to the life safety of passengers. Inaddition, with increasing complexity of the metro vehicle equipment in recentyears, of which corresponding failure rate is also rising. The brake system isone of the most important subsystems in the key equipment of metro vehicle,how to do fault diagnosis efficiently, fast and accurately is an important issuewhich is worthy of our study.This paper summarizes the development of fault diagnosis of rail transittechnology at home and abroad, and on the basis of the vehicle in-depth study ofthe Guangzhou subway operator, the application and failure of key equipment inthe current subway vehicles are summarized to identify the characteristics andlaws of the failure. On the basis of the above-mentioned requirements analysisand investigation, the paper described and designed a set of vehicle brakesystem fault diagnosis system, include the metro and the ground, established thefunctional structure of the various subsystems.Information collected is the basis of fault diagnosis. In the paper, firstanalyzes the source of diagnostic information and collection devices ofinformation, sensors, Information pretreatment equipment, which is to establishthe necessary foundation for subsequently diagnostic analysis, and monitoringpoints and monitoring information category of metro vehicle brake system isgiven in later chapters. This article also proposed the method of faultclassification of metro braking system which incorporated the failure data of themaintenance management system in depot and then integrated the way of faultclassification into the diagnostic model.Support Vector Machine (SVM: Support Vector Machine) is a new patternrecognition method developed on the basis of statistical learning theory,proposed by Vapnik based on statistical learning theory, the theory of VCdimension and structural risk minimization principle. There are many uniqueadvantages during solving the small sample, nonlinear machine-dimensional pattern recognition problems in the performance using SVM. In this paper, it isresearched on the principle of support vector machine, and the advantages anddisadvantages of several multi-classification methods. Finally it is selected oneof the binary tree support vector machine classification of fault diagnosis for thisarticle. And then it is established a diagnostic model of binary tree supportvector machine, which is realized at last.Finally, the software of the metro braking system fault diagnosis system isimplemented. First describes each functional module and system database, andthen shows the operation of the system interface. In the paper, the distributedmonitoring points of the subway braking system fault diagnosis is realized.Based on the proposed fault classification method, using the binary tree supportvector machine diagnostic model to accurately distinguish and classify thefailure of the braking system, which support the relevant direction study in thenear future and improve the depot historical knowledge base, provideconvenient maintenance of metro vehicles, which with good use.