Study On Fault Diagnosis Of Urban Railcar Door Based On Random Forest Algorithm

With the acceleration of urbanization,urban rail transit has developed rapidly as an efficient,rapid and large-capacity transport.The development of urban rail transit not only facilitates public travel,but also effectively reduces traffic congestion and environmental pollution.With the development of urban rail transit,the structure of rail transit becomes more and more complex,and the corresponding failure rate increases year by year.The community pays more and more attention on the safety of urban rail transit.Door system as an important part of the urban rail transit,its frequent failure will have a serious impact on the safe operation of urban rail transit.How to solve the failure of urban railcar door system to enhance the safe operation of the urban rail transit is a technical problem to be studied urgently.Among the various types of doors used in urban rail trains,the plug door is more and more widely used at home and abroad due to its good sealing performance and its design to meet aerodynamic requirements of vehicles.This subject intends to take the city rail train plug door as the research object.In view of the problems of the existing urban rail vehicle door fault diagnosis method,such as the slow diagnosis speed and a large number of troubleshooting data to be not properly used,a method of fault diagnosis of the rail vehicle door based on random forest is proposed.The method is used to establish the fault diagnosis model of the railcar door,and the feasibility of the method is verified through an example.Due to the use of stochastic forest method for fault diagnosis,its diagnostic performance will change with the splitting node selection rules.Therefore,in order to optimize the fault diagnosis model and improve the accuracy of fault diagnosis,this paper uses the information gain rate to improve the traditional stochastic forest algorithm,and proposes a stochastic forest fault diagnosis method based on information gain rate.And use this method to establish the railroad train door fault diagnosis model.The case study shows that the random forest method based on the information gain rate can diagnose the fault of the railcar door,and the diagnosis rate is higher and the effect is better.Finally,in order to apply the improved stochastic forest algorithm to practical engineering more,this paper established a fault diagnosis system for the railcar door,and verified the stability of the system in the test.Related work is mainly reflected in the following aspects:(1)Studied the structure of the sliding door and the working principle of each part,and analyzed the failure mode and failure reason of the sliding door.Failure modes,Effects Analysis(FMEA)are used to preprocess the failure data of urban rail transit train doors,and to establish FMEA tables and concise decision tables,which lays the foundation for in-depth research on door fault diagnosis methods.(2)A fault diagnosis method for urban railcar doors based on stochastic forest algorithm is proposed.The stochastic forest algorithm was used to train the train door fault data in the simplified decision table to establish the fault diagnosis model of the rail train door.The model can be used to predict and diagnose the fault data of the new railcar door.(3)In order to further optimize the fault diagnosis model of the railcar door,the information gain rate is introduced into the random forest to improve the traditional random forest algorithm,and a stochastic forest fault diagnosis method based on the information gain rate is proposed.And this method is used to establish the fault diagnosis model of the railcar door.Through the example analysis,the accuracy of fault diagnosis models established by these two methods in railway train door fault diagnosis was compared.Finally,the parameters of the stochastic forest model are analyzed and discussed,and the relationship between the stochastic forest model’s diagnostic performance and the number of single classifiers it contains is obtained.(4)Based on the fault diagnosis model of urban rail transit vehicle doors established in this paper,we choose QT as the system software development platform and Python language as the system programming language to build the fault diagnosis system to realize the fault diagnosis of the vehicle door.The system mainly includes database module,knowledge acquisition and fault diagnosis module.