Research On Simulation Modeling And Fault Location Of Passenger Train Brake Test System

The brake system of passenger train adopts air brake which has to be tested regularly to troubleshoot related faults and ensure the safety and reliability of the brake system.Among the common faults of the brake system,the emergency valve fault has the most severe impact and is the most difficult to locate.The wave speed method and the segmented search method are used in actual tests,but both of them have poor accuracy of location.The study of brake system fault laws requires customized experiments,which consume a lot of workforce and material resources.Using simulation to replace part of the experiments can reduce the experiment workload and reduce the blindness of the experiments which is a necessary technical method in the current engineering.How to improve the multi-classification algorithm to realize fault location is hot issue in recent years.This thesis will combine computer simulation and an improved multi-classification algorithm to realize the emergency valve fault location of the passenger train brake system.The main work of this thesis is as follows:(1)The structure of the 104 distribution brake is reasonably equivalent combining with the gas flow theory.With the help of advanced modeling and simulation tools,the single train brake system and the 18-carriage train brake system are constructed.The simulation model of single brake system passed the sensitivity test,stability test and emergency braking test.The simulation of the stability test of the train test is performed.The key indicators of the simulated wind pressure curve results are consistent with the test standards and actual test data,which verifies that the simulation model is safe and reliable.(2)In order to get further reasons for the emergency fault and to summarize the fault law,this thesis conducts a large number of simulations by modifying the internal parameters of the emergency valve.The curve of the relationship between the spring stiffness attenuation and the blocking allowance of the restricted hole-Ⅲ is drawn.The emergency valve fault of each section of the train is simulated based on the fault law and 1300 sets of emergency valve fault simulations are constructed.A complete fault data sample is constructed combining with actual test data and this data sample will be used for the fault classification modeling.(3)Aiming at the error accumulation problem in the binary tree support vector machine,this thesis designs a bottom-up classification tree structure to alleviate the error accumulation and perform the separation between classes.The reliability of the improved inter-class separability algorithm is verified in the three-class sample.(4)Aiming at the difficulty of selecting the optimal parameters of support vector machine and the premature convergence problem of particle swarm optimization algorithm,this thesis improves the particle swarm optimization combining genetic algorithm.The improved particle swarm optimization is used to optimize the parameters of support vector machine.The effectiveness of the algorithm is verified by multiple sets of UCI binary classification data sets.(5)Aiming at the problem of emergency valve fault location,this thesis improves the support vector machine multi-classification algorithm based on the optimal binary tree(IOBT SVM).The number of classification category is reduced through reasonable equivalence.After reasonably extracting the characteristics of the wind pressure curve,the fault classification tree structure is automatically created with the help of the inter-class separability algorithm.The comparison test results with other multi-classification methods verify the accuracy and the efficiency of classification in the IOBT SVM algorithm.