Research On Dynamic Security Early Warning Modeling And Algorithm Theory Of Rail Vehicles

With the rapid development of rail transit,the fault of rail vehicle and the hidden danger of safety are also increasing.The common faults are spring and damping faults of rail vehicle.The failure of the spring and damping is periodic and arbitrary.Therefore,I set up dynamic models of rail vehicle by taking the failure of spring and damping as the center of dynamic modeling and theoretical algorithm research.Fault models will set up on the dynamic models.Based on the data generated by the model,the characteristics of the spring and damping of the rail vehicle suspension system are studied.The statistical eigenvalues of the fault features are optimized.At the same time,Some algorithms of fault diagnosis are studied,and the neural network algorithm is improved based on genetic algorithm by myself.Firstly,the dynamic model and fault model of the track vehicle safety early warning are established by the Simulink of MATLAB.In view of the suspension device of the rail vehicle,different levels of fault are set up.Then setting up the fault model of the rail vehicle,including slight malfunction,moderate fault,serious fault and fracture failure of fault are divided in view of the suspension device of the track.Selecting the displacement spectrum of the six-level American orbit spectrum inversion as the input of the dynamic and fault model,and the simulation experiments are carried out.Secondly,according to the data produced by the model,Statistical parameters for fault tag are extracted.The Non dimensional statistical analysis method,the dimensionless statistical method are used to extract the fault characteristics of the rail vehicle.In the later period,the concept of fault feature relative contrast and self contrast are applied to the selection of fault features.The third step is to establish a screening evaluation index for the statistical eigenvalues of the suspension system of a rail vehicle to evaluate the difference between the fault features before and after the failure.The fourth step is to study the clustering algorithm,neural network algorithm and genetic algorithm.In the fault diagnosis of suspension system,neural network algorithm and neural network algorithm improved by genetic algorithm are applied.Finally,the application of the selected fault characteristics and the correct rate of setting up the fault on the experimental platform are verified by the experimental platform.The theory of security early warning and fault diagnosis algorithm is also studied in the thesis.At the same time,a scheme based on genetic algorithm is proposed to optimize the initial weights and thresholds of the neural network algorithm.Finally,the existing experimental platform is used to verify the optimization of the statistical characteristic parameters and the neural network algorithm based on the genetic algorithm optimization.The results of simulation and test show that the selection feature and the improved algorithm can improve the accuracy of the fault identification of the rail vehicle suspension system.