Research On Intelligent Sensing Algorithm Of Track Irregularity Based On Vehicle Dynamic Responses

Railway transportation is the most important traffic mode in China and is the lifeblood of the national economy and social development. With the substantial increase of train speed, carrying weight and traffic density, the dynamic interaction between vehicle and track is increasing. Damaging effects of vehicle on track is more serious, which even influence the train operation safety and ride comfort. Hence, it is of great significance to master the track status for ensuring the train operation safety and developing the maintenance plan. Track irregularity is studied in this dissertation, and an intelligent sensing method for track irregularity is proposed based on vehicle dynamic vibration responses, which combines with vehicle and track coupling model. Track dynamic irregularity can be acquired by this method when operating trains go through at the different speed. It can make up the deficiencies that track irregularity cannot be informed during the inspection cars detection cycle. So it has important theoretical and practical meaning for the train operation safety, track design and maintenance.Firstly, vehicle and track vibration characteristics are analyzed based on vehicle-track coupling system model under some kinds of conditions, such as different track condition, different speed, different vehicle types and different track irregularity. The simulation results further show the research significance and provide the data for the latter track irregularity estimation.A new algorithm based on micro genetic algorithm and vehicle-track coupling system model is proposed in the dissertation, which can acquire the track irregularity using the vehicle dynamic responses of operation train. In the algorithm track static irregularity is taken as an unknown parameter of nonlinear vehicle-track coupling system model, and then the solution of track irregularity is converted into a parameter estimation problem. Making the sum of error square between vehicle track model output and measurement value to be minimum is used as parameter estimation rule, and genetic algorithm is adopted to search the optimum solution in the solution space. Because the dynamic equations of vehicle-track coupling system are the large nonlinear equation set, an improved micro genetic algorithm is studied to reduce the computing time and workload. The restart step is removed from micro genetic algorithm and mutation operator is added. The optimal retention policy is adopted during the evolution process. The estimation results show the estimated track static irregularity error is in the allowable range.Track stiffness abrupt change caused by track infrastructure defections is the main reason which creates track stiffness irregularity. It can lead to rather large dynamic irregularity when the train goes through it. The dynamic responses of vehicle and track are analyzed when defections exist in the track substructure, and a recognition algorithm of track rigidity abrupt change is proposed based on Support Vector Machines, in which vehicle vibration responses are used. In order to improve the recognition precision, improved SVM is designed by GA and PSO methods to optimize the parameters of SVM. The recognition results show that these two methods are effective to enhance the accuracy and reduce calculation time and get the better results in track substructure defects recognition. Meanwhile, it makes a foundation for the more realistic track model.A nested algorithm based on GA and Unscented Kalman Filtering is investigated in the dissertation in order to improve the precision of track irregularity estimation, which is influenced by the error of measurement sensors and theoretical model. With the known characteristics of vehicle-track coupling system and measurement sensors, the dynamic responses of track are optimized to get higher estimation precision of track dynamic irregularity. The simulation results show that track static irregularity and all dynamic responses of vehicle and track are getting better after GA and UKF nested algorithm and the estimation error is decreased. Track dynamic irregularity, which operating trains go through at the different speed, can be estimated based on vehicle dynamic responses. The research in this dissertation provides a new solution to detection the track status from the vehicle devices.