Estimation of Track Irregularity Based on Genetic Algorithm and Unscented Kalman Filtering

Abstract

Track irregularity is the main excitation source of wheel-track interaction. Due to the difference of speed, axle load and suspension parameters between track inspection train and the operating trains, the data acquired from the inspection car cannot completely reflect the real status of track irregularity when the operating trains go through the rail. In this paper, an estimation method of track irregularity is proposed using genetic algorithm and Unscented Kalman Filtering. Firstly, a vehicle-track vertical coupling model is established, in which the high-speed vehicle is assumed as a rigid body with two layers of spring and damping system and the track is viewed as an elastic system with three layers. Then, the static track irregularity is estimated by genetic algorithm using the vibration data of vehicle and dynamic track irregularity which are acquired from the inspection car. And the dynamic responses of vehicle and track can be solved if the static track irregularity is known. So combining with vehicle track coupling model of different operating train, the potential dynamic track irregularity is solved by simulation, which the operating train could goes through. To get a better estimation result, Unscented Kalman Filtering (UKF) algorithm is employed to optimize the dynamic responses of rail using measurement data of vehicle vibration. The simulation results show that the estimated static track irregularity and the vibration responses of vehicle track system can go well with the true value. It can be realized to estimate the real rail status when different trains go through the rail by this method.