TRACK PROFILE ESTIMATION USING A HALF CAR MODEL AND UNSUPPORTED SLEEPER DETECTION CONSIDERING THE DIFFERENCE IN TRAIN VEHICLE WEIGHTS

Abstract

This study presents an efficient approach to estimate the track profile and detect the unsupported sleeper using train vehicle responses. A half car model is used to represent the bouncing and pitching motions of the vehicle, and an augmented state space model is defined that includes the input track profiles in the state vector. A Kalman filter and Rauch-Tung-Striebel smoothing are then applied to estimate the track profile using the vehicle response observations. In the Kalman filter procedure, both the process noise covariance matrix and the observation noise covariance matrix are sequentially updated using the Robbins-Monro algorithm. The half car model parameters are also updated deterministically by solving an optimization problem based on constraints on the estimated track profiles. Furthermore, unsupported sleepers are detected by a large difference between the track profiles estimated using train vehicles with different weights. A statistical index is proposed to quantitatively identify the unsupported sleeper and distinguish it from random errors between two track profile estimates. The proposed approach is demonstrated on a numerical example using multi-body simulation.