Adjustment of sampling locations in rail‐geometry datasets: Using dynamic programming and nonlinear filtering

Abstract

SUMMARYA track inspection car, which measures the shape of railway tracks (hereafter, rail geometry) while it is running on rails, discretizes the measurement results at nearly fixed spatial intervals. However, the distance between the discretized locations (spatial sampling intervals) may shorten or lengthen locally due to slipping or sliding of the car wheel, and this prevents the sampling locations from aligning with those of a dataset obtained with another measuring run. The authors developed an algorithm for approximately aligning the sampling locations of the measurement datasets obtained with different runs. First, they considered this problem as the selection of the series of data corresponding to each supervised data from a training dataset, which was constructed by interpolation in order to minimize the evaluation function of a number sequence representing data points. Next, they used the maximum likelihood method to identify the unknown parameters contained in the evaluation function. This problem uses two features of the evaluation function. The first is that the evaluation function is minimized by dynamic programming, and the obtained optimum sequence is equivalent to a maximum a posteriori (MAP) estimate in the Bayesian framework. The second is that by converting the evaluation function to a general state space representation, the log likelihood of the model that includes the parameters is obtained by a nonlinear filtering method. Also, to simplify the search for the identification, they devised a parameter search procedure for the parameters in the autoregressive (AR) model.

Similarity based position calibration for multisource track inspection data

“…Relevant methods share in common the properties of pair-wise and segment-wise aligning strategy, whereas the specific methodology can be generally classified into three kinds, i.e. Euclidean-distance-based, 1,8,9 dynamic-time-warping-based 2,3,10 and cross-correlation-based 11,12 methods. Li and Xu 8 utilize the squared sum of the difference between referential data and comparative data, which is equivalent to the Euclidean distance metric, as the objective function.…”

“…Li and Xu 8 utilize the squared sum of the difference between referential data and comparative data, which is equivalent to the Euclidean distance metric, as the objective function. Further, Kamiyama and Higuchi 9 add penalties to the same objective function, and treat the aligning task as an optimization problem, which is then iteratively solved by dynamic programming algorithm. In order to tackle the nonlinear shift in recorded positions, i.e.…”

Methods for aligning near-continuous railway track inspection data

“…Several sophisticated techniques have been developed by Xu et al., 3 Kamiyama and Higuchi, 4 and Xu et al. 5 that require significant computing processes.…”

Running position matching for the monitoring bogie and temporal subtraction analysis of derailment coefficient

“…In order to modify the position so that two waveforms can be well matched, dynamic programing (DP) matching, commonly used to match similar waveforms in numerous fields such as voice recognition (Sakoe and Chiba, 1978), could be effective. Kamiyama and Higuchi proposed a method to match the position of track inspection car using DP matching based on the track irregularity of gauge, which shows reproducibility without dramatical change in about a week (Kamiyama and Higuchi, 2006). However, the method cannot be applied to the monitoring bogie which does not measure the track irregularities.…”