Mileage Correction Model For Track Geometry Data From Track Geometry Car&Track Irregularity Prediction Model

To ensure both track elements and track geometry in good condition, railway maintenance of way departments often carry out maintenance and renewal works on track. There are three critical kinds of information (3W) to be accessed, comprising when and where to carry out maintenance works and what maintenance work to be conducted. With the guidance of condiction based track maintenance the key basis for obtaining3W is the track irregularity deterioration. Track Geometry Measurements of Track Geometry Car is one of the most important condition data sources for railway maintenance of way departments to monitor track irregularity. There are several key issues to be addressed to study the track irregularity deterioration according to TGM.This dissertation focused on two of these key issues. The first deals with the mileage error correction for TGM. The second focuses on the short-range prediction model for track irregularity. Research results in this dissertation will provide theorical and technical basics for implementing condition based track. Such maintenance technique will improve track system reliability and train travelling safety, lengthen track equipment life, and thus reduce Life Cycle Costs for track equipment.Based on resolutions of these two matters developed by scholars all over the world, firstly, employing Uniform Thresholding (UT), Map Matching, Cross Correlation, and Dynamic Programming, a novel, distinctive model was developed for correcting errors between measured mileages in TGM and track equipment mileages and is named Key Equipment based Mileage Error Correction method (KE-BMEC). These mileage errors are referred to as the first class mileage error throughout this dissertation. There are two steps for KE-BMEC to correct the first class mileage error. The first is to locate from TGM characteristic points of some equipment passed by TGC. For locating these characteristic points, TGM, railway station layout chart, and permanent-of-way equipment records stored in the database of Permanent-of-Way Management Information System (PWMIS) are used. PWMIS has been implemented and used throughout the entire rail network of China since2007. Recorded mileages of all sampling points in TGM are corrected according to mileages of these located characteristic points stored in PWMIS database. From performance analysis results for KE-BMEC, two main conclusions are arrived at as follows:(a) the first class mileage error of processed TGM by KE-BMEC is reduced significantly and thus is far smaller than the one processed by GPS based mileage correction system implemented in TGC; and (b) mileage errors over same sampling points between different inspections of TGC are below1meter in normal circumstances. These mileage errors over same sampling points between different inspections are referred to as the second mileage error.Secondly, employing Cross Correlation, Dynamic Time Warping (DTW), and Dynamic Programming, a novel model for correcting the second class mileage error was developed and is named TGM based Mileage Error Correction method (TGM-BMEC). DTW is used for correcting TGM mileage errors for the first time. For decreasing the second class mileage error as much as possible, TGM processed by KE-BMEC is treated further with TGM-BMEC. These TGM processed by KE-BMEC but not treated by TGM-BMEC are referred to as pending TGM, whereas the latest TGM processed by both KE-BMEC and TGM-BMEC as reference TGM. There are two steps for TGM to achieve its aim. The first step is to locate the sampling point in the reference TGM with the smallest distance from the one in the pending TGM. And the located sampling point is referred to as the corresponding sampling point. The second is to update for all. sampling points in the pending TGM the mileage according to the mileage of corresponding sampling points. Performance analysis results for TGM-BMEC illustrate that the second class mileage error corrected with TGM-BMEC is smaller than one sampling interval,0.25m, and is less than the one corrected with the existing mileage error correction models.Finally, based on track irregularity condition evolution characteristics, utilizing TGM processed with KE-BMEC and TGM-BMEC, a novel, distinctive short-range prediction model for track irregularity was developed and is abbreviated to TI-SRPM. TI-SRPM is designed for predicting track irregularity values on each day in a future short period over sampling points. The future short period is determined with the inspection interval of TGC. The following conclusions are drawn from performance analysis results for TI-SRPM as follows:(a) values of predicted track irregularity amplitudes by TI-SRPM are very close to ones measured by TGC;(b) at least80%track irregularity exceptions can be noticed one inspection interval of Track Geometry Car in advance and the prediction reliability for exceptions with bigger amplitudes is higher than that for exceptions with smaller amplitudes; and (c) track condition indices over many track section lengths can also be available one inspection interval in advance and these diverse indices provide data for refined management of track irregularity.