Data-Driven Wheel Wear Modeling and Reprofiling Strategy Optimization for Metro Systems

Abstract

With the rapid developments in metro systems worldwide, more research concerning optimization of maintenance actions is needed, because the availability and service state of a metro system directly influences the daily activity of a city and its people. In particular, the prediction of wear and maintenance optimization of wheels is significant. Maintenance costs for a rail track subsystem represent more than half the total maintenance costs for a metro line. A hard rail–soft wheel compromise extends the life of the rails and increases the wheel replacement frequency with economic benefits. An improved strategy for predicting and maintaining wheel wear will allow agencies to improve reliability, enhance safety, and maximize wheel life while minimizing relevant costs. In this study, historical data are used to analyze wheel wear curves, and the flange thickness and wheel diameter are identified as the most important profile parameters. A new data-driven model of wheel wear trends is given for variations in wheel diameter and flange thickness. An approach for optimizing the wheel reprofiling strategy is based on this model and determines the optimum reprofiling point that maximizes wheel life while minimizing relevant costs. An initial case study on the Shanghai, China, metro network shows that the proposed approach can provide a reasonable solution for optimization of the reprofiling strategy.

A hybrid optimization strategy for the maintenance of the wheels of metro vehicles: Vehicle turning, wheel re-profiling, and multi-template use

“…vehicle turning, wheel re-profiling, and multi-template use in a complementary manner. This approach improves the methodology of using re-profiling alone that was proposed before, 11 and it performed better in the initial tests.…”

“…Based on these measurements, a data-driven optimization strategy for wheel re-profiling was developed. 11 Several important insights were obtained from this historical data analysis, which motivated the study described in this paper.…”

“…In this strategy, re-profiling starts when the worn wheel profile is outside of its acceptable range. The second group was operated with the re-profiling strategy proposed in Zhu et al 11 In this strategy, an optimal re-profiling point was introduced. The third group was operated with the hybrid strategy developed in this paper.…”

“…Starting from 2009, its wheel-sets have been measured and recorded regularly. In particular, the historical data from the period 2009 to 2013 were used to generate an optimal hybrid strategy for wheel maintenance, while the historical data from 2014 were used to test the hybrid strategy and compare it with the strategy proposed in Zhu et al 11 Based on the insights already obtained, two important measures (Sd and D) were used for the development of the strategy. First, with the data from the period 2009 to 2013, the improved system was used to model the wheel wear curves.…”

“…Thanks to the abundant historical wheel profile data recorded by the Shanghai Metro system, we have been able to develop a data-driven wheel wear model that can be used to provide a description of the wheel wear process, on the basis of which we propose an optimized wheel re-profiling strategy for metro vehicles. 11 An initial case study on the Shanghai Metro network showed that the approach provided a reasonable solution for the optimization of re-profiling strategies. 12 However, acknowledging that re-profiling is not the only strategy for dealing with the wheel wear, in the past 3 years we have developed a more advanced optimization approach that combines another two strategies, namely, vehicle turning and multi-template use , to give as near an optimal solution as possible.…”