Abstract: Based on the wheel wear data of Guangzhou Metro, this paper analyzes the characteristics of wheel wear and re-profiling at first. With the application of the theory of Gaussian processes, a new data-driven model of wheel wear is given in terms of the variation of wheel diameter and flange thickness. A control limit policy for the wheel re-profiling decision is also proposed. According to the wheel wear model, the Monte Carlo simulation model/procedure of the re-profiling strategy is described in detail. Utilizing the Monte Carlo simulation, this wheel re-profiling strategy is optimized in order to maximize the expected lifetime of wheel finally. The results indicate that wheel wear is higher when wheels have a thicker or thinner flange, and the decreasing rate of the wheel diameter is independent of the thickness of wheel flange. Furthermore, the better selection for the wheel re-profiling strategies is that wheels should be restored to 30 mm when the flange is worn down to 27～28.5 mm, which could extend the expected lifetime of wheels to 13.2～15.3 years.

Key words: Gaussian process, Metro wheel, Optimization, Re-profiling strategy, Wear model