Optimisation of hardness profiles in high-speed train axlebox bearings

Abstract

To produce bearings with more reasonable hardness profiles, a finite element model was established to simulate different heat treatment processes. Based on the simulated hardness profiles, an artificial neural network model was established to fit the relationship between the heat treatment process parameters and the hardness profiles. Genetic algorithm was used to optimise the hardness profiles and the confirmation test was conducted. The results showed that the optimised hardness profile was closer to the optimisation target. The mean absolute error of hardness was reduced by 68.16% and the root mean square error was reduced by 66.72% after optimisation. This method can be used for hardness profile optimisation and provides a reference for the optimisation of the bearing heat treatment process.