To address the frequent carbody lateral acceleration (CLA) overrun caused by turnout grinding defects after high-speed train wheel reprofiling, and improve high-speed turnout grinding quality as well as train operation safety and ride comfort, this paper takes a specific turnout as the research object and systematically conducts field tests, dynamic simulation, intelligent rail profile optimization and grinding strategy development. Field data are used to analyze carbody acceleration and rail profile characteristics; a vehicle–turnout coupled dynamic model considering nonlinear wheel–rail contact is established to analyze the influence of grinding defects. The genetic algorithm (GA), particle swarm optimization-non-uniform rational B-spline (PSO-NURBS) fitting, and Newton–Raphson-based optimizer-back propagation-particle swarm optimization-grey wolf optimizer-non-dominated sorting genetic algorithm III (NBRO-BP-PSOGWO-NSGA III) hybrid algorithm with clear coupling logic are adopted to optimize the target grinding profile. An improved multi-objective sparrow search algorithm-density peaks clustering (IMSSA-DPC) intelligent grinding model is constructed and applied to 20 groups of turnouts in field practice. Results show that the maximum CLA exceeds the 0.6 m/s 2 alarm limit for seven consecutive days at 300 km/h, with only 40% profile qualification rate. The model matches well with measured data, confirming that grinding defects cause profile distortion and raise CLA overrun risk. The optimized profile has better dynamic indices than the new rail profile in the full speed range. Field application shows profile deviation is controlled within ±0.1 mm, and CLA remains stable below the alarm threshold in 12-month monitoring, effectively solving the overrun problem.
Lin et al. (2026) studied this question.