Modified Nondominated Sorting Genetic Algorithm for Optimizing Turnout Stiffness in High-Speed Railways

The stiffness of high-speed railway turnout systems critically governs wheel–rail dynamic interaction and the longevity of rail infrastructure. However, traditional stiffness design methods often fail to consider the nonuniform stiffness distribution and its impact on wheel–rail interaction. This study presents a modified nondominated sorting genetic algorithm for optimizing the track stiffness of high-speed railway turnouts. The optimization incorporates rail deflection gradient, wheel–rail vertical force, and rolling contact fatigue index as performance objectives, constrained by turnout stiffness gradients and rail deflection change rates. Design variables include the stiffness parameters of six substructures covering mainline and divergent lines, with their bounds defined in accordance with engineering standards. The modified nondominated sorting genetic algorithm II (NSGA-II) is employed to derive a representative Pareto-optimal solution set. These solutions are further validated through explicit finite element modeling and dynamic simulation under realistic operational conditions. The results demonstrate that the optimized stiffness layout effectively improves mechanical performance. Specifically, the maximum stiffness decreases by 22.5% (from 109.11 to 84.55 kN / mm ), the longitudinal deflection gradient decreases by 58.8% (from 0.345 to 0.142 mm / m ), and the turnout stiffness ratio decreases from 2.18 to 1.73. Corresponding dynamic performance enhancements include an 8% reduction in wheel–rail contact forces and a 20% attenuation of vehicle vibration acceleration. Overall, the proposed modified multiobjective optimization approach enables more uniform stiffness transitions, reduces mechanical impacts in critical turnout regions, and contributes to enhanced structural durability and safety in high-speed railway operations.