Optimization design of curved rail profile for heavy-haul railways based on multi-period optimization method

Rail grinding is widely employed in heavy-haul railways to mitigate abnormal rail wear. However, frequent grinding can result in even more significant material loss than regular wear. This paper presents a method for adjusting track layout parameters to alleviate the severe rail wear problem on curved lines. First, field experiments and simulation analysis have been used to analyze the impact of track parameters on wheel-rail contact and the feasibility of parameter adjustment. A numerical optimization model has been established based on Genetic Algorithm-Levenberg Marquardt- Backpropagation neural networks (GA-LM-BP neural network), with the track parameters as the independent variable and the goal of reducing wear as the objective. The chaotic microvariation adaptive genetic algorithm has been used to obtain an optimized solution set. Finally, the optimization effects are revealed by comparing the rail wear characteristics obtained from the original values and the optimal solution set.