基于多目标优化策略的曲线钢轨非对称打磨设计

Design of asymmetric grinding for curved rail based on multi-objective optimization strategy

铁路曲线钢轨打磨一直是养护维修工作的重点之一,虽然钢轨的磨损不可避免,但适当打磨型面能有效减缓磨损。为了降低打磨去除量,改善轮轨接触性能,提高铁路系统的安全性和经济性,提出一种基于多目标协方差矩阵自适应优化策略(MO-CMA-ES)的曲线段打磨型面设计方法。首先,通过主成分分析方法(PCA),从钢轨磨耗型面中选取代表性的磨耗型面作为初始种群。接着,采用NURBS曲线拟合算法对这些型面进行曲线拟合,以建立参数化模型。随后,使用MOCMA-ES算法进行多目标优化,其中目标函数包括降低打磨去除量、轮重减载率和轮轴横向力。根据每个型面在不同目标函数下的适应度值,进行种群的进化与连续迭代,直到满足终止条件,即适应度函数值不再显著变化。最后,从最终的种群中选取目标适应度值最小的子代个体作为最优解。优化结果显示,优化型面降低了45.3%的打磨去除量,同时内外轨轨顶的打磨下降高度分别降低了0.84 mm和0.23 mm。静态轮轨接触方面,优化后的轮轨接触点分布更加均匀,且轨距角过渡区域的跳跃现象得到改善。动力学性能方面,优化型面降低了列车车体的横向加速度和轮轨横向力。此外,有效减小了轮对的横向位移,并微幅降低了轮轨最大接触应力。研究结果为曲线钢轨打磨提供了一种智能化的设计方法与参考。

Rail grinding of railway curves has always been a key focus of maintenance and repair work.While rail wear is inevitable,proper grinding profiles can effectively slow rail wear.To reduce the removal volume of grinding,improve the wheel-rail contact performance,and enhance the safety and economy of the railway system,a design method for curved rail grinding profile based on Multi-Objective Covariance Matrix Adaptation Evolution Strategy(MO-CMA-ES)was proposed.First,representative wear profiles were selected as the initial population using Principal Component Analysis(PCA)from rail wear profiles.Subsequently,a curve mathematical model was established by fitting these profiles using the NURBS curve fitting algorithm.Then,the MO-CMA-ES algorithm was used for multi-objective optimization,with the objective functions including reducing the removal volume of grinding,wheel weight reduction rate and axle transverse force.The population evolution was carried out with successive iterations based on the fitness values of each profile under different objective functions iteratively until the termination condition is satisfied,i.e.,the fitness function value no longer significantly changes.Finally,the offspring individuals with the smallest target fitness values were selected from the final population as the optimal solution.The optimized results show that the rail profile reduces the grinding removal by 45.3%,and the grinding reduction heights at the inner and outer rail tops are reduced by 0.84 mm and 0.23 mm,respectively.In terms of static wheel-rail contact,the optimized wheel-rail contact point distribution is more uniform,and the jumping phenomenon in the transition area of the gauge angle is improved.In terms of dynamic mechanical properties,the optimized rail profile reduces the lateral acceleration of the train body and the lateral force between the wheel and rail.Additionally,it effectively reduces the lateral displacement of the wheelset and slightly decreases the maximum contact stress between the wheel and rail.These findings provide an intelligent design method and reference for curve rail grinding.