基于GA-MADS混合算法的既有铁路平面线形自动重构

研究目的:既有铁路平面线形重构是铁路养护维修与增改建设计的重要基础,重构结果对列车运行安全、养护改建工程量将产生重要影响。既有方法通常先识别确定交点坐标,再逐交点优化半径缓长,属局部重构,难以实现全局优化;同时对约束的处理还不全面,重构的线形需要经过大量人工调整方可应用。对此,本文提出一种遗传算法混合网格自适应直接搜索(GA-MADS)的既有线路平面重构方法,实现线路平面整体自动优化重构。研究结论:(1)在遗传算法中融入MADS,有效克服了遗传算法在优化中随机性大、早熟收敛的缺点,实现平面线形整体重构;(2)各类约束在优化过程中被自动处理,无需人工后续调整;(3)研究成果已在各大铁路设计院6 000余公里的既有线增改建设计中成功应用,结果表明该方法可以自动产生满足工程约束的优化重构线形,大幅提高效率和质量。

Automatically Recreating a Horizontal Alignment Geometry of Existing Railways Based on a GA-MADS Hybrid Algorithm

Research purposes: Horizontal alignment recreation of exiting railway is an important basis for the railway maintenance and reconstruction. The recreation results greatly affect the train operation safety and the amount of maintenance and reconstruction works. Existing methods usually first identify the point of intersection( PI) and determine its coordinates. And then,the radius and transition curve length of each PI are optimized successively.However,through this recreation method,each PI can only be optimized locally. It is difficult to achieve global optimization. Moreover,constraints are not handled comprehensively,a lot of manual adjustments are needed before the recreation alignment can be applicable. In this regard,this paper proposes a horizontal alignment recreation method of existing railway based on genetic algorithm and mesh adaptive direct search algorithm( GA-MADS),which can realize the automatic optimization and recreation of the overall railway horizontal alignment.Research conclusions:( 1) The integration of MADS into genetic algorithm effectively overcomes the shortcomings of genetic algorithm of large randomness and premature convergence in optimization,which realizes the global recreation of the railway horizontal alignment.( 2) Multiple constraints can be handled automatically in the optimization process without manual subsequent adjustment.( 3) The research results have been successfully applied to the extension and reconstruction design of more than 6,000 km existing railway. The results show that the proposed method can automatically generate optimized recreation railway horizontal alignment which satisfies multiply engineering constraints,and improves design efficiency and quality greatly.