Recreating horizontal alignments of existing railways with a hybrid analytic and harmony search algorithm

After operating a long time, existing railway tracks gradually deviate from their initially designed alignments due to the wheel-track interaction, thereby compromising operational performance and safety. Therefore, existing railway alignments should be recreated through periodic maintenance to enhance train operations and guarantee operational safety. Currently, most methods in this field first identify tangent alignment sections and then curves based on measured points along the railway. Afterward, these identified geometric components are optimized to fit the actual alignment. However, for existing railways that are fairly circuitous, the above strategy is limited since the curves' geometric characteristics dominate those of tangents in recreating alignments. Thus, the precision of geometric identification is reduced and the recreated alignment may significantly deviate from the actual railway. To solve this problem, a three-stage optimization method is proposed. At the first stage, a dynamic threshold method is developed to flexibly identify the geometric attributions of all measured points based on the existing railway. For the second stage, an adaptive analytic method is proposed which combines two geometric recreation approaches that are devised for curves and tangents to determine a preliminarily recreated horizontal alignment. Moreover, to enhance the recreation accuracy of local alignment sections, a harmony search algorithm is customized as the third stage to further optimize the recreated alignment. Finally, the effectiveness of the proposed method is verified through detailed data and sensitivity analyses applied to a realworld case.