Multi-population competition genetic algorithm for assessing long-span cable-supported bridge girder’s maximum deflections and rotation angles under live loads: A direct optimization task solution

Abstract

This study addresses the stability problem of long-span cable-supported bridges (CSBs) under live loads, which requires an accurate estimation of maximum girder deflection and rotation angle. In contrast to the cumbersome influence line method or analytical method, which ignores the structural nonlinearity of this bridge type or uses too many constraint conditions, we convert this problem into an optimization task. Since the number of segments of distributed live loads under which maximum girder deflection and rotation angle occur (i.e., the number of optimization variables) is unknown due to CSB’s structural complexity, a multi-population competition genetic algorithm (MPCGA), inspired by the population competition theory in ecology, is applied. It incorporates the Lotka-Volterra competition model to depict the changing sizes of the competing populations. We designed the interspecies migration and exchange mechanism for the above engineering problem and ran ANSYS to compute individual fitness. This algorithm offers high accuracy and efficiency in solving the maximum girder deflection and rotation angle of the long-span CSB, the positions where the maximum girder deflection and rotation angle occur, and the corresponding live load patterns. Finally, the proposed method is validated by a case study of a hybrid CSB with a main span of 1400 m. The calculation results obtained via the conventional influence line and proposed methods are compared, proving the latter’s supremacy.