Prestress multi-objective optimization method of the large opening cable dome with rigid inner ring

Abstract

The prestress optimization of cable dome structure holds significant practical importance in engineering applications. However, the intricate relationship between prestress levels and structural responses in complex structures renders the optimization process challenging. To address this, a surrogate model of structural response was first developed in this study based on a Backpropagation (BP) neural network, enabling the establishment of a nonlinear mapping between prestress and both structural displacements and support reactions. This approach markedly reduces the reliance on computationally expensive finite element analyses during the optimization process. Building upon this, an Artificial immune algorithm (AIA) and an improved Genetic–artificial immune algorithm (GA-AIA) were employed to conduct multi-objective prestress optimization of the structure. Considering the multi-objective nature of the problem, a Pareto-based optimization framework was adopted to obtain the Pareto front solutions that simultaneously account for vertical displacement, support reaction, and prestress levels. The optimal solution was then selected from the Pareto set using the coefficient combination method. The results demonstrate that the proposed neural network surrogate model exhibits high accuracy and computational efficiency, and that the presented optimization strategy possesses excellent global search capability and strong engineering applicability.