Cost-Constrained Safety Factor and Tolerance Optimization for Thin-Walled Cylindrical Shells

Abstract

Traditional safety factor design adopts empirical fixed values, leading to unnecessary structural redundancy and failing to quantify failure probability. Monte Carlo Simulation (MCS) brings extremely high computational costs, while existing methods cannot fully optimize the relationship among machining tolerance, manufacturing cost and safety factor. To address these issues, this paper presents a refined design method combining Optimal Latin Hypercube Design, PSO-BP neural network surrogate model and African Vultures Optimization Algorithm (AVOA). The PSO-BP model replaces time-consuming finite element simulations. Combined with stress-strength interference theory considering confidence levels, an optimization model is established with machining tolerance as design variables and manufacturing cost as well as reliability as constraints, and solved via AVOA. The results show that under cost limits and high reliability requirements, load-carrying stability is improved by approximately 22.5 percent. The optimized safety factor is 1.03068, a reduction of 26.38 percent compared with the traditional value of 1.4. The proposed method realizes optimal tolerance allocation and structural lightweighting under manufacturing cost constraints.