An adaptive framework for reliability-based topology optimization under multi-source uncertainties

The lightweight design of continuum structures is essential in engineering, yet their complex service environments inevitably introduce the challenges of multi-source uncertainties. Reliability-Based Topology Optimization (RBTO) methods often face difficulties when addressing these uncertainties, frequently experiencing numerical oscillations under highly nonlinear constraints and exhibiting limited computational efficiency during the search for volume fractions. To address these issues, this paper proposes a novel Particle Swarm Optimization-Adaptive Topology and Reliability Optimization (PSO-ATRO) double-loop framework. The inner loop employs a hybrid PSO-ADMV strategy, which adaptively determines the optimal gradient descent step size to mitigate numerical oscillations during the Most Probable Point (MPP) search process. Simultaneously, the outer loop introduces a surrogate-assisted PSO-MIST strategy to seek the optimal material volume allocation. The effectiveness of the proposed framework is validated through several numerical examples, including a robotic arm link under stochastic loading. Comparative results demonstrate that the proposed method effectively avoids numerical divergence issues while strictly satisfying the reliability limits, providing application value for the safe and lightweight design of robotic structures operating in complex, uncertain environments.