Bayesian interactive search algorithm and incorporated improved fly-back approach for optimization of skeletal structures with dynamic constraints

Abstract Optimizing structural systems for multiple natural frequencies is a significant challenge due to the high nonlinearity and sensitivity of these problems to configuration changes. As traditional search algorithms often struggle, this study employs a Bayesian-reinforced search algorithm to address these complexities. The highly nonconvex nature of the search space for these problems also makes constraint handling crucial for effective performance. To this end, a new constraint handling approach was developed using the fly-back mechanism integrated with a quadratic search technique. The developed strategy is integrated with interactive search algorithm. The search capability of the proposed integrated method was evaluated on a suite of structural optimization problems, including planar, spatial, and dome structures. The numerical results demonstrate that the developed approach significantly reduces the required computational cost while yielding promising results in terms of solution accuracy and process stability.