Objective Conversion-Based Multiobjective Evolutionary Algorithm for Optimal Design of Braced Steel Frame Structures

This study proposes a multiobjective evolutionary algorithm based on objective conversion (MOEA/OC) for braced steel frame structure optimization. The algorithm innovatively integrates three mechanisms. A dynamic interval division strategy is employed to decompose the multiobjective problem into subproblems with adaptive constraints, eliminating the traditional reliance on dominance relationships. This strategy, combined with an archive-guided differential evolution operator, enables mutation and crossover operations driven by elite solutions, preserving diversity in the offspring while enhancing algorithm convergence. Benchmark tests were conducted, and the algorithm was compared with the nondominated sorting genetic algorithm II and the multiobjective evolutionary algorithm based on decomposition, demonstrating superior robustness and convergence. In a six-story frame example, the MOEA/OC algorithm outperforms the other two algorithms in terms of speed in finding Pareto optimal solutions. For the 12-story braced steel frame case, the structure designed using MOEA/OC achieves a cost saving of 43%, with inter-story drift angles controlled between 0.0029 and 0.0128, balancing economy and performance and demonstrating the potential of MOEA/OC in complex structural optimization. Additionally, the frequency distribution of braces in each bay of the steel frame layout on the Pareto-optimal solutions was analyzed using statistical counting methods, providing data-driven design recommendations for designers.