Bio-Inspired Computational Models for Multi-Objective Optimization in Complex Engineering Systems

Multi-objective optimization is now an indispensable tool in engineering and scientific practice, where two or more conflicting objectives have to be optimized subject to complex constraints. The classic evolutionary algorithms (EAs) and swarm intelligence (SI) schemes are competitive, but have many disadvantages, including early convergence, low diversity maintenance, and extreme expensive computational requirements, especially in high-dimensional spaces. To limit these shortcomings, this paper presents a hybrid bio-inspired optimization model that combines genetic algorithm (GA) crossover-mutation operators, particle swarm optimization (PSO) velocity-position updates, and differential evolution (DE) mutation strategies, and adaptive parameter control. The hybridization makes use of the exploration capability of GA, the fast convergence of PSO, and the strong local search of DE to be able to achieve a faster convergence and better diversity across the Pareto front. The framework is tested on six benchmark functions (ZDT1-ZDT6) as well as a real-world welded beam design problem. Experimental evidence indicates that the suggested method performs better than NSGA-II, MOPSO and MODE in terms of average hypervolume increase of 7.8 %, convergence time decreases on average by 12.5% , and spread diversity improves by 35 %. The results indicate that the framework suggested offers a scalable and a robust approach to solving complex multi-objective optimization problems.