A Hybrid Evolutionary Algorithm for Production Process Optimization

In the production process, spare parts are assembled to become products. To improve product quality and reduce cost, it is required to determine whether to detect the quality of spare parts, semi-finished products, and finished products, and to disassemble unqualified products. Among these, the assembly of components constitutes a high-dimensional 0–1 nonlinear integer decision-making problem characterized by multiple steps, factors, and dynamic changes. It requires considering the defect rates of parts and finished products at each step, as well as decisions on whether to conduct inspections or disassemble, aiming to achieve overall cost minimization. This is a NP-hard problem, bring challenges to traditional algorithms. In this paper, we propose a hybrid algorithm HGA-PSO, which integrates the global search capability of GA and the local search capability of PSO. The algorithm employs GA subpopulations and PSO subpopulations for preliminary exploration by dynamically adjusting the usage ratios of GA and PSO. The top-performing individuals in the overall population undergo further refinement using PSO, while the introduction of randomness helps maintain population diversity, thereby further enhancing the algorithm's performance. Experimental results demonstrate that the HGA-PSO algorithm outperforms both GA and PSO on various instances in terms of solution quality.