Optimizing single-objective and bi-objective models for economic-statistical design of RGS plan using particle swarm optimization

This study investigates the optimization of repetitive group sampling (RGS) plans, focusing on minimizing implementation costs, reducing risks, and improving computational efficiency. Two single-objective models are introduced for the economic-statistical design (ESD) of RGS. The appeal of the ESD grows when the desired plan simultaneously approaches the ideal operating characteristic (OC) curve. Thus, the minimum angle method (MAM) is employed to formulate bi-objective Cost-MAM and ASN-MAM models. Addressing the limitations of traditional grid search approaches, the research proposes a novel particle swarm optimization (PSO)-based algorithm to enhance computational efficiency. Numerical examples, case studies, and a sensitivity analysis demonstrate that the Cost-MAM model significantly mitigates risk with only a marginal cost increase. A key insight is that achieving an optimal RGS with the lowest ASN does not necessarily lead to cost minimization. Furthermore, PSO algorithm outperforms both genetic algorithm and grid search approaches in terms of solution quality and computational efficiency.