Key points are not available for this paper at this time.
In flexible manufacturing, simultaneously minimizing production time and cost while guaranteeing product quality under uncertainty remains challenging. This paper addresses the quality-aware flexible scheduling in multi-process system (QFSMPS). We propose a digital-twin (DT)-enabled, multi-stage dynamic optimization framework that decomposes the global workflow into sequential sub-workflows, each encompassing processing, inspection, and optional rework. A closed-loop mechanism models quality uncertainty via probability distributions and iteratively applies rework until quality thresholds are satisfied. A greedy heuristic generates initial schedules; these are refined by tabu-enhanced local search and evaluated by the Monte Carlo simulation to estimate time, cost, and expected quality. An adaptive, composite objective function balances the three criteria. Real-time feedback from the physical line continuously adjusts budget allocations and predicted metrics, ensuring global constraints on total time, cost, and minimum quality are met. Computational experiments on automotive benchmarks demonstrate that the proposed method outperforms non-dominated sorting genetic algorithm (NSGA)-II, multi-objective evolutionary algorithm based on decomposition (MOEA/D), and quality-optimal greedy algorithms, converging faster and more robustly across problem scales. The DT feedback markedly improves responsiveness and scheduling accuracy, indicating strong potential for deployment in dynamic manufacturing environments.
Luo et al. (Sat,) studied this question.