A DRGO framework for dynamic optimization of rework assembly systems

Rework and assembly systems are prevalent in modern manufacturing, yet their optimization remains challenging due to dynamic feedback loops formed by material recovery. This paper constructs a comprehensive dynamic optimization model for the entire production process, integrating decisions on procurement, inspection, assembly, and rework/disassembly handling, with the goal of maximizing profit per unit cost. To solve this model, a novel Dynamic Roulette-wheel Greedy Optimization (DRGO) algorithm is proposed. DRGO integrates the local search capability of a greedy strategy with the stochastic exploration of roulette-wheel selection in each decision iteration, enabling it to adapt to system state evolution and efficiently search for the optimal decision sequence. The effectiveness and adaptability of the proposed model and the DRGO algorithm are validated through simulation experiments across various production scenarios. The results demonstrate that the method can generate economically rational dynamic strategies responsive to different cost structures and quality levels. This study provides a scientific and practical decision-making tool for enhancing operational efficiency and profitability in complex manufacturing environments.