A study of a matrix manufacturing system scheduling method considering equipment degradation

Equipment reconfigurability and equipment degradation endow scheduling in Matrix Manufacturing System (MMS) with enhanced physical significance and more complex scheduling constraints. Equipment reconfigurability introduces setup time as the third temporal parameter in addition to traditional machining duration and logistics time, while equipment degradation-induced dynamic variability and real-time characteristics of real processing time supersede theoretical processing durations. This study innovatively addresses equipment degradation in MMS by proposing a Scheduling Problem in MMS with Real Processing Time (SPMMS-RPT). First, a mathematical model for SPMMS-RPT is constructed, comprehensively considering the impacts of equipment degradation and predictive maintenance (PM) on temporal parameters. Subsequently, a Double Deep Q Network (DDQN)-based solving algorithm designed, featuring a degradation-aware state space, a discrete scheduling rule-based action space, and a reward function incorporating virtual feedback. Experimental validation demonstrates the effectiveness of the proposed model and algorithm. Results indicate that the DDQN-based SPMMS-RPT solution achieves superior makespan and shorter computation time compared to conventional methods.