Predictive control for operation and maintenance in smart manufacturing systems with multiple operating modes

Owing to dynamic changes in the demands of production in different stages of manufacturing, smart manufacturing systems require the adaptive selection of the operating modes of each device and the allocation of their workloads. In this case, the varying workloads of devices result in different levels of degradation and health status. Load balancing is a crucial factor that influences both operating and maintenance planning. However, there is a lack of research on predictive control for operation and maintenance (O&M) in smart manufacturing systems with multiple operational modes. In this paper, driven by task and health information, a proactive O&M decision framework is proposed to dynamically regulate the health status of smart manufacturing systems. We formulate a multi-stage integer linear programming (MSILP) model with constraints based on generalized disjunctive programming (GDP) to enable optimal decision-making regarding operating modes and maintenance planning, with the aim of minimizing the cost of maintenance while ensuring task reliability by considering the component dependencies, resource maintenance, and degradation under different operating modes. Furthermore, we introduce a matheuristics algorithm for rolling horizon control to accommodate customized market demands. Finally, we analyze the subsea production system as an illustrative example to examine how task requirements and rolling horizons influence smart manufacturing systems in a customer-oriented market environment.