Hierarchical Control for Continuous Flow Simulation of Manufacturing Systems

This paper aims at developing hierarchical control architecture for continuous-flow simulation in order to regulate production in multiple-part-type flow shops. The approach uses a continuous-flow approximation to model the discrete flow of parts in a manufacturing system. The control strategy specifies how to allocate limited system capacity among all the part types to follow the solution of the continuous-flow model as closely as possible. The control objectives are to keep the actual production close to the demand, while maintaining the average work-in-process inventory and lead time to satisfactory levels. The control architecture is hierarchical. It allows combining different decisions into a unified model. This architecture is composed of basic-level distributed fuzzy logic controllers supervised by a higher level decision-maker. At the bottom level of the hierarchy, individual decisions are based on local information and expert’s knowledge to adjust the machine’s processing rate. At the top level of the hierarchy, the supervisory controller combines both local information and global performance indicators in order to tune the action of the lower distributed fuzzy controllers. The global performance indicator used in the supervisory level evolves in a tolerance interval defined by the normal operating conditions of the process. When a performance indicator value is outside of the predefined tolerance interval, an abnormal behaviour occurs. In this case, the supervisor allocates the production capacity or reduces the production throughput according to the aggregated global performance indicators. Simulation results through continuous-flow simulator of production network are presented to illustrate the feasibility of the proposed approach.