Production scheduling for an auto-component manufacturing plant using model predictive control

A two-level approach for production scheduling is suggested. The first level uses Model Predictive Control (MPC) to allocate resource capacities, while the second level does the sequencing. A plant-wide linear optimization problem is formulated for the first level. Next, smaller independent clusters within the plant are identified and production is sequenced for these. This approach has the advantages of centralized allocation and being linear can solve large size problems. The sequencer takes care of local effects, that otherwise would make the MPC problem complex. The rolling horizon feature of MPC helps working under uncertain conditions.