ON-LINE SCHEDULE OPTIMIZATION FOR MIXED-BATCH/CONTINUOUS PLANTS

Abstract A reactive, on-line scheduler has been developed. It controls path and sequencing choices, start times of phases, batch sizes and flows, in order to optimize production. Constraints may involve state variables, which makes it especially useful for mixed-batch/continuous plants. In reactive mode it acts as a model-based predictive controller and deals on-line with disturbances and failures. The reactiveness of the scheduler is achieved by using a combination of: non-monotonic reasoning, event driven reasoning, hierarchical control loops, progressive reasoning, anticipatory reasoning, parallelism and heuristics. Due to the reactive behaviour of the scheduler, an increase of production compared to non-reactive schedules is obtained.