Planning on Discrete Event Systems using parallelism maximization

摘要

This work deals with the production planning problem in Discrete Event Systems, using the Supervisory Control Theory to establish the search space and developing two heuristics based on the maximization of the parallelism to find sequences that minimize makespan. The role of the Supervisory Control Theory is to provide the set of all safe production sequences, given by the closed-loop behavior. The two heuristics are based on the idea that controllable events should be executed as soon as they are allowed (maximizing parallelism) but only temporally feasible candidates are evaluated. The proposed methodology delivers solutions that are robust to uncertainties in the model parameters that represent time intervals required for plant operations. Although heuristic procedures are not guaranteed to reach exact optimal solutions in general, we present a case study where it happens for all batch sizes. The efficiency in terms of computation time is also illustrated by the case study.