Reactive scheduling using timed automata models and integration with sequential control logic

The increasing demand for customized products and the pressure to produce them on time at a reasonable price call for an efficient and economic scheduling of the production processes. One crucial characteristic that makes the scheduling challenging is the presence of uncertainty on the operational level. Highly dynamic production environments demand for optimization techniques that can compute new decisions on the operational levels within very limited computation time in order to react to unforeseen events. In current industrial practice, the scheduling in dynamic production environments is either done manually by plant managers or by heuristics. The quality of those schedules is often rather poor since process related restrictions can hardly be integrated efficiently. The solution approach to use the framework of timed automata combines the advantage of modular graphical modeling with a high performance graph-search algorithm that finds good and feasible solutions in a very short time. This thesis proposes a reactive scheduling framework based on embedding the timed automata-based approach for nominal schedule calculation into a moving window approach to handle the different uncertainties of dynamic production environments for the scheduling layer. The reactive scheduling framework is successfully applied to different case studies; detailed results are discussed. Furthermore, it is connected to low-level sequential control logic to form a hierarchical integrated control system which is able to detect and handle delays and failures in process operations on the different control levels; the approach leads to an as-fast-as-possible execution strategy which robustly handles different kinds of uncertainties in the operation.