Regelung flexibler Fertigungssysteme
Control of flexible manufacturing systems
It is well known, that the present techniques for production scheduling in flexible manufactoring systems (FMS) are not satisfactory (compare e.g. (1)). Some production planning systems (PPS) may be able to produce detailled schedules for the production of a fixed workload in a certain period, but in a dynamically changing environment (changes of due dates, introduction of new orders, breakdown of machines or lack of personell) such a schedule soon becomes obsolete. The enormous complexity of the task on the PPS level on the other hand prohibits a repetitive replanning on this level. Thus extending the task of detailled production scheduling within the PPS is not feasible. The only possible way to improve the situation is to decentralize the scheduling function and to assign it to the control units of small segments of the system, e.g. flexible work cells or certain sets of machines (workshops). This paper is concerned with the scheduling problem in subsystems of a large production pro cess. Even for such subsystems, the task to schedule a number of operations on the machines (which we assume to possess overlapping functionality with possibly different effectivity) is by far too complex to calculate the exact optimum in the time available in a dynamically changing environment with disturbances of all kinds present. The algorithms which we have developped have a two-layer structure where first a basic schedule is calculated using modified priority rules. This basic schedule is then improved by intelligent search strategies with the goal of just-in-time production with minimal residence time. The research reported here was supported by the Deutsche Forschungsgesellschaft (grant DFG En 152-3).