抄録
Scheduling of manufacturing process is one of the significant problems in FMSs. Requirements for the FMSs' scheduling are stated as follows: 1) It is easy to formulate a problem. 2) Semi-optimal solution can be found within short time. 3) It is possible to search the action in a reactive manner (action search depending only on a current state of the system). The property 3) is required to adapt to unexpected changes in production requests and/or machine breakdowns. However, no scheduling method which meets these requirements, has been proposed yet. This paper presents a new scheduling method based on problem formulation with Petri net and reactive graph search. Firstly, a new modeling method based on Petri net is proposed, in which physical behavior in FMSs and processing information of job are described independently. Secondly, a new reactive scheduling based on RTA* algorithm and supervisors (dispatching rules) is proposed. In the proposed framework, machine and AGV allocations can be scheduled simultaneously. Finally, usefulness of the proposed algorithm is confirmed through numerical experiments.
