Multi-agent Systems for Planning and Controlling of Industrial Manufacturing Processes 2002-01-2124

This paper deals with the conception and modelling of overlapping business processes which base upon software agents. The regarded processes arise in connection with the research project CoagenS, in which learnable production networks of series fabrication will be conceived and realized. The R&D project CoagenS is supported by the BMBF1 within its programme: “Research for the production of tomorrow.”

By the use of a software agent the in-house purchase processing of the Supply Chain Management is arranged more efficiently. The benefit of software agents results from its special flexibility and the capability to learn, which allows solving a problem without a central coordinator. By means of attributes such as autonomy, reactivity, sociability, and productivity software agents dispose of the possibility to plan and to track a multitude of complex functions independently. The different skills of the software agents are coming along with certain benefits like the existence of a de-central organisation structure. Thus software agents are predestined for the planning as well as for the controlling of industrial fabrication processes. With the help of the Multi-Agent Systems the communication structure between associated suppliers and buyers will be improved. Consequently, time consuming manual and therefore error-prone data input disappears. The implementation of agent systems takes place in an existing PPS-system. Thus synergies evolve, which optimise the process of the ordering completion and simultaneously overcome asymmetries of information between the associated partners. Moreover all participants of the value added chain -supplier-carrier-customer- profit from these positive effects.