Dependable reconfiguration of mobile manufacturing systems

This paper presents a dependable reconfiguration mechanism for mobile agent-based manufacturing systems. A concept is presented which long-term goal is the integration of unknown components and the functionality improvement of mobile manufacturing systems. Based on a four layered agent architecture, the system reconfiguration is performed in a concurrent system state to maintain system productivity during reconfiguration. The integration of unknown components is activated by a wrapping mechanism. Functionality descriptions of inserted components are supplied by local ontologies. After a conceptual matching between the local ontology and the system's global ontology, all newly created ontological concepts are classified to refine semantic integration. Finally a real-world clearance decides whether to use the integrated components completely, in parts or not at all. Likewise the detection of an internal lack of functionality causes a system reconfiguration. After evaluating the lack of functionality, related concepts can be figured out by the help of the used ontologies. By applying the layered learning mechanism the combination of known concepts leads to new system functionalities within hardware limits. The resulting functionalities have to be integrated the same way as functionalities of newly inserted components.