Agent-Based Decentralised Process Planning and Evolutionary Change Propagation

Abstract

The increasing dynamics of manufacturing systems promote the utilization of decentral system architectures, such as Multi-agent Systems (MAS). These are especially applicable for dynamic systems that exhibit structural and behavioural changes during runtime. This paper develops a MAS that is capable of decentralised process planning that does not require any a priori definition of compatibility or interfaces between machines. Furthermore, the paper outlines how the producibility graph that is derived by the agents during process planning is updated and shared between the agents to keep the knowledge bases of the agents updated at all times.