An Agent-based Inference System for FMS Dispatching

Abstract

In the era of mass customization and ever-changing market, it is important for enterprises to develop the dynamic and quick-response planning techniques in their flexible manufacturing systems (FMSs). This research paper presents an agent-based inference system, called Intelligent Decision Inference System (IDIS), to support existing FMS for production dispatching. When system integration is required between two separate operating systems, there are many issues to be taken into consideration, including system communication method, data flow design, and application program control. In the past, the most universal method was to integrate databases with two or more systems. However, this method had a main weakness of “data confusion and inconsistent”. Therefore, this research combines the agent, JESS, and related IT technologies to systematically integrate the IDIS with FMS. The agents can easy communicate and deliver the messages, data, or information among the different systems and automatically scan the signals of the programmable logic controller (PLC)/sensors periodically. The method applies web service techniques as a standard information exchange mechanism and the agent communication language (ACL) of the Foundation for Intelligent Physical Agents (FIPA) for multi-agents to communicate/coordinate with each other. The agents, implemented in Java Agent Development Framework (JADE), are designed to decide whether a pallet should be moved in the shop floor of the FMS. Finally, one explicit example about how the cooperative multi-agent supporting the IDIS and FMS integration is presented. The unique contribution of this research is in the integration of expert systems and an agent framework to apply in real-time shop floor control such as FMS.