Using multi-agent system to build a smart workshop for achieving self-organized production

At present, the manufacturing control system is expected to be more flexible and agile. Simultaneously, using a multi-agent manufacturing system has been considered as a fundamental approach to solving the problems of dynamics and uncertainty in the part processing environment. However, the difficulty in deployment and lack of global performance guarantee measures limit the use of the multi-agent system in actual manufacturing sites. For these issues, this paper proposes a technology of an “agent computing node” to improve the realization of a multi-agent control system for smart workshops and achieve self-organized control of production. Moreover, a supervisor layer for performance monitoring and rules adjustment is proposed and investigated. During production, the self-organizing rules of agent computing nodes in heterarchical control architecture are optimized in real-time through this layer so that the agent computing nodes are guided to make decisions that are beneficial to global performance.