Researches On Multi-Agent System Based Industrial Process Monitoring Technology

Industrial process monitoring is a nessesary enabler for normal operation of industrial processes. Operators'inapproprate judgments or actions possible may be caused by fatigue, inexperience and other personal factors, therein resulting in production inefficiencies, waste of resources, substandard products or even accidents.Inspired by multi-agent system, this paper proposes a novel idea to combine a variety of process monitoring techniques, which offers an opertunity for them to contribute advantages, share resources and perform tasks.First, based on the analysis of the industry development of intelligent monitoring system, a multi-agent schem is proposed to design industrial mornitoring system.Given a deep insight into the existing process mornitoring methods such as PCA technique, control system performance assessment and parameter trend analysis, the corresponding agents are utilized to encapsulate them. Taking advantages of multi-agent system, an archituecture of industrial process monitoring system based on multi-agent system is presented.In this paper, various specifications of agent technology in FIPA, Java, an object-oriented programming language as well as a Java-based agent development platform-JADE are employed to realize the multi-agent industrial monitoring system.Finally, the proposed industrial process monitoring system is applied to Tennessee-Eastman Process, performing experimental simulations. The real-time data generated from TEP are stored in the database, which are shared by each agent of industrial monitoring system to best perform the monitoring tasks.The studies show that the multi-agent system of industrial process monitoring system is able to well monitor industrial process, bringing in individual merits of specific monitoring techniques, as well as compensating against shortages, sharing resources and saving costs. A good application prospect could be expected for this approach.