Agent-oriented Analysis, Modeling And Integration Of Process Operation Systems

With the development of information technology and economic globalization,process industries tend to be more networked and digitalized. More efforts arefocused on how to integrate production, management and market for optimizedprocess industries and supply-chain. Combining with the system science, computerand network technologies, an agent-oriented analysis (AOA) is proposed for theintegrated process operation system in this paper. By AOA, the process operationsystem is divided into two parts: agents and objects, associated by message. Objectsare the equipments, units and process, called process models. Agents are theengineering tasks on the actual process, also called operation task models. Theintegration of the process operation systems is accomplished by hardware integration,information integration and task integration. The hardware integration provides thecomputer network platform for the requirements of source sharing and co-calculation.The information integration means standardized process data, process models andprocess knowledge, which is understand and used by different domain engineersconventionally. At the same time, the information is transferred easily between thediverse application software provided by different vendors. Furthermore, the differentcomputer aided process operations (CAPOs) are cooperated by the task integration tooptimize the whole operation system.Structural method and object-oriented method are used widely in the processsystem analysis. A new analysis method, agent-oriented analysis (AOA), is proposedafter investigation of the two methods mentioned above. AOA divides the processsystem into two parts, objects and agents. Objects are modeled by first principlemethod, artificial neural network, statistical regression and fuzzy logic technique.Recently, many researchers combine different modeling methods to remedy theshortage of single model. Consideration of lots of operation data and complexnon-linear characters of real production, a PCA based output integration recurrentneural network is presented for the dynamic chemical process. The proposedmodeling method is applied successfully in the product flow rate prediction. For theagents in the process operation system, a π-calculus is used to depict the rationalcharacters of agents. To implement the agent, the standard software structure isproposed, including the internal database, the executor, the functional modules andthe communication module. In addition, the agent communication platform,...