Extensible multi-agent system for optimal design of complex systems using analytical target cascading

Abstract

Analytical target cascading (ATC) has emerged as an integrated approach and methodology for optimal design of complex hierarchical systems with good convergence. It facilitates collaborative design problem solving and enables distributed computing to extend both the capability and capacity. ATC is very powerful yet requiring comprehensive understanding and efforts in setting up specific projects. This paper presents a generic and extensible information infrastructure called atcPortal in the form of a web portal where the ATC analyst defines the ATC problem using a special-purpose XML-based language called atcXML, conducts the ATC analysis, and obtains the analytical results. atcPortal not only reduces the overheads for analysts to set up the projects for practical applications, but also allows the researchers to experiment and extend the ATC methods under different system structures and/or different ATC strategies.