The Application of Multi-Agent Systems to the Design of an Intelligent Geometry Compressor

Abstract

AimsThe specific aim of this research is to demonstrate how a multi-agent systems (MAS) approach may be applied to the design of an axial flow compressor in order to optimise performance and considerably enlarge the useful operating envelope of the machine.This application was proposed by Rzevski (1998) as one of a number of potential engineering applications for multi-agent systems and termed an 'Intelligent Geometry Compressor' (IGC).The underlying hypothesis of the research is that multi-agent systems offer a novel, and potentially effective, alternative approach to the design of engineering systems and machines which are required to operate in a largely autonomous and intelligent manner. BackgroundMachines provide solutions to specific engineering problems found in a wide range of industrial applications.Here, the meaning of the term machine is a device or piece of equipment, comprising mechanical, electronic and software components, which carries out some significant, purposeful function typically involving the transmission or conversion of energy.Examples of machines in industry include robotic manipulators, automatic guided vehicles, machine tools, assembly and test equipment, power plant such as turbines and compressors, material handling and process control equipment.A general consequence of the continuing growth in capability per unit cost of computer, communications and information technology is the potential for enhancing machine functionality and control.This, in turn, offers opportunity to extend and optimise the performance of existing machines or to design totally new machines for new applications which hitherto would have been impractical or prohibitively expensive.Technical advantages sought include increased operating range, greater energy efficiency, higher precision, faster response, greater durability and reduced physical size whilst economic motivation lies in reduced lifetime costs of machines and increased productivity.However, this enlargement of machine performance demands a more complex, adaptable and autonomous behaviour than that previously achieved.Machines that exhibit this capability are invariably described as 'intelligent'.