wilamowski-b-m-irwin-j-d-industrial-communication-systems-2011

16-6 Industrial Communication Systems
the graphical interface of two tools provided by JADE, namely the Remote Management and the
Sniffer agents. The last one is a debugging tool that allows tracking messages exchanged in a JADE
environment using a notation similar to Unified Modeling Language (UML) sequence diagrams.
Jadex, as an extension of Jade, is one of the few examples of a platform that also provides support for
the reasoning capabilities of agents (cf. http://jadex.informatik.uni-hamburg.de/). Its reasoning engine
implements the Belief-Desire-Intention model.
16.3 agents and Multi-Agent Systems in Industry
Classical centralized solutions perform well in static industrial environments in which neither the physical
layout of the plant nor the product portfolio varies. Thus, their rigid control architecture does not
have to be modified, for example, to deal with new unknown situations. Modern industries, however,
pose different challenges to which the centralized approach cannot respond. The main ones are ever
changing, unpredictable order flows, which, in order to be fulfilled, need a dynamic shop floor. Both
require more flexible approaches that can be naturally provided by MAS. In fact, MAS are designed
as a group of autonomous interacting units that pursue (directly or indirectly) a common goal. This is
exactly what modern manufacturing systems need.
Altogether, a MAS-based manufacturing control has to satisfy the following characteristics:
• Reliability and robustness: The system must respond properly to disturbances.
• Adaptability to changes: A new physical layout, changes on product designs, extension of the
product portfolio, demand changes, and other novelties should not prevent the MAS from carrying
out its task.
• Scalability: It should offer an easy coupling of new parts or systems and should easily allow integrating
new automation devices or other information systems (e.g., the scheduling system with
the plan execution, the supply chain manager, and so on).
• Easy upgradeability to new technologies: In case of a new version of platforms, protocols, or brand
new technologies, the system should offer an easy upgrade procedure.
• Real-time response: It should be able to react in real time to disturbances or changes.
• Portability and platform independence: The MAS should not depend on a certain platform but be
able to work under different platforms.
16.4 application Areas
Responding to the modern trends in automation such as mass customization and dynamic configuration,
over the last decade distributed systems found its place in several industrial areas, like manufacturing
control, air traffic control, production planning, logistics, supply chain management, or traffic
and transportation. Some of them are combined under the umbrella of the plant automation pyramid:
enterprise resource planning (ERP), manufacturing execution systems (MES), and field control. Due to
the rigid nature of the ERP systems as well as machine-oriented field levels, distributed MESs are the
ones where industrial agents are most widely used.
One of the best-known architectures for distributed manufacturing execution is the Product–
Resource–Order–Staff Architecture (PROSA). It is a holonic reference architecture based on three types
of basic holons: product, order, and resource (cf. [24]). Additionally, typically to all distributed architectures,
the necessity to observe and control the shop floor and to provide an interface to the enterprise
level systems evolved into the creation of a fourth special type of components. In PROSA, it is the
staff holon, which assists and supervises the basic holons. Another architecture based on the holonic
manufacturing systems (HMS) concept—MetaMorph and its follow-up MetaMorph II—uses a mediator
centric federation architecture for intelligent manufacturing (cf. [37]). It uses a team mediator that
provides communication mechanisms to different systems and components and takes over some of the
© 2011 by Taylor and Francis Group, LLC