Agent-Based Intelligent Manufacturing System for the 21

Agent-Based Intelligent Manufacturing System
for the 21st Century
Bing Qiao, Jianying Zhu
Mechatronic Engineering Institute
Nanjing University of Aeronautics and Astronautics
Nanjing 210016, P. R. China
Email: [email protected]
Abstract: The intelligent manufacturing system is a solution to the problems of the 21st
century’s manufacturing industry. In this paper, the historical evolution of
manufacturing industry is outlined. According to the literature published during the last
decade, 34 modern manufacturing systems and production modes and 35 mathematical
methods for modern manufacturing system modeling are listed and summarized. An
intelligent object called Manufacturing Agent (MA) is proposed based on the
conceptual model of an agent that is being discussed and studied in Distributed
Artificial Intelligence (DAI) technology, and an agent-based intelligent manufacturing
system architecture is presented.
Key words: Agent, Intelligent manufacturing, Multi-agent system
1. Introduction
The coming century has been named as the age of knowledge economy. Manufacturing
industry is facing more and more challenges. Though different technical bottlenecks for
modern manufacturing have been or are being investigated, the real bottleneck is,
however, the lack of knowledge (Ann 1999). Efficient acquisition and utilization of
knowledge have been considered as the trump to win the competition in the knowledge
Fig. 1: The significance of the modern manufacturing system
The base of the knowledge economy is the knowledge industry that is supported by the
production equipment and modes, science and technology, management philosophy,
etc., produced by the modern manufacturing system and technology. Fig.1 briefly
illustrates the significance of the modern manufacturing system and its relationship to
the knowledge economy.
During the historical development of industry, human society has experienced three
industrial revolutions. Currently, the world is undergoing a Hi-Tech industrial revolution
with information technology as its main feature. As summarized in Table1, the
manufacturing industry of the next century will be characterized by intensively
concurrent engineering based on information technologies such as digitalization,
computer network, artificial intelligence and the like. The focus of enterprise production
is shifting from quantity/quality to service including quality, price, and after service.
People are paying more attention to the individualization, involvement and responsive
requirements of product. The management of the enterprise is being transformed from a
centralized and independent model to cooperative and coordinated management, in
which the creativity of people will be given full play. The 21st century’s manufacturing
industry will be green. It does not damage but preserves and beautifies the
environment. It not only utilizes but also saves and renews natural resources. It will
play an important role in the harmonious development of humankind and nature.
Table 1: Features of the three economic periods and their comparison
1st industrial revolution
(steam engine)
2nd industrial revolution
Large scale economy
Large scale production
Production flowline
Interchangeability Manufacturing
Unit technology
Big batch
Developed industrial
3rd industrial revolution
(atomic technology)
Hi-technology industrial
revolution (information
Speed economy
Integrated production
Automatic line
Integrated Manufacturing
Synthetic technology
Focus of enterprise
Relationship to natural
Relationship to
Utilization of natural
Damages environm-ent
Saving natural resource
Pays attention to environment
Knowledge economy
Intensive production
Concurrent engineering
Distributed networks
Intelligent manufacturing
Digitization & network
Intelligent technology
Individualization, involvement and rapid response
Services (quality, price,
after service, etc.)
Cooperative, full play
of personal creativity
Preserve & renew
natural resource
Preserves & beautifies
Industrial revolution
(marked technology)
Economy pattern
Main production mode
Main production factor
Production process
Production features
Key technology
Requirements to
The manufacturing systems for the next century should possess the features of agility,
intelligence, rapid response and favor high quality products, small batch sizes,
individualization requirements, consumer involvement, and environmental
consciousness (Koren et al. 1999, Lu et al. 1999, Kimura et al. 1998, Zhang et al.1997,
Wiendahl and Scholtissek 1994). The manufacturing systems characterized by the above
features have been named as intelligent manufacturing systems on which extensive
studies have been conducted.
2. Modern Manufacturing Systems and Production Modes
Over the past one or two decades, a great amount of effort has been invested by the
industrial world and academia to explore the advanced manufacturing system and
technology that will deal with manufacturing complexities originating from
globalization of markets, stringent competition environment and overload of constraints
on the manufacturing environment. And many manufacturing systems and production
modes have been proposed. According to the literature published by CIRP and other
manufacturing periodicals during the latest decade, nearly 34 modern manufacturing
systems and production modes have been proposed. A list is given below:
2. GT (Group Technology)
3. TQC (Total Quality Control)
4. MRP-I (Material Requisite Planning)
5. MRP-II (Manufacturing Resource Planning)
6. FMS (Flexible Manufacturing System)
7. JIT (Just In Time)
8. CAPP (Computer Aided Process Planning)
9. DFM, DFT, … … DFX
10. CIMS (Computer Integrated Manufacturing System)
11. CDP (Customer Driven Production)
12. BR (Business Reengineering)
13. CE (Concurrent Engineering)
14. EFM (Environment Friendly Manufacturing)
15. LP (Lean Production)
16. AM (Agile Manufacturing)
17. MAMS (Multi-Agent Manufacturing System)
18. VM (Virtual Manufacturing)
19. HMS (Holonic Manufacturing System)
20. BMS (Biological Manufacturing System)
21. LCE (Life Cycle Engineering)
22. GM (Green Manufacturing)
23. CM (Collaborative Manufacturing)
24. RM (Remote Manufacturing)
25. TPM (Total Production Maintenance)
26. OAMS (Open Architecture Manufacturing System)
27. IMS (Intelligent Manufacturing System)
28. FE (Fractal Enterprise)
29. P3 IS (Pre-Planned Product Improvement System)
30. ERP (Enterprise Resource Planning)
31. SOPS (Self-Organized Production System)
32. RMS (Reconfigurable Manufacturing Systems)
33. GM (Global Manufacturing)
34. NGMS (Next Generation Manufacturing System)
Some of the above manufacturing systems and production modes have been maturely
applied to practical manufacturing areas and have dramatically reduced the lead time of
products. Some are still undergoing investigation or conceptualization. Behind these
manufacturing systems and production modes lies the continuous pursuit of design
intelligence, manufacturing intelligence and management intelligence, or the enterprise
intelligence as a whole.
3. Intelligent Mathematics for Modern Manufacturing System Modeling
In order to model increasingly complex manufacturing systems, numerous mathematical
methods, especially the so-called non-conventional mathematical tools, have been
widely used. Some applications of these mathematical methods in the manufacturing
arena have achieved exciting progress. Teti(1997) presented a list of intelligent
computing methods for manufacturing systems and summarized their respective
application aspects in design, planning, production and system level activities. Zhu
reviewed 35 mathematical methods employed in manufacturing system modeling and
manufacturing process controlling:
Expert System (Knowledge Based System)
Pattern Recognition
Graph Theory
Similarity Theory
Optimization Theory
Game Theory
Time Series Analysis
Wavelet Analysis
Computer Vision
Natural Language Processing
Knowledge Representation
Heuristic Search
Constraint Based Search
Confidence Theory
Qualitative Reasoning
Reasoning Technologies
Machine Learning
Machine Proving
Multiple-Valued Logic
Fuzzy Logic
Artificial Neural Network
Petri Networks
Immune Networks
Genetic Algorithms
Artificial Life
Associative Memory
Blackboard Architecture
Multi-Agent Systems
Non-Classical Control Theory
Operations Research
System Engineering
Simulated Annealing
Combinatorial Mathematics
34. Fractal Theory
35. Chaos Theory
In the practical applications, these methods are usually combined with each other to deal
with planning, design, process control, and system integration, etc. Research examples
can be frequently found in technical literature. The wide application of these intelligent
mathematical methods or their combinations in manufacturing will definitely promote
the development of manufacturing system modeling and provide new solutions to the
complexities manufacturing.
4. Agent and Agent-Based Manufacturing
Currently, the open architecture Multi-Agent System (MAS) has become the main
research direction of DAI. Agents are now being widely discussed by researchers in
mainstream computer science (Wooldridge and Jennings 1995). Though there is not a
universally accepted definition of agent, it does not prevent people from expanding the
application space of agent technology. In practical application an agent is usually
considered as a self-directed program object which has its own value system and the
means to solve certain subtasks independently and then communicate its solution to a
greater problem solving process, either on its own initiative or on request from other
agents (Sikora and Shaw 1997, Wiendahl and Ahrens 1997). Fig.2 is a conceptual
model of an agent, which comprises 4 components:
Knowledge processor, a knowledge base system that stores and processes the
necessary knowledge for an agent to play the role the agent society has designed for
Perception, a channel for an agent to receive information from the external world.
Effector, an interface for an agent to modify or influence the state of the agent
Communication, a mechanism for an agent to exchange views with other members
in the agent society.
Objectives, a list of roles for an agent to play.
Fig. 2: Conceptual structure of an agent
The MAS is an open and distributed system that is formed by a group of agents
combined with each other through a network for cooperatively solving a common
problem. The MAS has been applied in many areas. The modern manufacturing system,
which is a highly decentralized system, is one of the typical representatives of MAS
application. A great deal of research effort has been devoted to agent-based
manufacturing. Some leading manufacturers and government agencies in the USA are
claiming that agent-based manufacturing is the future for US manufacturing and agent
technology is the fundamental technology for implementing the agile manufacturing
vision (Baker 1998).
There is multifold advantage in using an agent-based approach for manufacturing. First,
manufacturing information is stored and processed in a distributed manner as opposed
to being stored in one large program. Second, it makes the incremental improvement of
the manufacturing system possible through the learning and cooperation in agents.
Third, it provides a promising method for enterprise integration.
5. MA and Agent-Based Intelligent Manufacturing System Architecture
The application of agent technology in many manufacturing aspects have been
investigated, however, a general agent model suitable for manufacturing applications
still remains to be constructed.
5.1 Intelligent Manufacturing Agent
In manufacturing, an agent represents an object with certain intelligence, which is either
a physical object such as a worker, a machine tool, and the like or a logical object such
as an order, a task, etc. In order to establish an agent-based manufacturing system, the
model of the Manufacturing Agent (MA) must first be defined.
Fig.3 is the conceptual model of MA. Procedure repository, inference mechanism,
knowledge base, and perception processor constitute the knowledge processing unit for
the MA. The self-learning, self-organizing, and self-adapting are the optimizers of the
MA. The simulation expert provides behavior evaluations for the MA. Working
memory stores the temporary data produced by the other parts of the MA. The
communication mechanism manages the message exchanging tasks of the MA with the
other members of the system. The coordinator coordinates the internal functions of the
MA, meanwhile it receives the coordination requests from other MAs.
Fig. 3: Model of MA
The main features of MA are:
Self similarity: Every MA has a similar structure.
Self learning: The MA has coordinated learning ability that contributes to an
emergent improvement of the whole system behavior.
Self adaptation: The MA has the ability to adjust its behavior intention according
to different rules.
Self organization: The MA can organize or configure its internal procedures to deal
with different tasks.
Self maintenance: The MA can monitor the state of itself and keep itself in the
most efficient condition.
5.2 Agent-Base Intelligent Manufacturing System Architecture
There are three commonly studied multi-agent systems: functional, blackboard, and
heterarchical architectures. The heterarchical multi-agent architecture shows a greater
promise in the application of manufacturing for its self-configuration, scalability, fault
tolerance, reduced complexity, increased flexibility, reduced costs, massive parallelism,
Internet compatibility, and virtual enterprise formation (Baker 1998). In a heterarchical
multi-agent manufacturing system, MAs communicate as peers and there are no fixed
master/slave relationships. The objectives of the system are achieved through
cooperation of MAs. According to the general structure of the manufacturing
enterprises, we propose the agent-based manufacturing system architecture depicted by
Fig. 4: Agent-Based Architecture for Manufacturing Enterprise
In Fig.4, the whole operation logic of a manufacturing enterprise is divided into 4 parts:
central part, management, planning, and production. Each part consists of a group of
MAs. The central MAs are designed for the leadership of the enterprise to coordinate
and control the operation of the whole enterprise. The management MAs execute the
management functions of the enterprise. The planning MAs arrange and allocate the
available and potential resources to support the profit seeking objectives of the
enterprise. And the production MAs conduct shop floor control and management of the
manufacturing. All these MAs are connected to the distributed enterprise network or
Intranet. This MA network is heterarchical in nature. Each MA has its own behavior
objectives and local knowledge about the manufacturing process. In this agent-based
manufacturing system, humans communicate with MAs through a special MA, for
example a Personal Assistant (PA) ( Pan and Tenenbaum 1991).
6. Concluding Remarks
This paper has outlined the historical evolution of manufacturing and enumerated the
main features of 21st century manufacturing. 34 modern manufacturing systems and
production modes and 35 mathematical methods for modern manufacturing system
modeling are listed and summarized. An intelligent object called Manufacturing Agent
(MA) is proposed based on the conceptual model of the software agent, and an agentbased intelligent manufacturing system architecture is presented.
In a multi-agent system, the objective of the system is achieved through the cooperation
of agents. Each agent has only local knowledge about the whole agent environment and
has only local behavior ability, however, the cooperative and coordinated interaction of
multiple agents usually results in a macro efficient system behavior that is robust,
flexible and error-tolerant. This kind of emergent intelligent behavior can be found in
many biological systems such as the ant colony, bird flocks, fish schools, and wolf
packs (Baker 1998). What makes agent-based manufacturing attractive is just this
emergent behavior intelligence.
To successfully employ agent-based technology in manufacturing, some fundamental
problems should be researched. They include the coordination mechanism of MAS, the
communication protocol of agent, behavior intelligence of agent, human involvement in
agent system, the formalized modeling of agent-base manufacturing, etc. It is
completely believable that the agent will have an impact on future manufacturing.
7. Acknowledgement
This research is supported by the Chinese National Science Foundation under grant
contract no. 59990470.
Ann K (1999) Teams-What’s Next?: From Fragmentation and Consciousness to Responsiveness by
Competence Management for Modular Manufacturing Learning, Annals of the CIRP, Vol. 48/2
Baker D (1998) A Survey of Factory Control Algorithms That Can Be Implemented in a Multi-Agent
Heterarchy: Dispatching, Scheduling, and Pulling, Journal of Manufacturing Systems, Vol.17/4
Kimura F, Lipson H, Shpitalni M (1998) Engineering Environments in the Information Age: Research
Challenges and Opportunities, Annals of the CIRP, Vol. 47/1
Koren Y, Heisel U, Jovane F, Moriwaki T, Pritschow G, Ulsoy G, and Brussel HV (1999) Reconfigurable
Manufacturing Systems, Annals of the CIRP, Vol. 48/2
Lu SC-Y, Shpitalni M, Gadh R (1999) Virtual and Augmented Reality Technologies for Product
Realization, Annals of the CIRP, Vol. 48/2
Monostori L, Markus A, Brussel HV, Westkampfer E (1996) Machine Learning Approaches to
Manufacturing, Annals of the CIRP, Vol. 45/2
Pan JYC, Tenenbaum JM (1991) An Intelligent Agent Framework for Enterprise Integration, IEEE Trans.
Syst., Man and Cyber., Vol. 21/6
Sikora R, Shaw MJ (1997) Coordination Mechanisms for Multi-Agent Manufacturing Systems:
Applications to Integrated Manufacturing Scheduling, IEEE Trans. Syst., Man and Cyber., Vol. 44/2
Teti R, Kumara SR (1997), Intelligent Computing Methods for Manufacturing Systems, Annals of the
CIRP, Vol. 46/2
Wiendahl H-P, Ahrens V (1997) Agent-Based Control of Self-Organized Production Systems, Annals of
the CIRP, Vol. 46/1
Wiendahl H-P, Scholtissek P (1994) Management and Control of Complexity in Manufacturing, Annals
of the CIRP, Vol. 43/1
Wooldridge M, Jennings NR (1995) Agent Theories, Architectures, and Languages, Intelligent Agents,
Zhang HC, Kuo TC, Lu H (1997) Environmentally Conscious Design and Manufacturing: A State-of-theArt Survey, Journal of Manufacturing Systems, Vol. 16/5