DOI 10.1007/s10845-011-0606-5
A multi-agent collaborative maintenance platform applying game
theory negotiation strategies
Amy J. C. Trappey · Charles V. Trappey · Wei-Chun Ni
Received: 14 December 2009 / Accepted: 6 March 2011 / Published online: 23 November 2011 © Springer Science+Business Media, LLC 2011
Abstract Engineering asset management (EAM) is a broad discipline with distributed functions and services. When engineering assets are capital intensive, management requires specialized expertise for diagnosis, prognosis, maintenance and repairs. The current practice of EAM relies on self main-tained experiential rules with coordinated collaboration and outsourcing for maintenance and repairs. In order to enhance the life long asset value and efficiency (from the stakeholder’s viewpoint) and after sales service quality (from the asset provider’s viewpoint), this research proposes a collaborative maintenance platform that integrates real time data collection with diagnostic and prognostic expertise. The collaborative system combines and delivers services among asset oper-ation sites (the maintenance demanders), the service center (the intermediary coordinator), the system providers, the first tier maintenance collaborators, and the second and lower tier parts suppliers. Multi-agent system technology is used to integrate different systems and databases. Agents with autonomy and authority work to assist service providers and coordinate communications, negotiations, and maintenance decision support. Finally, game theory is used to design the decision models for strategic, tactical, and operational deci-sion making during collaborative maintenance practices.
A. J. C. Trappey· W. -C. Ni
Department of Industrial Engineering and Engineering Management, National Tsing Hwa University, Hsinchu, Taiwan C. V. Trappey (
B
)Department of Management Science, National Chiao Tung University, Hsinchu, Taiwan
e-mail: trappey@faculty.nctu.edu.tw
Keywords Engineering asset management· Maintenance decision· Intelligent multi-agent system · Game theory · Negotiation mechanisms
Introduction
Engineering asset management (EAM) includes stages of continuous lifecycle management including the design, con-struction, use, maintenance, repair, disposal, and recycling of assets (CIEAM 2008). Enterprises often focus on opera-tional aspects of engineering assets, since these assets may fail due to lack of proper maintenance which in turn causes production or service delays. Therefore, poor EAM incurs financial loss of the asset itself as well as production delays and loss of services.
Many academic studies review asset diagnosis based on asset condition monitoring and life cycle prediction (Majidian and Saidi 2007). Engineering assets and equip-ment are shutdown for repair when maintenance person-nel become aware of a malfunction. Repair work is often reactive to a cause rather than based on real time condi-tion monitoring, prediccondi-tion, diagnostics and pre-scheduled maintenance (Yao et al. 2005). Further, traditional asset management approaches are not well suited for predict-ing equipment failure (Sun et al. 2006). Therefore, many enterprises cooperate with research institutes to improve diagnostic and forecasting skills. Research in these areas has helped enterprises solve complicated engineering asset problems and reduce the cost of equipment malfunctions. The demand for high level maintenance services has led to the creation of new services offered by system providers offering specialized maintenance consulting. These service companies negotiate costs and repair specifications with equipment holders and maintenance chain participants and
fulfill maintenance requests according to a time schedule. Consequently, maintenance scheduling, coordination, and communication are critical for the success of the business model and ongoing sales relationship. This research studies the problem of how to effectively coordinate the operations among different service providers and asset holders. A gen-eral EAM decision model is developed using game theory. A case is used to describe the relationships between a network of power transmission and distribution equipment providers and to demonstrate the EAM methodology and information technology (IT) solutions.
The proposed preventive maintenance decision support system is applied to the EAM of a company managing large electrical transformers. Multi-agent system (MAS) tech-nology enables autonomous, social, goal-oriented, reactive, rational choice, self-learning, and better enables information exchange, communication, and decision support (Hossack et al. 2003;McArthur et al. 2005). Agent behavioral mod-eling is used to design and develop negotiation mechanisms which strengthen cooperation across the maintenance chain. The research implements an intelligent collaborative pro-cess system that supports maintenance chain monitoring and repair as well as negotiation between members.
Literature review
The literature covering EAM, agent technology for asset management, and game theory are reviewed. The directions for collaborative and intelligent EAM are identified. More-over, game theory and negotiation mechanisms are studied to select appropriate models for integrated maintenance chain applications.
Engineering asset management
Integrated Engineering Asset Management (IEAM) is a con-tinuous process covering the asset life cycle from the concep-tual design, construction, manufacturing, operational usage, maintenance, rehabilitation, disposal, and recycling (CIEAM 2008). Many engineering assets, such as water pipeline sys-tems and electrical transmission syssys-tems, are large scale facil-ities which directly affect the quality of life. A key research issue concerns ways to extend the operational life of assets and the means to effectively and efficiently arrange routine, emergent, and preventive maintenance based on the assets condition.Li et al.(2005) demonstrate in their field research that maintenance costs range from 15% for manufacturing companies to 40% for iron and steel companies, percentages which translate to almost $200 billion US dollars every year. Bangemann et al.(2006) and Han and Yang(2006) noted that successful predictive maintenance requires the integra-tion of condiintegra-tion monitoring, prognosis, diagnosis expertise,
maintenance schedule coordination, human resource alloca-tion, and maintenance part preparation. They also note that MAS are suitable for the integration of these elements that are widely distributed among organizations.
An EAM framework focuses on real time condition mon-itoring techniques to provide continuous data flows for rule based systems to derive timely maintenance schedules. Bretthauer et al. (1998) proposed an integrated mainte-nance scheduling (IMS) system which separates maintemainte-nance scheduling into device specific and system specific levels. The device specific level collects maintenance requests, ana-lyzes potential maintenance actions and maintenance times from individual machines. At the system specific level, the intelligent condition estimation (ICE) module determines feasible maintenance schedules. Fuzzy logic is used to derive the final maintenance schedule to enhance maintenance teams’ decision making capabilities. Further, Yang et al. (2008) use another approach, namely genetic algorithms, to generate cost-effective maintenance schedules that predict machine degradation.
Fu et al. (2004) recognize that traditional maintenance concepts and scheduling systems are outmoded for dynamic engineering environments. They identified monitoring and forecasting, diagnosis and prognosis, and maintenance deci-sion making as the key elements in maintenance manage-ment systems. In their research, these elemanage-ments are integrated with a trained artificial neural network (ANN) and a predic-tion model to build a maintenance management system for a dynamic environment.
Agent technology for integrated asset management
The linkage and communication efficiency among mainte-nance chain participants ensures flexible and accurate asset management decision making. Scholars have applied multi-agent technology for system integration and heterogeneous decision making. Li et al. (2005) proposed a basic agent framework for maintenance coordination using various agent roles, such as monitoring, diagnostic, prognostic and main-tenance decision making agents. Several frameworks of MAS for collaborative e-maintenance applications were also designed byIung(2003);Han and Yang(2006), andHsiao et al.(2008). Further,Jenab and Zolfaghari(2008) developed an analytical model to measure the performance of the virtual collaborative e-maintenance architecture.
The current research trend in the agent based EAM focuses on integrating goals, objectives and constraints between maintenance chain participants (i.e., asset owners, operators, equipment providers and maintenance parts suppliers). The functions and resources required for preventive maintenance (condition monitoring, data transformation and transmission, asset condition database, diagnosis and prognosis expertise,
Table 1 Major categories of game theory
Static Dynamic
Complete information Static games of complete information: Dynamic games of complete information:
Nash equilibrium Subgame Perfect Nash Equilibrium
Incomplete information Static games of incomplete information: Dynamic games of incomplete information: Bayesian Nash equilibrium Perfect Bayesian Nash equilibrium
human resources allocation and maintenance parts prepara-tion) must be considered when decisions are made among different organizations. For our proposed collaborative main-tenance platform, participants from different organizations in the maintenance chain are inter-connected by agents to increase communication and negotiation efficiency to predict equipment malfunctions without geographic constraints. Game theory and negotiation mechanisms
In 1940, game theory was proposed to model the way humans think as a practical solution for problems in economics and other social sciences (Nagarajan and Sosic 2008). Game the-ory is used to manage conflicts and help decision makers develop strategies that yield benefits and encourage coopera-tion. Since conflict and cooperation in and of itself may influ-ence decision making, a decision maker chooses a strategy by studying their own interests, conducting dependence anal-ysis, and predicting others’ behaviors. Consequently, game theory provides mathematical models that generate outcomes when there are conflicting interests among participants and relies on the competitive and dynamic features of informa-tion (Kreps 1990). Game theory considers the game state (static or dynamic) and information completeness (complete or incomplete). Table1 shows that there are four types of games including Nash Equilibrium, Subgame Perfect Nash Equilibrium, Bayesian Nash Equilibrium and Perfect Bayes-ian Nash Equilibrium.
Nash equilibrium
A set of strategies selected by participating members to reach the best suitable conditions in game theory. In other words, if all members select their strategies, one cannot raise his own benefits without changing the others’ strategies under the Nash equilibrium.
Subgame perfect Nash equilibrium
In game theory, a subgame perfect equilibrium (or subgame perfect Nash equilibrium) is a refinement of a Nash equilib-rium used in dynamic games. A strategic profile is a subgame perfect equilibrium if it represents the Nash equilibrium of every subgame of the original game. If the players played any
smaller game that is only one part of the larger game then their behaviour represents a Nash equilibrium of that smaller game. Their behavior is then a subgame perfect equilibrium of the larger game.
Bayesian Nash equilibrium
In a Bayesian game, rational players (or risk neutral players) seek to maximize their expected payoff given their beliefs about the other players. When players are risk averse or risk seekers, the assumption is that players will maximize their utility. A Bayesian Nash equilibrium is defined as strategic profiles and beliefs specified for each player to maximize the expected payoff given their beliefs about the other players’ types and strategies.
Perfect Bayesian Nash equilibrium
A perfect Bayesian equilibrium is a strategy profile and a belief system such that the strategies are sequentially rational given the belief system is consistent, wherever possible.
Current practices for a collaborative maintenance chain Maintenance and repair are critical issues for large companies that are heavily invested in engineering assets that have long depreciation and complicated and costly repair procedures. Electrical transformers for power plants match this descrip-tion since the life of a transformer can be considerably short-ened though poor maintenance. Further, the excessive loss of transformers through poor asset management can easily lead to financial ruin. Thus, we use electrical transformers as the case to compare the current practice (as-is) and the improved (to-be) general decision models in the construction of a multi-agent collaborative maintenance platform. In this section, we analyze the existing problems with current trans-former maintenance procedures and form the framework for improved decision models.
Case description
The network system for power transmission and trans-formation consists of electrical power generators, voltage
transformers, and connecting power cables. Figure1depicts the processes for power generation by wind turbine, power collection, power transmission, power distribution and power consumption. The wind turbine uses a rotating blade to drive a power generator, which yields about 600 volts. The power collection transformers collect the power from wind turbines and steps up the voltage to around 35,000 volts. After receiv-ing power from the collection transformer, the large size transformer (the case-study transformer) steps up the voltage to about 220 kilovolts. High voltage transmission from the power grid helps minimize power loss. Before consumption by end users, the power transmission facility must step down or lower the high power transmission voltage. High voltage
power users, such as manufacturing companies, operate their own step down transformers to match their needs. For home and other users, the high voltage power is first stepped down by power distribution centers, and then again by small local transformers to provide 110 volts. The critical issue is to determine the maintenance schedule for the numerous trans-formers in the system to ensure a reliable and uninterrupted power supply. The profit and reliability of the electric power system is used to measure the competitive status of the utility company and also provides an indicator of a nation’s social welfare.
Common maintenance for the power transmission and distribution network relies on the facility’s staff to identify
Fig. 1 Transmission and distribution for wind turbine power
maintenance demands. In most cases, the staff issue main-tenance orders after the equipment shutdown has occurred. In order to overcome crisis reactions to problems, this research studies a new approach to maintaining large size voltage transformers (Fig. 2). The research constructs a negotiation model for an agent based collaborative mainte-nance chain applying game theory to maximize the system benefits.
As-Is maintenance analysis
The maintenance procedure begins with accident repair and is followed by daily maintenance and then periodic mainte-nance. Since it is difficult to prevent unexpected shutdowns in equipment, system monitoring and data analysis using agents are employed. The participants of the maintenance chain are shown in Fig.3.
Asset operation sites: The stakeholder and operator of the voltage transformer utilize the engineering asset for produc-tion or services and is the asset maintenance service receiver. System provider: The engineering asset producer or mainte-nance service provider performs jobs such as routine, emer-gency and preventive maintenance for the asset operation sites.
First-tier collaborators: The first-tier collaborators provide maintenance crews and maintenance parts for the jobs.
The as-is voltage transformer’s current maintenance model is shown in Fig.4. In this diagram, the asset oper-ation sites and the system providers prioritize their objec-tives while conducting maintenance tasks, which can cause resource limitations for the entire system. After the as-is analysis, two improvements are implemented for routine maintenance and emergency repair. According to the above analysis, several problems occur in the current maintenance practice. Asset operation sites 1 Asset operation sites 2 Asset operation sites 3 System provider A System provider B System provider C First tier collaborators First tier collaborators First tier collaborators
Fig. 3 The current maintenance chain structure
Ro u tin e m ain te n an ce Eme rg en t repa ir Go to the asset site to serve
Periodic
maintenance
Record service data
Collaborators System provider operation sites Asset
Daily maintenance
Confirm Service finished
Check demand Notify
Service finished YES Maintenance schedule Shutdown occurred NO Human resource/ maintenance parts allocation Go to the asset site to serve
Confirm Service finished Record service
data
Fig. 4 The as-is operational sequence for current maintenance
proce-dures Asset operation sites 1 Asset operation sites 2 Asset operation sites 3 First tier collaborators First tier collaborators First tier collaborators Service center System provider A System provider A System provider A
Fig. 5 The to-be maintenance chain structure
Initiator Participant X X X X inform-start-of-auction cfp-1 not-understand propose reject-proposal accept-proposal cfp-2 inform-2 request n n n n n-1 1 1 1 1 1 1 1
Fig. 7 The operational
sequence for the to-be
collaborative maintenance chain
Record service data Service center Historical service data consult Condition monitoring/ Data analysis Record service data P reventive maintenance Monitorin g predictive maintenance Go to the asset site to serve
Periodic
maintenance demand
Collaborators System provider Asset
operation sites
Daily maintenance Confirm Service
finished
Check demand
Emergency maintenance negotiation mechanism
Go to the asset site to serve
Service finished YES Confirm Service finished Maintenance schedule Condition monitoring/ Data analysis Exceptional situations occurred Failure cause diagnosis NO Human resource/ maintenance parts allocation Check time/period Equipment reliability and system
profit optimization mechanism
Condition monitoring/ Data analysis Condition monitoring/ Data analysis Periodic maintenance Exceptional situations occurred Exceptional Situations occurred
Table 2 Goals and agent task analysis
Goal Agent type How to achieve the goal
Data/signal extraction Monitoring Agent (MA) • Condition monitoring system Asset Agent (AA) • Data/ signal transformation
Accurate diagnosis Diagnosis Agent (DA) • Collect knowledge from diagnosis experts
• Integrate expert knowledge into diagnosis knowledge base • Provide diagnosis result
Reliable prognosis Prognosis Agent (PA) • Collect knowledge from prognosis expert
• Integrate expert knowledge into prognosis knowledge base • Provide prognosis result
• Provide system profit optimization mechanism Timely and
reli-able maintenance
Service System Agent (SSA) • Arrange maintenance provider Maintenance Decision Support Agent
(MDSA)
• Provide emergency maintenance negotiation mechanism System-Provider Maintenance Scheduling
Agent (SMSA)
• Provide expected maintenance time • Provide expected maintenance start time • Have enough maintenance resources Personalized interface Human-Resource Agent (HRA) • Provide personalized work list Inventory management Spare Part Agent (SPA) • Notification of procurement
Fig. 8 Case diagram for the
agent based collaborative maintenance chain Maintenance Timing Coordination Maintenance Dispatching Maintenance Coordinator Condition
Monitoring Diagnosis andPrognosis
Condition Monitor Diagnosis Expert Asset Owner Maintenance Provider Maintenance Timing Coordination Maintenance Dispatching Maintenance Coordinator Condition
Monitoring Diagnosis andPrognosis
Asset Owner
Maintenance
Provider
Fig. 9 Agent relationship
diagram incorporating different scenarios DA PA SPMSA IA HRA MA AA MA MA REQUEST INFORM REQUEST INFORM REQUEST INFORM CFP,INFORM,RESPONSE PROPOSE,REQUEST CFP,INFORM,RESPONSE PROPOSE,REQUEST RESPONSE REQUEST CFP,INFORM,RESPONSEREQUEST RESPONSE REQUEST RESPONSE REQUEST REQUEST RESPONSE Asset operation site
SSA
MDSA
S
ervice centerSystem provider First
-
tier collaboratorsCFP,INFORM,RESPONSE PROPOSE,REQUEST
Lack of life cycle information for the voltage tranform-er. Besides daily maintenance and periodic maintenance, there is no preventive maintenance.
Lack of historical maintenance records and condition monitoring data.
When asset operation sites notify the system provider of a malfunction, the service delivery time delays often further damage the equipment.
The to-be model for collaborative maintenance
In order to solve the problems mentioned above, this research proposes a service center model. The service center coordi-nates the expert diagnosis and dispatch of repair services. The
center monitors the engineering asset data constantly and integrates the supply and demand decisions between asset operation sites and the system providers (Fig.5).
There are three steps for the maintenance decision-making including the strategic, tactical, and the operational decisions (Tien 2005;Hipel et al. 2007). The strategic level includes the preparation and recovery stages and considers the organiza-tional and financial impacts of the decisions. In this research, the maintenance cost is evaluated based on a reliability cen-tered asset maintenance approach (Bertling et al. 2005).
Total maintenance cost [cost/year]: TCPM(S) = CPMf(S) + CPMP M(S) + CPMint(S)
CPMf(S) The cost of failure (repair costs)
(planed maintenance or replacement of a transformer in advance of failure)
CPMint(S) The cost of interruption (power interruption direct costs and affects the user who incurrs must be compensated via a penalty payment) S Preventive maintenance strategy
After converting to the marginal cost CPMii(S) [cost/ MWh], the model is combined with the Cournot-Nash equi-librium model (CNE) (Chattopadhyay 2004) to consider both equipment reliability and system profit.
Maximize t αt − 1 2βtYt − i,t Gi,tCPMit(S) − i,t 1 2βtG 2 i,t Subject to i Gi,t = Yt∀ t Gi,t ≤ Gmaxi,t × 1− σi,t ∀ (i, t) t σi,t = ∀ i t Gmaxi,t × 1− σi,t − Gi,t ≥ Rt∀ t Gi,tYt ≥ 0 σi,t = 0 or 1
αt, βt Linear demand equation parameters for t
Gi,t Transformer i load in t [MW]
Gmaxi,t Transformer i max load capacity in t [MW]; Yt Total power supply in t [MW]
σi,t Binary maintenance decision variable for i in t
Number of periods transformer needs to be maintained Rt Reserve requirement in period t
The decision making for preventive maintenance mod-els the voltage transformer’s loading. In the other words, the model determines the maintenance time and the service period (σi,t) according to the system losses while the
trans-former is down (CPMii(S)). Thus, the model accounts for both system reliability and profit.
The tactical level decision making includes prediction and prevention. The decision marking considers medium term problems and associated objectives. The Java Agent Devel-opment Framework (JADE 2007) is deployed to construct a MAS for enabling a collaborative maintenance chain. Each agent makes decisions that are influenced directly by the envi-ronment or indirectly through communication and negotia-tion. An agent can decide to cooperate or compete with other agents and an agent is programmed to effectively solve prob-lems. For agent communication and negotiation, the Founda-tion for Intelligent Physical Agents (FIPA) English aucFounda-tion protocol (Fig.6) is used to negotiate maintenance expenses and determine the successful bidder.
MA AA SSA DA SPMSA X X X X 1 1 1 1 n 1 n 1 1 1 n-1 1 ~ ~ n-1 1 n ~ ~ inform-start-of-auction cfp-1 not-understand propose reject-proposal accept-proposal cfp-2 inform-2 request inform cfp propose request request propose
Fig. 10 Sequence of negotiations for a repair proposal
MA AA SSA PA MDSA n n 1 1 1 1 1 inform cfp-1 inform cfp propose request
SPMSA SPA HRA
n n 1 1 1 n n n n 1 1 1 1 request propose request propose propose inform inform inform cfp cfp request 1
Fig. 11 Sequence for condition based preventive maintenance
The operational level includes the detection and response stages. This stage requires timely decision making. A mainte-nance decision support agent (MDSA) and a service system agent (SSA) are used for constructing emergency mainte-nance negotiations. The steps of the negotiation mechanism are described as follows.
Step 1, Initialization. When an exceptional situations occurs, the MDSA provides the maintenance costs and compensation to the SSA.
Step 2, Feasibility Analysis. Check all constraints, including service time, service region, and technology constraints.
Step 3, Evaluation of the maintenance cost, time and period. The danger level is evaluated and the mainte-nance time and cost are determined.
Step 4, Updates of maintenance decisions. MDSA and SSA negotiate a settlement and a response.
The operational sequence diagram for collaborative mainte-nance is shown in Fig.7.
Agent-based collaborative maintenance chain platform Strategic, tactical and operational levels of multi-agent designs are depicted in the following sub-sections. The implementation of the agent-based platform are described in detail.
MAS requirement analysis with strategic level decision making
The proposed information platform is supported by soft-ware agents to autonomously communicate and negotiate among the maintenance chain participants. The IT platform design, the expected goals, and the agent roles are summa-rized in Table2. The case diagram for the MAS is depicted in Fig.8.
Agent relationship analysis with tactical level decision making
After defining the goals of the MAS, the organizations with related resources and knowledge are determined. The inte-grated relationships incorporating different scenarios are shown in Fig.9.
Agent conversation design for operational decision making To complete the design and development of MAS for the collaborative maintenance chain, the blueprints for agent conversations are defined. Figure 10 depicts the negoti-ation for a repair proposal when an asset malfunctions.
Figure 11portrays the condition based preventive mainte-nance sequence.
Agent-based collaborative maintenance chain MAS implementation
Using detailed requirement analysis (from strategic level decisions), agent relationship analysis (from tactical level decisions), and agent conversation design (from operational level decisions), the branches and leaves of the collabora-tive maintenance chain are carefully depicted. Afterward, this research uses the agent development tool, JADE, which follows the FIPA specifications and provides a Graphi-cal User Interface (GUI) to enable the development and debugging of the agent-based system. The web interfaces are designed for all maintenance platform users, including the engineering asset owners and users (at the asset oper-ation site), the service center, the system providers, and the lower tier collaborators and parts suppliers, for real time data displays and decision support. All engineering assets across dispersed locations and their key condition parameters (e.g., temperatures, loads, and noise) are con-stantly monitored and displayed by a web-based dashboard interfaces as shown in Fig. 12. Historical data are held in the service center database and can be viewed using the Internet. For example, the equipment malfunction records may be reviewed by the asset owners as well as the pre-ventive maintenance schedules which are updated period-ically based on the decision model described in section “The to-be model for collaborative maintenance”. Finally, Fig.13depicts the action view of the autonomous agents that
Fig. 13 The execution
sequence of agents working in the background of the collaborative maintenance platform
interactively execute the predefined tasks in the system background.
Conclusions
This research provides an engineering asset maintenance architecture using agents to support maintenance chain col-laboration and decision making. A collaborative business model integrates maintenance chain members with an inter-mediary service center. The collaborative business model links asset sites, system providers, and first tier and lower tier collaborators. The platform provides a reliability and profit optimization mechanism with an emergency negotia-tion mechanism to enable intelligent and collaborative main-tenance procedures. Mainmain-tenance resources, asset condi-tions, multi-goals, and constraints, are considered to achieve improved maintenance system benefits.
Acknowledgments This research was partially supported by the Industrial Technology Research Institute (ITRI) and the National Sci-ence Council. The authors would like to express their appreciation to the JIM reviewers’ valuable comments and suggestions for the preparation of the final manuscript.
References
Bangemann, T., Rebeuf, X., Reboul, D., Schulze, A., Szymanski, J., Thomesse, J. P., Thron, M., & Zerhouni, N. (2006). PROTEUS— creating distributed maintenance systems through an integration platform. Computers in Industry, 57(6), 539–551.
Bertling, L., Allan, R., & Eriksson, R. (2005). A reliability-centered asset maintenance method for assessing the impact of mainte-nance in power distribution systems. IEEE Transactions on Power Systems, 20(1), 75–82.
Bretthauer, G., Gamaleja, T., Handschin, E., Neumann, U., & Hoff-mann, W. (1998). Integrated maintenance scheduling system for electrical energy system. IEEE Transactions on Power Deliv-ery, 13(2), 655–660.
Chattopadhyay, D. (2004). A game theoretic model for strategic maintenance and dispatch decisions. IEEE Transactions on Power Systems, 19(4), 2014–2021.
CIEAM (CRC for Integrated Engineering Asset Management). (2008). Available at http://www.cieam.com/, Accessed on March 15, 2008.
Fu, C., Ye, L., Liu, Y., Yu, R., Iung, B., Cheng, Y., & Zeng, Y. (2004). Predictive maintenance in intelligent control mainte-nance management system for hydroelectric generating unit. IEEE Transactions on Energy Conversion, 19(1), 1–8.
Han, T., & Yang, B. S. (2006). Development of an e-maintenance system integrating advanced techniques. Computers in Indus-try, 57(6), 569–580.
Hipel, K. W., Jamshidi, M. M., Tien, J. M., & White, C. C., III (2007). The future of systems, man, and cybernetics: Appli-cation domains and research methods. IEEE Transactions on Systems, Man, and Cybernetics, 37(5), 726–743.
Hossack, J. A., Menal, J., McArthur, S. D. J., & McDonald, J. R. (2003). A multiagent architecture for protection engi-neering diagnostic assistance. IEEE Transactions on Power Sys-tems, 18(2), 639–647.
Hsiao, D. W., Trappey, A. J. C., Ma, L., Fan, Y.-C., & Mao, Y.-C. (2008). Agent-based integrated and col-laborative engineering asset management. Materials Science Forum, 594, 481–493.
Iung, B. (2003). From remote maintenance to MAS-based e-main-tenance of an industrial process. Journal of Intelligent Manufac-turing, 14(1), 59–82.
JADE (Java Agent DEvelopment framework). (2007).http://jade.tilab. com, Accessed May 28, 2007
Jenab, K., & Zolfaghari, S. (2008). A virtual collaborative maintenance architecture for manufacturing enterprises. Journal of Intelligent Manufacturing, 19(6), 763–771.
Kreps, D. M. (1990). Game theory and economic modeling. New York: Oxford University Press Inc.
Li, Y., Chun, L., & Ching, A. N. Y. (2005). An agent-based platform for web-enabled equipment predictive maintenance. In Proceedings of the 2005 IEEE/WIC/ACM international conference on intelli-gent aintelli-gent technology (IAT’05) (pp. 132–135).
Majidian, A., & Saidi, M. H. (2007). Comparison of fuzzy logic and neural network in life prediction of boiler tubes. International Journal of Fatigue, 29, 489–498.
McArthur, S. D. J., Booth, C. D., McDonald, J. R., & McFa-dyen, I. T. (2005). An agent-based anomaly detection archi-tecture for condition monitoring. IEEE Transactions on Power Systems, 20(4), 1675–1682.
Nagarajan, M., & Sosic, G. (2008). Game-theoretic analysis of coopera-tion among supply chain agents: Review and extensions. European Journal of Operational Research, 187, 719–745.
Sun, Y., Ma, L., Mathew, J., & Zhang, S. (2006). An analytical model for interactive failures. Reliability Engineering & System Safety, 91(5), 495–504.
Tien, J.M. (2005). Viewing urban disruptions from a decision infor-matics perspective. Journal of Systems Science and Systems Engi-neering, 14(3), 257–288.
Yang, Z., Djurdjanovic, D., & Ni, J. (2008). Maintenance sched-uling in manufacturing systems based on predicted machine degradation. Journal of Intelligent Manufacturing, 19(1), 87–98. Yao, Y. H., Lin, G. Y. P., & Trappey, A. J. C. (2005). Using knowl-edge-based intelligent reasoning to support dynamic equipment diagnosis and maintenance. International Journal of Enterprise Information Systems, 2(1), 17–29.
FIPA (Foundation for Intelligent Physical Agents). (2008). Agent com-munication language specification,http://www.fipa.org, Accessed June 1, 2008.
