Content uploaded by Khalid Amechnoue
Author content
All content in this area was uploaded by Khalid Amechnoue on Mar 25, 2017
Content may be subject to copyright.
Royaume du Maroc
Congrès Scientif ique International
MANAGEMENT ET
INGENIERIE DES SYSTEMES
LA REVUE SCIENTIFIQUE INTERNATIONALE : MANAGEMENT ET INGENIERIE DES SYSTEMES
2
CONGRES INTERNATIONAL :
MANAGEMENT ET INGENIERIE DES
SYSTEMES
Actes
sélectionnés à la cinquième édition du congrès
international MIS 2017 à Rabat (Maroc)
(24 et 25 fevrier 2017)
Les articles publiés dans cette revue scientifique n‘engagent que la
responsabilité de leurs auteurs.
Dépôt légal : 2012 PE 0120
ISSN : 2028 – 9421
Président du congrès
Directeur de la revue « MIS »
Docteur Aziz SOULHI, Professeur à l’ENSMR
LA REVUE SCIENTIFIQUE INTERNATIONALE : MANAGEMENT ET INGENIERIE DES SYSTEMES
177
MULTI-AGENT SYSTEM FOR PROCUREMENT OF RAW MATERIAL IN
SUPPLY CHAIN
Ms I. ACHATBI *, Pr. K. AMECHNOUE**, Ms S. AOULAD ALLOUCH***
* Laboratory of Innovative Technologies, E-mail i.achatbi@gmail.com
**National School of Applied Sciences-Tangier, E-mail kamechnoue@gmail.com
*** Laboratory of Innovative Technologies, E-mail saloua.aoulad@gmail.com
Abstract
Supply chain involves a set of nodes, geographically distributed throughout the globe and linearly ordered
by the material, information or financial flow in order to analyse or synthesize a specific set of logistic
functions and costs. One of supply chain problems is the procurement management which take the interest
of researches in domain of Supply chain management. Multi agent systems are appropriate for distributed
decision and domains that involve interactions between different organisations with proprietary information,
the fact that agents are autonomous, pro-active, sociable and reactive. In this paper, we give an overview of
Multiagent-based supply chain management covering the interaction between agents inside production node,
to solve the problem related to the procurement of raw material. The ontology, in turn, is used to represent
the domain knowledge and enable the agent to share available knowledge and identify new knowledge.
Key words: Multi-agent system, Supply chain, Ontology.
1.
Introduction
Since the advent of globalization and the evolution of organizations, in a world increasingly competitive,
to more responsiveness, agility and flexibility, collaborative work between all actors in the supply chain is
required, and each firm modifies its behaviour to adapt to market and competition evolutions.
Supply chain is a set of companies acting to design, engineer, market, manufacture, and distribute
products and services to end-consumers [1]. In general, this set of firms is structured as a network. This
network gathered basic material suppliers, manufacturers, logistic centres, warehouses, intermediaries,
transport companies, wholesalers, retailers, and other participants. Supply chain management SCM is
concerned with planning and scheduling the different activities of organization across the supply chain, from
raw material procurement to complete finished goods delivery. Supply chain management involves
planning, supply, production, delivery and return, each stage has its own processes, issues and solutions [2].
From the industry viewpoint, production planning, inventory management and transport are the main
areas of finding significant problems in SCM. Legacy software systems have been developed for SCM, such
as planning system (APS), enterprise resources planning (ERP) and e-commerce systems. More and more,
multiagent systems are seen as a new technology for improving or replacing technologies used in both
transactional and analytical information technologies. The ability to negotiate is the unique feature of agents
that distinguishes them from other software. In fact, agents are best suited for applications that are modular,
decentralized, changeable, ill-structured and complex.
Agents working cooperatively can communicate with each other by means of a common ontology.
Ontology is machine readable, and supports agent communication by defining and providing a shared
vocabulary to be used in the course of communication. Ontologies not only provide a definition of the terms
that can be used in communication; ontologies also provide the definition of the world in which an agent
grounds its actions. Different agents of a system can reach a shared understanding by committing to the
same ontology. Two important functions of ontologies are that they:
1) Enable agents to work cooperatively to communicate with each other,
2) Make the available information more accessible to automated agents.
This work encompasses the procurement and inventory management of raw materials. Raw materials and
supplied components constitute 60% to 70% of finished goods‘ cost[3]. Delivery times of suppliers and
distribution reliability affecting more than production time on the stock level and the quality of service of
each manufacturer[4].Thus, ensuring timely and sufficient commodity saturation at a warehouse is
LA REVUE SCIENTIFIQUE INTERNATIONALE : MANAGEMENT ET INGENIERIE DES SYSTEMES
178
important, in order to avoid surplus of goods in stock (overstock) and deficiency of goods in warehouses
(out-of-stock).
The rest of this paper is structured as follows. Section 2 reviews the related research works on multi-
agent-based supply chain management. In section 3 we present the coordination in supply chain. Section 4
presents the global supply chain network. Section 5 describes ontology and agent technology. Section 6
shows the generic multi agent system architecture. Section 7 provides multi agent system construction.
Finally, conclusion is drawn in Section 8.
2.
Related works
Multi-agent technology has been used in many areas but industry applications have taken the earliest
advancement of agent technology when compared to others. In fact, the distributed manufacturing units
have the same characteristics as agents[5], they allow relaxing the constraints of centralized, planned,
sequential control[6].
The major focusing points for researches on agent-based architecture are the negotiation protocols and
agent decision-making models. For the last one, there is no one universal model for each separate case.
The authors of [7] specified that the multiagent software system is suitable for tasks including interaction
between different organizations with different goals and available information. Each node in the supply
chain is represented by a separate agent; the agents cooperate with each other to implement the functionality
of the system.
MASCOT (MultiAgent Supply Chain cOordination Tool) is a reconfigurable, multilevel, agent-based
architecture for planning and scheduling aimed at improving supply chain agility. It coordinates production
among multiple facilities, and evaluates new product/subcomponent designs and strategic business decisions
with regard to capacity and material requirements across the supply chain[8].
OCEAN (Organization and Control Emergence with an Agent Network) is a control system with an
open, decentralized and constraints-based architecture in which there is responsiveness, and distribution of
production resources and technical data. This system was designed to react to environment dynamics in
order to show that cooperation at the global level may emerge from competitions at the local level.
The authors of [9] proposed an agent-based distributed architecture for simulation in decision-making
processes within the supply chain context. Agents in this architecture use a set of negotiation protocols to
make decisions collectively in a short time.
A multi-agent approach to model supply chain dynamics is proposed in [10]. In their approach, a supply
chain library of software components, such as retailers, manufacturers, inventory policy, and so on, has been
developed to build customized supply chain models from the library.
The authors of [11] specified that there is a new architecture of supply chain management at the tactical
and operational levels in recent years. The supply chain is represented by a set of intelligent software agents,
where everyone is responsible one or more of its activities, and the agents interact with each other to plan
and carry out their duties.
A multi-agent system application for supply chain node cooperation is proposed in [12], the system is
developed to solve the problem of procurement and inventory of raw material for microchips manufacturing
enterprise.
3.
Coordination in supply chain management
Supply chain involves several partners who collaborate to produce goods with best results. Participants in
the supply chain should not behave as competitors but as partners sharing the same goal and taking risks in
common.
The lack of coordination in supply chain may decrease the competitiveness of the entire chain. This turns
into higher costs, tighter profit margins, delivery delays, order losses and poor quality of service to
customers. As a result, the bullwhip effect can be identified as a major problem in the domain of SCM. This
effect consists in an amplification of the order variability. This variability is a problem because it makes
demand more unpredictable [13].
LA REVUE SCIENTIFIQUE INTERNATIONALE : MANAGEMENT ET INGENIERIE DES SYSTEMES
179
nd export as exchange rates, customs duties, in
tudying the opportunities to source all around t
ndered cost [20]. This cost must integrate all p
rtation cost, the cost of customs, and cost of stor
in with four nodes: suppliers, manufacturers, di
4.
Ontology and agent technology
a- Agent:
An agent is a computer system that is situated in some environment, and that is capable of autonomous
action in this environment in order to meet its design objectives[17]. Agents can also interact with other
agent (sometimes humans) via some kind of agent communication language (ACL). An agent state consists
of components such as beliefs, decisions, capabilities and obligations. As an agent state is more
sophisticated, it is also referred to as a mental state. The main difference between agent and objects is the
autonomy of agents. In fact, while objects encapsulate some state on which their methods can perform
actions, and in particular the action of invoking another object‘s method, an object has control over its
behaviour. That is, if an object is asked to perform an action, it always does so, while an agent may refuse.
Concerning this point, Wooldridge recalls the slogan ―Objects do it for free; agents do it because they want
to‖. There are different levels of complexity of implementation the agent. Such complexity depends on the
task that agents have to carry out and on the environment surrounding them. Different classifications of
agent architectures are proposed: Simple reflex agents, Model-based reflex agents, Goal-based agents,
Utility- based agents, and Learning agents. This multiple agents are detailed in [18].
b- Ontology:
As a novel knowledge organization concept, ontology is more than just a vocabulary and taxonomy of
terms, it provides a set of well-founded constructs that can be leveraged to build meaningful higher level
knowledge and relationships between terms [19]. By describing a set of concepts and the relationships
between them, ontology can construct both the hierarchical architecture of the negotiation knowledge and
the descriptive logics of negotiation regulations and activities. The key idea of ontology is to have
agreement explicitly interpreted by software tools rather than just being implicitly interpreted by a human.
According to the ontology structure, inference rules can be defined to guide agents negotiation behaviours
to adapt to various negotiation environments and personal attitudes. It can be seen that ontology is a
promising solution to organize negotiation knowledge and facilitate agent reasoning ability.
5.
Global supply chain network
If sites of supply chain are located in different countries, we talk about the global supply chain. Thus, the
aspects relating to the import a surance and legislation should
be taken into account.
Global Sourcing involves s he world. This choice is made
by calculating procurement re rocurement logistics elements
as the purchase price, transpo age along the flow. The figure
1 represent a global supply cha stributors, and retailers.
Figure 12: Global supply chain network
LA REVUE SCIENTIFIQUE INTERNATIONALE : MANAGEMENT ET INGENIERIE DES SYSTEMES
180
We consider in this paper, the procurement of raw material in a global supply chain. In this case, various
levers are introduced to optimise procurement such as purchases and procurement transportation.
The purchasing function evolves greatly in recent years. In the automobile industry, the value of a vehicle
comes largely from suppliers (2/3 in 2000 and 3/4 to nearly 2015, according to AMR Research). Thus, the
main objective of purchasing is systematically reducing purchasing costs.
In view of the rising costs of energy, optimizing procurement transport has become a major competitive
economic challenge. Two levers are presented: Use of ―Ex-work price‖ and ―milk run‖ generalization.
When transport becomes intercontinental, the use of ocean freight is necessary. International transport has
also its optimization levers to reduce the cost. We can mention: improvement of transportation, approach
and distribution, the cost of freight, customs optimization. The problem of how to find the cheapest transport
solution can be seen as a shortest (cheapest) path problem, in the transportation network [21].
6.
Multi agent system architecture
This section presents our solution for procurement of raw material in global supply chain. Multiagent
system hands the decentralization in solving the problem, which is important in our case because agents are
distributed in different node of supply chain.
The proposed solution will be able to model complex system for supply chain management to address the
issues of communication among the parties involved in supply chain, it consists of: SCM Agent,
procurement agent, Seller Agent, TransportProduct Agent, in manufacturing enterprise; Supplier Agent,
MaterialSeller Agent, TransportSupplier Agent in supplier node; and finally in the buyer side, the buyer
Agent is present (Figure 2).
To ensure process production of finished goods, timely coordination of the elements of orders with
suppliers is required, namely lists of materials needed, their quantity, price, and delivery time.
Seller agent receives an order to produce such merchandise with such quantity from the Buyer Agent.
SCM agent command Procurement Agent to buy raw materials, the last one agrees with Supplier agent
about price and delivery time.
Once the supplier is selected, if transportation of raw material isn‘t assured by the supplier (transport not
available), TransportProduct agent triggers the negotiation with Transport Agent to insure transportation of
raw materials from supplier to the warehouse for raw material in production node. The outputs of
TransportProduct Agent are supplier node, delivery node, product type, quantity, and delivery time.
Figure 13: Multi Agent system architecture
LA REVUE SCIENTIFIQUE INTERNATIONALE : MANAGEMENT ET INGENIERIE DES SYSTEMES
181
The negotiation between agents inside manufacturing node is ensured by Ontology Production. Ontology
Material, Ontology Transport, and Ontology FinishedGood are common ontologies to assure the interaction
of agents from different organizations.
7.
Multi agent system construction
As a promising technology JADE[22] will be used to automate procurement management process and
Eclipse Luna as integrated development environment to construct the multiagent system. Despite of other
multi-agent technologies JADE is available as free software component hence development cost is marginal.
Ability to perform under limited resource environment, installation and access through the mobile devises
increased the rapid growth of multi-agent technology. Instead of the standalone environment the agent is
accessible through the web interface with minimum bandwidth. Protégé and Ontology Bean Generator to
create a domain ontology and to transform it into JADE classes, MySQL to support the database, ACL
(agent communication message) messages to transfer information, share knowledge and negotiate with each
other using FIPA negotiation protocols.
8.
Conclusion
In this paper, we present multi agent system resolving the problem of coordination to purchase the raw
material and select transporter, the system can be applied to forecast trends in supply chain functions and to
maximize business objective of an organization and finally satisfy the end- customer requirements.
Negotiation between different agents is ensured by ontologies, and JADE platform to implement the
proposed architecture.
References
[1] J. A. Muckstadt, D. H. Murray, J. A. Rappold, and D. E. Collins, ―Guidelines for collaborative supply
chain system design and operation,‖ Inf. Syst. Front., vol. 3, no. 4, pp. 427–453, 2001.
[2] J. J. Oberholzer and R. A. BOTHA, ―„How Agent Technologies Could be Used in Supply Chain
Management‖,‖ in Proceedings of SAICSIT, 2003, pp. 111–121.
[3] L. Ouzizi, ―Planification de la production par co-décision et négociation de l‘entreprise virtuelle,‖ Metz,
2005.
[4] R. L. Harmon and L. D. Peterson, Reinventing the factory II: managing the world class factory, vol. 2.
Simon and Schuster, 1992.
[5] S. Viswanathan and R. Piplani, ―Coordinating supply chain inventories through common replenishment
epochs,‖ Eur. J. Oper. Res., vol. 129, no. 2, pp. 277–286, 2001.
[6] H. V. D. Parunak, ―Applications of distributed artificial intelligence in industry,‖ Found. Distrib. Artif.
Intell., vol. 4, pp. 139–164, 1996.
[7] Y. Chen, Y. Peng, T. Finin, Y. Labrou, S. Cost, B. Chu, R. Sun, and R. Willhelm, ―A negotiation-based
multi-agent system for supply chain management,‖ Work. Notes Agents, vol. 99, 1999.
[8] N. M. Sadeh, D. W. Hildum, D. Kjenstad, and A. Tseng, ―Mascot: an agent-based architecture for
coordinated mixed-initiative supply chain planning and scheduling,‖ in In Workshop on Agent-Based
Decision Support in Managing the Internet-Enabled Supply-Chain, at Agents‘ 99, 1999.
[9] E. H. Nfaoui, Y. Ouzrout, O. El Beqqali, and A. Bouras, ―An approach of agent-based distributed
simulation for supply chains: Negotiation protocols between collaborative agents,‖ in The 20th annual
European Simulation and Modelling Conference (2006), 2006, pp. 290–295.
[10] J. M. Swaminathan, S. F. Smith, and N. M. Sadeh, ―Modeling supply chain dynamics: A multiagent
approach,‖ Decis. Sci., vol. 29, no. 3, pp. 607–632, 1998.
[11] V. Kumar and S. Srinivasan, ―A review of supply chain management using multi-agent system,‖ Int. J.
Comput. Sci. Issues, vol. 7, no. 5, pp. 198–205, 2010.
[12] D. Plinere and A. Borisov, ―A negotiation-based multi-agent system for supply chain management,‖
Sci. J. Riga Tech. Univ. Comput. Sci., vol. 45, no. 1, pp. 128–132, 2011.
[13] H. L. Lee, V. Padmanabhan, and S. Whang, ―Information distortion in a supply chain: The bullwhip
effect,‖ Manag. Sci., vol. 43, no. 4, pp. 546–558, 1997.
LA REVUE SCIENTIFIQUE INTERNATIONALE : MANAGEMENT ET INGENIERIE DES SYSTEMES
182
[14] D. J. Thomas and P. M. Griffin, ―Coordinated supply chain management,‖ Eur. J. Oper. Res., vol. 94,
no. 1, pp. 1–15, 1996.
[15] S. Kreipl and M. Pinedo, ―Planning and scheduling in supply chains: an overview of issues in practice,‖
Prod. Oper. Manag., vol. 13, no. 1, pp. 77–92, 2004.
[16] T. Boyacı and G. Gallego, ―Coordinating pricing and inventory replenishment policies for one
wholesaler and one or more geographically dispersed retailers,‖ Int. J. Prod. Econ., vol. 77, no. 2, pp. 95–
111, 2002.
[17] M. Wooldridge, An introduction to multiagent systems. John Wiley & Sons, 2009.
[18] S. J. Russell, P. Norvig, J. F. Canny, J. M. Malik, and D. D. Edwards, Artificial intelligence: a modern
approach, vol. 2. Prentice hall Upper Saddle River, 2003.
[19] K. W. Cheung, C. H. Li, E. C. Lam, and J. Liu, ―Customized electronic commerce with intelligent
software agents,‖ IDEA Group Publ., pp. 150–176, 2001.
[20] D. J. Bowersox, D. J. Closs, and M. B. Cooper, Supply chain logistics management, vol. 2. McGraw-
Hill New York, NY, 2002.
[21] P. Davidsson, J. Holmgren, J. A. Persson, and L. Ramstedt, ―Multi agent based simulation of transport
chains,‖ in Proceedings of the 7th international joint conference on Autonomous agents and multiagent
systems-Volume 2, 2008, pp. 1153–1160.
[22] W. Um, H. Lu, T. J. Hall, and others, ―A study of Multi-agent based supply chain modeling and
management,‖ iBusiness, vol. 2, no. 4, p. 333, 2010.
