The dynamic variations in demand patterns and produ ct mix, driven by unpredictable changes in a global market, are placing manufactur ing systems under significant pressure. In order to remain competitive, manu factu...The dynamic variations in demand patterns and produ ct mix, driven by unpredictable changes in a global market, are placing manufactur ing systems under significant pressure. In order to remain competitive, manu facturing organisations must satisfy demands timely. This implies that companies must increase product varieties, reduce time-to-market, shorten product-life cycles and at the same time maintain good quality and reduce investment costs. Conventional methodologies for planning and control have been found to be inadeq uate in meeting these challenges. Agile manufacturing is the state-of-the-art concept that provides enterprises with the opportunity to react rapidly and cos t-effectively to changes that occur in their environment. Several paradigms suc h as Holonic Manufacturing Systems (HMS), Bionic Manufacturing Systems (BMS) and Fractal Factory have been developed to enable manufacturing systems achieve agi lity by integrating manufacturing activities into a coordinated framework. Despi te the differences in their origin (HMS from social organisation, BMS from biolo gy and Fractal Factory from Mathematics), these paradigms have overlapping conce pts and one of the most important is hierarchical organisational structure. This paper presents a conceptual hierarchically structured multi-agent architec ture for manufacturing systems’ modelling. Multi-Agent Systems (MAS) provide su itable techniques for implementing the above concepts and as a branch of Distrib uted Artificial Intelligence (DAI), have characteristics that have been explored in various applications. Such characteristics include self-organisation, flexi bility, scalability, and robustness. The proposed architecture provides a suit able decision-making framework where each agent represents a node in the hier archic tree of manufacturing systems such as the company as whole, each plant wi thin the company, each cell or line within the plant, each machine in a cell or line. Each agent has the ability to perceive and evaluate changes that occur in the manufacturing environment, interact with other agents in the system in order to reach an optimal decision, and act based on that decision. In other words, agents respond timely to unexpected changes by continuously co-ordinating t heir activities, and allocating manufacturing resources dynamically based on act ual shop-floor situation. The flexibility of this architecture also lies in its ability to accommodate both homogenous and heterogeneous agents, and its capabi lity for the dynamic addition and removal of agents using a conceptual intellige nt communication mechanism similar to the blackboard messaging system. A Bidding -Based Scheme (BBS) would be used to generate and evaluate alternative scenario at run-time. In addition, this architecture can be extended to meet the requir ements of enterprise integration.展开更多
Agile manufacturing execution systems (AMES) are used to help manufacturers optimize shop floor production in an agile way. And the modeling of AMES is the key issue of realizing AMES. This paper presents an agent-bas...Agile manufacturing execution systems (AMES) are used to help manufacturers optimize shop floor production in an agile way. And the modeling of AMES is the key issue of realizing AMES. This paper presents an agent-based approach to AMES modeling. Firstly, the characteristics of AMES and its requirements on modeling are discussed. Secondly, a comparative analysis of modeling methods is carried out, and AMES modeling using an agent-based approach is put forward. Agent-based modeling method not only inherit the favorable features of traditional object-oriented modeling method such as data encapsulation, modularity and so on, but also has the ability to construct intelligent, rational and autonomous agent which can cooperate together to realize the goal of agile operation. A general agent architecture used in AMES modeling is described. Under this architecture, an agent can be divided into domain-independent components and domain-specific components which helps solve problems such as information overload, incomplete information handling and soft decision-making. Furthermore, an AMES model using four types of agents, i.e., interface agent, information agent, resource agent and management agent, is established. Thirdly, a snapshot of AMES model is provided in the case study. Especially, an agent-based cooperating process of task scheduling in AMES is illustrated in detail. Finally, the advantages and disadvantages of this modeling approach are discussed as well.展开更多
文摘The dynamic variations in demand patterns and produ ct mix, driven by unpredictable changes in a global market, are placing manufactur ing systems under significant pressure. In order to remain competitive, manu facturing organisations must satisfy demands timely. This implies that companies must increase product varieties, reduce time-to-market, shorten product-life cycles and at the same time maintain good quality and reduce investment costs. Conventional methodologies for planning and control have been found to be inadeq uate in meeting these challenges. Agile manufacturing is the state-of-the-art concept that provides enterprises with the opportunity to react rapidly and cos t-effectively to changes that occur in their environment. Several paradigms suc h as Holonic Manufacturing Systems (HMS), Bionic Manufacturing Systems (BMS) and Fractal Factory have been developed to enable manufacturing systems achieve agi lity by integrating manufacturing activities into a coordinated framework. Despi te the differences in their origin (HMS from social organisation, BMS from biolo gy and Fractal Factory from Mathematics), these paradigms have overlapping conce pts and one of the most important is hierarchical organisational structure. This paper presents a conceptual hierarchically structured multi-agent architec ture for manufacturing systems’ modelling. Multi-Agent Systems (MAS) provide su itable techniques for implementing the above concepts and as a branch of Distrib uted Artificial Intelligence (DAI), have characteristics that have been explored in various applications. Such characteristics include self-organisation, flexi bility, scalability, and robustness. The proposed architecture provides a suit able decision-making framework where each agent represents a node in the hier archic tree of manufacturing systems such as the company as whole, each plant wi thin the company, each cell or line within the plant, each machine in a cell or line. Each agent has the ability to perceive and evaluate changes that occur in the manufacturing environment, interact with other agents in the system in order to reach an optimal decision, and act based on that decision. In other words, agents respond timely to unexpected changes by continuously co-ordinating t heir activities, and allocating manufacturing resources dynamically based on act ual shop-floor situation. The flexibility of this architecture also lies in its ability to accommodate both homogenous and heterogeneous agents, and its capabi lity for the dynamic addition and removal of agents using a conceptual intellige nt communication mechanism similar to the blackboard messaging system. A Bidding -Based Scheme (BBS) would be used to generate and evaluate alternative scenario at run-time. In addition, this architecture can be extended to meet the requir ements of enterprise integration.
文摘Agile manufacturing execution systems (AMES) are used to help manufacturers optimize shop floor production in an agile way. And the modeling of AMES is the key issue of realizing AMES. This paper presents an agent-based approach to AMES modeling. Firstly, the characteristics of AMES and its requirements on modeling are discussed. Secondly, a comparative analysis of modeling methods is carried out, and AMES modeling using an agent-based approach is put forward. Agent-based modeling method not only inherit the favorable features of traditional object-oriented modeling method such as data encapsulation, modularity and so on, but also has the ability to construct intelligent, rational and autonomous agent which can cooperate together to realize the goal of agile operation. A general agent architecture used in AMES modeling is described. Under this architecture, an agent can be divided into domain-independent components and domain-specific components which helps solve problems such as information overload, incomplete information handling and soft decision-making. Furthermore, an AMES model using four types of agents, i.e., interface agent, information agent, resource agent and management agent, is established. Thirdly, a snapshot of AMES model is provided in the case study. Especially, an agent-based cooperating process of task scheduling in AMES is illustrated in detail. Finally, the advantages and disadvantages of this modeling approach are discussed as well.
