New Algorithm for FMS Scheduling Based on Multiagent System Architecture

Redefined benefit-driven function is used to study the dynamic scheduling of FMS based on multiagent architecture. Each agent is dedicated to a work center, i.e. a set of the manufacturing system. In one hand, each agent selects locally and dynamically the dispatching rule(DR) that seems to be most suited to the operating conditions, production objectives and current shop status. On the other hand, each task should bring certain amount of benefit for the manufacturer. So, it is reasonable to have the dynamic scheduling of FMS relying upon multiagent architecture using the benefit-driven function as a strategy. Well, today's manufacturing corporation, especially the high & new technology one and deep machining one, the cost of their products is mainly determined by how much the knowledge is input From this viewpoint, we redefined the benefit-driven function, hi the end, this approach is compared with other existing DRs on a job-shop problem, already used in other research works.     

基于改进??学习的知识化制造自适应动态调度策略

针对知识化制造系统生产环境的不确定性, 构建一个基于多Agent 的知识化动态调度仿真系统. 为了保证设备Agent 能够根据当前的系统状态选择合适的中标作业, 提出一种基于聚类-动态搜索的改进??学习算法, 以指导不确定生产环境下动态调度策略的自适应选择, 并给出算法的复杂性分析. 所提出的动态调度策略采用顺序聚类以降低系统状态维数, 根据状态差异度和动态贪婪搜索策略进行学习. 通过仿真实验验证了所提出动态调度策略的适应性和有效性.

     

Development of two-level decision tree-based real-time scheduling system under product mix variety environment

Most of the research on machine learning-based real-time scheduling (RTS) systems has been aimed toward product constant mix environments. However, in a product mix variety manufacturing environment, the scheduling knowledge base (KB) is dynamic; therefore, it would be interesting to develop a procedure that would automatically modify the scheduling knowledge when important changes occur in the manufacturing system. All of the machine learning-based RTS systems (including a KB refinement mechanism) proposed in earlier studies periodically require the addition of new training samples and regeneration of new KBs. Hence, previous approaches investigating machine learning-based RTS systems have been confronted with the training data overflow problem and an increase in the scheduling KB building time, which are unsuitable for RTS control. The objective of this paper is to develop a KB class selection mechanism that can be supported in various product mix ratio environments. Hence, the RTS KB is developed by a two-level decision tree (DT) learning approach. First, a suitable scheduling KB class is selected. Then, for each KB class, the best (proper) dispatching rule is selected for the next scheduling period. Here, the proposed two-level DT RTS system comprises five key components: (1) training samples generation mechanism, (2) GA/DT-based feature selection mechanism, (3) building a KB class label by a two-level self-organizing map, (4) DT-based KB class selection module, and (5) DT-based dynamic dispatching rule selection module. The proposed two-level DT-based KB RTS system yields better system performance than that by a one-level DT-based RTS system and heuristic individual dispatching rules in a flexible manufacturing system under various performance criteria over a long period.     

Learning policies for single machine job dispatching

Reinforcement learning (RL) has received some attention in recent years from agent-based researchers because it deals with the problem of how an autonomous agent can learn to select proper actions for achieving its goals through interacting with its environment. Each time after an agent performs an action, the environment's response, as indicated by its new state, is used by the agent to reward or penalize its action. The agent's goal is to maximize the total amount of reward it receives over the long run. Although there have been several successful examples demonstrating the usefulness of RL, its application to manufacturing systems has not been fully explored. In this study, a single machine agent employs the Q-learning algorithm to develop a decision-making policy on selecting the appropriate dispatching rule from among three given dispatching rules. The system objective is to minimize mean tardiness. This paper presents a factorial experiment design for studying the settings used to apply Q-learning to the single machine dispatching rule selection problem. The factors considered in this study include two related to the agent's policy table design and three for developing its reward function. This study not only investigates the main effects of this Q-learning application but also provides recommendations for factor settings and useful guidelines for future applications of Q-learning to agent-based production scheduling.     

An agent-based service-oriented integration architecture for collaborative intelligent manufacturing

The rapidly changing needs and opportunities of today's global market require unprecedented levels of interoperability to integrate diverse information systems to share knowledge and collaborate among organizations. The combination of Web services and software agents provides a promising computing paradigm for efficient service selection and integration of inter-organizational business processes. This paper proposes an agent-based service-oriented integration architecture to leverage manufacturing scheduling services on a network of virtual enterprises. A unique property of this approach is that the scheduling process of an order is orchestrated on the Internet through the negotiation among agent-based Web services. A software prototype system has been implemented for inter-enterprise manufacturing resource sharing. It demonstrates how the proposed service-oriented integration architecture can be used to establish a collaborative environment that provides dynamic resource scheduling services.     

Application of reinforcement learning for agent-based production scheduling

Reinforcement learning (RL) has received some attention in recent years from agent-based researchers because it deals with the problem of how an autonomous agent can learn to select proper actions for achieving its goals through interacting with its environment. Although there have been several successful examples demonstrating the usefulness of RL, its application to manufacturing systems has not been fully explored yet. In this paper, Q-learning, a popular RL algorithm, is applied to a single machine dispatching rule selection problem. This paper investigates the application potential of Q-learning, a widely used RL algorithm to a dispatching rule selection problem on a single machine to determine if it can be used to enable a single machine agent to learn commonly accepted dispatching rules for three example cases in which the best dispatching rules have been previously defined. This study provided encouraging results that show the potential of RL for application to agent-based production scheduling.     

基于合同网协议交互投标的动态调度方法研究

柔性制造系统使生产加工路径有很多可选性，所以调度系统必须考虑机器调度问题。分配规则调度是一种最基本、最具影响力的动态调度方法。然而，分配规则调度方法很少考虑机器顺序选择。兼顾工件选择和机器选择两方面，本文运用交互投标过程，构建基于合同网协议调度的协商规则。研究作业车间动态调度问题，提出并构建了5种合同网规则调度方法。通过实验分析结果表明，基于合同网交互投标模式的规则调度能够大大改善调度系统性能，提高设备的利用率和设备负荷平衡指标。     

Multi-agent Mediator architecture for distributed manufacturing

A generic Mediator architecture for distributed task planning and coordination has been developed using multi-agent paradigms. In this approach, agents function autonomously as independent computing processes, and dynamic virtual clusters coordinate the agent's activities and decision making. This coordination involves dynamically created coordination agents and resource agents concurrently. The Mediator architecture contains three levels of these coordination agents: the template mediator, the data-agent manager, and the active mediator. The template mediator is the top-level global coordinator. This agent contains both the templates and the cloning mechanism to create the successively lower-level agents. Task plans are decomposed successively into subtasks, which are allocated to dynamically created agent clusters coordinated through data-agent managers and active mediators. Coordination of agent activity takes place both among the clusters and within each cluster. The system dynamically adapts to evolving manufacturing tasks, with virtual agent clusters being created as needed, and destroyed when their tasks are completed. The mediator architecture and related mechanisms are demonstrated using an intelligent manufacturing scheduling application. Both the machines and the parts involved in this production system are considered as intelligent agents. These agents use a common language protocol based on the Knowledge Query Manipulation Language (KQML). The generic Mediator approach can be used for other distributed organizational systems beyond the intelligent manufacturing application it was originally developed for.     

A multi-agent based approach to dynamic scheduling with flexible processing capabilities

A multi-agent based system is proposed to simultaneous scheduling of flexible machine groups and material handling system working under a manufacturing dynamic environment. The proposed model is designed by means of \(\hbox {Prometheus}^{\mathrm{TM}}\) methodology and programmed in \(\hbox {JACK}^{\mathrm{TM}}\) agent based systems development environment. Each agent in the model is autonomous and has an ability to cooperate and negotiate with the other agents in the system. Due to these abilities of agents, the structure of the system is more suitable to handle dynamic events. The proposed dynamic scheduling system is tested on several test problems the literature and the results are quite satisfactory because it generates effective schedules for both dynamic cases in the real time and static problem sets. Although the model is designed as an online method and has a dynamic structure, obtained schedule performance parameters are very close to those obtained from offline optimization based algorithms.     

An agent-oriented approach to resolve scheduling optimization in intelligent manufacturing

Agent technology is considered as a promising approach for developing optimizing process plans in intelligent manufacturing. As a bridge between computer aided design (CAD) and computer aided manufacturing (CAM), the computer aided scheduling optimization (CASO) plays an important role in the computer integrated manufacturing (CIM) environment. In order to develop a multi-agent-based scheduling system for intelligent manufacturing, it is necessary to build various functional agents for all the resources and an agent manager to improve the scheduling agility. Identifying the shortcomings of traditional scheduling algorithm in intelligent manufacturing, the architecture of intelligent manufacturing system based on multi-agent is put forward, among which agent represents the basic processing entity. Multi-agent-based scheduling is a new intelligent scheduling method based on the theories of multi-agent system (MAS) and distributed artificial intelligence (DAI). It views intelligent manufacturing as composed of a set of intelligent agents, who are responsible for one or more activities and interacting with other related agents in planning and executing their responsibilities. In this paper, the proposed architecture consists of various autonomous agents that are capable of communicating with each other and making decisions based on their knowledge. The architecture of intelligent manufacturing, the scheduling optimization algorithm, the negotiation processes and protocols among the agents are described in detail. A prototype system is built and validated in an illustrative example, which demonstrates the feasibility of the proposed approach. The experiments prove that the implementation of multi-agent technology in intelligent manufacturing system makes the operations much more flexible, economical and energy efficient.