Agent learning in autonomic manufacturing execution systems for enterprise networking

In enterprise networks, companies interact on a temporal basis through client–server relationships between order agents (clients) and resource agents (servers) acting as autonomic managers. In this work, the autonomic MES (@MES) proposed by Rolón and Martinez (2012) has been extended to allow selfish behavior and adaptive decision-making in distributed execution control and emergent scheduling. Agent learning in the @MES is addressed by rewarding order agents in order to continuously optimize their processing routes based on cost and reliability of alternative resource agents (servers). Service providers are rewarded so as to learn the quality level corresponding to each task which is used to define the processing time and cost for each client request. Two reinforcement learning algorithms have been implemented to simulate learning curves of client–server relationships in the @MES. Emerging behaviors obtained through generative simulation in a case study show that despite selfish behavior and policy adaptation in order and resource agents, the autonomic MES is able to reject significant disturbances and handle unplanned events successfully.     

Toward a cloud-based manufacturing execution system for distributed manufacturing

This paper illustrates the needs and challenges for the management of distributed manufacturing in a multi-company supply chain and processes these further as features of new IT systems. Requirements are collected from manufacturing companies and combined with insights from literature in the field of current ERP/MES system drawbacks, advantages, needs and challenges. The findings show that the needs and challenges in data integration inside SME networks are closely related to the limitations of current supply chain solutions. Current ERP-solutions lack extended enterprise support and a shared cloud-based approach. On the other hand, current MES solutions can operate the manufacturing process, but not for distributed manufacturing. As an answer to the requirements, we made a proposal for the core of architecture for next generation of MES solution in this position paper. Moreover, a pilot software tool has been developed to support the needs related to real time, cloud-based, light weight operation.     

Agent-based distributed manufacturing control: A state-of-the-art survey

Manufacturing has faced significant changes during the last years, namely the move from a local economy towards a global and competitive economy, with markets demanding for highly customized products of high quality at lower costs, and with short life cycles. In this environment, manufacturing enterprises, to remain competitive, must respond closely to customer demands by improving their flexibility and agility, while maintaining their productivity and quality. Dynamic response to emergence is becoming a key issue in manufacturing field because traditional manufacturing control systems are built upon rigid control architectures, which cannot respond efficiently and effectively to dynamic change. In these circumstances, the current challenge is to develop manufacturing control systems that exhibit intelligence, robustness and adaptation to the environment changes and disturbances. The introduction of multi-agent systems and holonic manufacturing systems paradigms addresses these requirements, bringing the advantages of modularity, decentralization, autonomy, scalability and re-usability. This paper surveys the literature in manufacturing control systems using distributed artificial intelligence techniques, namely multi-agent systems and holonic manufacturing systems principles. The paper also discusses the reasons for the weak adoption of these approaches by industry and points out the challenges and research opportunities for the future.     

The model-based product agent: A control oriented architecture for intelligent products in multi-agent manufacturing systems

The multi-agent control strategy has been previously shown to improve the flexibility of complex, dynamic manufacturing systems. One key component of this strategy is the product agent. The product agent is responsible for autonomously guiding a physical part in the manufacturing system based on its production goals. Though the product agent has been described in previous works, a fully developed software architecture for the product agent that uses a model-based optimization approach has not been proposed. In this work, a product agent architecture with the capabilities to explore the local environment, plan and schedule events based on its knowledge, and request desired actions from the resources in the system is presented and tested.     

A holonic approach to dynamic manufacturing scheduling

Manufacturing scheduling is a complex combinatorial problem, particularly in distributed and dynamic environments. This paper presents a holonic approach to manufacturing scheduling, where the scheduling functions are distributed by several entities, combining their calculation power and local optimization capability. In this scheduling and control approach, the objective is to achieve fast and dynamic re-scheduling using a scheduling mechanism that evolves dynamically to combine centralized and distributed strategies, improving its responsiveness to emergence, instead of the complex and optimized scheduling algorithms found in traditional approaches.     

Applications of agent-based systems in intelligent manufacturing: An updated review

Agent technology has been recognized as a promising paradigm for next generation manufacturing systems. Researchers have attempted to apply agent technology to manufacturing enterprise integration, enterprise collaboration (including supply chain management and virtual enterprises), manufacturing process planning and scheduling, shop floor control, and to holonic manufacturing as an implementation methodology. This paper provides an update review on the recent achievements in these areas, and discusses some key issues in implementing agent-based manufacturing systems such as agent encapsulation, agent organization, agent coordination and negotiation, system dynamics, learning, optimization, security and privacy, tools and standards.     

基于Agent的复杂系统分布仿真建模方法的研究

基于Agent的分布仿真是研究大型复杂系统的一种有效的、重要的方法。为了减小复杂系统仿真的复杂度,增加仿真模型的重用和可维护性,需要研究基于Agent分布仿真的建模方法。首先对复杂系统及其特性进行了分析,对基于Agent的仿真进行了全面的论述,然后对基于Agent的复杂系统仿真中的复杂系统建模分析、Agent建模分析以及Agent的分布进行了分析,给出了基于Agent的复杂系统分布仿真的建模步骤,最后给出了在此建模思想指导下的金融证券市场的建模过程。     

A multi-agent based agile manufacturing planning and control system

In today’s manufacturing enterprise, the performance of customer service level (e.g., short ordering-to-delivery time, low price) is highly dependent on the effectiveness of its manufacturing planning and control system (MPCS). However, most of the current MPCS, employed the hierarchical planning approach, may have some drawbacks, such as structural rigidity, difficulty of designing a control system, and lack of flexibility. Currently, RFID (Radio Frequency Identification) technology has been applied to enhance the visibility, accountability, track ability and traceability of manufacturing system whenever the accurate and detailed manufacturing information (e.g., raw material, WIP, products in factory and products in the down streams) of products will be followed in real-time basis by RFID technique. In addition, a multi-agent approach may be applied in a distributed and autonomous system which allows negotiation-based decision making. Therefore, the objective of this research is to study the application of RFID technique and multi-agent system (MAS) in developing an agent-based agile manufacturing planning and control system (AMPCS) to respond to the dynamically changing manufacturing activities and exceptions.In AMPCS, RFID-based manufacturing control (R-MC) module plays the role of controlling the manufacturing system in which production items and manufacturing resources attached with RFID tag may actively feedback production status to and receive production and operations schedule from advanced manufacturing planning (AMP) module. In addition, a bidding process and algorithm is developed to generate operations schedule by using the characteristics of MAS. Performance analysis (PA) module is responsible not only for evaluating the scheduling results but also for evaluating the performance of production execution. The development of an AMPCS for an automated manufacturing cell demonstrates that the integration of RFID technique and MAS in developing an agile manufacturing planning and control system can really possess the characteristics of visibility, accountability, track ability, responsiveness, and flexibility in a distributed and dynamic manufacturing system.     

RFID-enabled real-time manufacturing execution system for mass-customization production

Mass-customization production (MCP) companies must fight with shop-floor uncertainty and complexity caused by wide variety of product components. The research is motivated by a typical MCP company that has experienced inefficient scheduling due to paper-based identification and manual data collection. This paper presents an RFID-enabled real-time manufacturing execution system (RT-MES). RFID devices are deployed systematically on the shop-floor to track and trace manufacturing objects and collect real-time production data. Disturbances are identified and controlled within RT-MES. Planning and scheduling decisions are more practically and precisely made and executed. Online facilities are provided to visualize and manage real-time dynamics of shop-floor WIP (work-in-progress) items. A case study is reported in a collaborating company which manufactures large-scale and heavy-duty machineries. The efficiency and effectiveness of the proposed RT-MES are evaluated with real-life industrial data for shop-floor production management in terms of workers, machines and materials.     

基于SOA的制造执行系统技术研究*

设计了制造执行系统面向服务的多层体系结构,研究了服务构件、服务总线、业务流程标准在制造执行系统中的应用,在此基础上提出了面向服务的制造执行系统开发方法。经验证,面向服务的制造执行系统提高了车间业务的敏捷程度,有利于快速实施和推广。     

支持快速扩散制造的制造执行系统研究

结合武器装备快速变批量生产的需求,对武器装备快速扩散制造的概念和支撑理论进行了介绍;针对快速扩散制造过程中若干核心问题,提出了支持快速扩散制造的制造执行系统。介绍了制造执行系统实施的三个关键技术:支持快速扩散制造的生产计划与动态调度技术;基于资源平台的物料全生命周期管理及应用集成技术;基于工作流的制造过程自动导航与驱动技术。最后对制造执行系统的总体体系结构和功能模型展开探讨和研究。该系统在航天制造企业应用的效果显著,证明了该系统整体方案的可行性和有效性。     

Optimal cycle production of a manufacturing system subject to deterioration

A single part type, single machine, fluid model manufacturing system is considered. The machine is subject to a deterministic deterioration process, which depends on the operation rate. The objective is to minimize a long term average cost index which penalizes inventory surplus and backlog. The optimal policy determined presents some similarities with the policy conjectured optimal in the past for a similar Markovian formulation. The considered problem can be applied to several real contexts, and not just from the manufacturing domain, as briefly discussed in the paper.     

Integration of neurocomputing and system simulation for modeling continuous improvement systems in manufacturing

The use of a neural network embedded in a larger general-purpose simulation system (GPSS) simulation used to model continuous improvement systems (CIS) policies in a factory setting is described. The neural network is used to accelerate the identification of an effective CIS policy by providing a more realistic simulation framework. The interface between general simulation theory and neural network simulation is examined. Neural networks, when embedded in larger general-purpose simulations, are found to offer the potential for improving on the capabilities of those simulations, in particular manufacturing simulations for continuous improvement of production processes.     

Intelligent products: The grace experience

Product intelligence is a new industrial manufacturing control paradigm aligned with the context of cyber-physical systems and addressing the current requirements of flexibility, reconfigurability and responsiveness. This paradigm introduces benefits in terms of improvement of the entire product׳s life-cycle, and particularly the product quality and customization, aiming the customer satisfaction. This paper presents an implementation of a system of intelligent products, developed under the scope of the GRACE project, where an agent-based solution was deployed in a factory plant producing laundry washing machines. The achieved results show an increase of the production and energy efficiency, an increase of the product quality and customization, as well as a reduction of the scrap costs.     

Intelligent control system simulation of an agricultural robot

An intelligent control system for an agricultural robot which performs in an uncertain and unstructured environment was modelled as distributed, autonomous computing modules that communicate through globally accessible blackboard structures. The control architecture was implemented for a robotic harvester of melons. A CAD workstation was used to plan, model, simulate and evaluate the robot and gripper motions using 3-D, real-time animation. The intelligent control structure was verified by simulating the dynamic data flow scenarios of melon harvesting. Control algorithms were evaluated on measured melon locations.Picking time was reduced by 49% by applying the traveling salesman algorithm to define the picking sequence. Picking speeds can be increased by a continuous mode of operation. However, this decreases harvest efficiency. Therefore, an algorithm was developed to attain 100% harvest efficiency by varying the vehicle's forward speed. By comparing different motion control algorithms through animated visual simulation, the best was selected and thereby the performance improved.Journal Paper No. 13043, Agricultural Experiment Station, Purdue University, W. Lafayette, IN 47907, U.S.A. This research was supported by Grants No. US-1254-87 and US-1682-89 from BARD, the United States-Israel Binational Agricultural Research and Development Fund.     

An intelligent modeling and analysis method of manufacturing process using the first-order predicate logic

Manufacturing process refers to machining sequence from raw materials to final products. Process plan has important effects on manufacturing process. In general, process designer relies on his experience and knowledge to arrange the process plan. For a complex part, it takes long time and effort to determine process plan. In this paper, an intelligent modeling and analysis method using the first-order predicate logic is proposed to evaluate the manufacturing performance. First, the logic predicates used to represent the process plan are defined according to the machining methods, and the predicate variables are discussed in detail. Consequently, the process plan can be represented in the form of the first-order predicate logic. Second, a type of element model composed of four nodes and four links is put forward in order to construct the process model. All components in this element model are respectively explained, and the mapping relationship between element model and predicate logic is described in detail. According to engineering practices, logic inference rules are suggested and the inference process is illustrated. Hence, the manufacturing process model can be constructed. Third, the process simulation is carried out to evaluate the performance of manufacturing system by using measures such as efficiency, the machine utilization, etc. Finally, a case study is given to explain this intelligent modeling method using the first-order predicate logic.     

Achieving dynamic, multi-commander, multi-mission planning and execution

The Multi-Agent Distributed Goal Satisfaction (MADGS) system facilitates distributed mission planning and execution in complex dynamic environments with a focus on distributed goal planning and satisfaction and mixed-initiative interactions with the human user. By understanding the fundamental technical challenges faced by our commanders on and off the battlefield, we can help ease the burden of decision-making. MADGS lays the foundations for retrieving, analyzing, synthesizing, and disseminating information to commanders. In this paper, we present an overview of the MADGS architecture and discuss the key components that formed our initial prototype and testbed. Eugene Santos, Jr. received the B.S. degree in mathematics and Computer science and the M.S. degree in mathematics (specializing in numerical analysis) from Youngstown State University, Youngstown, OH, in 1985 and 1986, respectively, and the Sc.M. and Ph.D. degrees in computer science from Brown University, Providence, RI, in 1988 and 1992, respectively. He is currently a Professor of Engineering at the Thayer School of Engineering, Dartmouth College, Hanover, NH, and Director of the Distributed Information and Intelligence Analysis Group (DI²AG). Previously, he was faculty at the Air Force Institute of Technology, Wright-Patterson AFB and the University of Connecticut, Storrs, CT. He has over 130 refereed technical publications and specializes in modern statistical and probabilistic methods with applications to intelligent systems, multi-agent systems, uncertain reasoning, planning and optimization, and decision science. Most recently, he has pioneered new research on user and adversarial behavioral modeling. He is an Associate Editor for the IEEE Transactions on Systems, Man, and Cybernetics: Part B and the International Journal of Image and Graphics. Scott DeLoach is currently an Associate Professor in the Department of Computing and Information Sciences at Kansas State University. His current research interests include autonomous cooperative robotics, adaptive multiagent systems, and agent-oriented software engineering. Prior to coming to Kansas State, Dr. DeLoach spent 20 years in the US Air Force, with his last assignment being as an Assistant Professor of Computer Science and Engineering at the Air Force Institute of Technology. Dr. DeLoach received his BS in Computer Engineering from Iowa State University in 1982 and his MS and PhD in Computer Engineering from the Air Force Institute of Technology in 1987 and 1996. Michael T. Cox is a senior scientist in the Intelligent Distributing Computing Department of BBN Technologies, Cambridge, MA. Previous to this position, Dr. Cox was an assistant professor in the Department of Computer Science & Engineering at Wright State University, Dayton, Ohio, where he was the director of Wright State’s Collaboration and Cognition Laboratory. He received his Ph.D. in Computer Science from the Georgia Institute of Technology, Atlanta, in 1996 and his undergraduate from the same in 1986. From 1996 to 1998, he was a postdoctoral fellow in the Computer Science Department at Carnegie Mellon University in Pittsburgh working on the PRODIGY project. His research interests include case-based reasoning, collaborative mixed-initiative planning, intelligent agents, understanding (situation assessment), introspection, and learning. More specifically, he is interested in how goals interact with and influence these broader cognitive processes. His approach to research follows both artificial intelligence and cognitive science directions.     

A decentralized model for flow shop production with flexible transportation system

The recent advances in technology sectors often clash with traditional organizational paradigms which can limit or make difficult an efficient implementation in the real world. In this paper we show how it is possible to exploit the advantages of innovative technologies in manufacturing when these are supported by new and efficient methods for production management. More in details, we face a flow shop scheduling problem in a shoe manufacturing system in which overtaking of jobs is allowed thanks to an innovative transportation system. Overtaking means that a job can be put in waiting state and another job can surpass it, allowing the change of the scheduling sequence. Preemption is not allowed. The objective function of the problem is the minimization of the maximum lateness. We propose a decentralized model, based on multi-agent system theory, to represent the production cells of the plant and to include the potentiality offered by overtaking of jobs at decisional level. The adoption of a decentralized approach increases the system flexibility since each machine is able to solve its local scheduling problem. Adding or removing machines to the plant will not imply a change in the scheduling algorithms. The outcomes of this work are reached firstly through a formulation of the problem with three flow shop scheduling models, secondly through a comparison of the models with respect to different performance indicators. The results highlight as the decentralized approach is able to reach comparable performances with the centralized one for a relevant number of instances. Moreover sensitivity analysis shows as in the decentralized model the computational time required to solve bigger instances increases less quickly than in the case of centralized ones. Finally, simulations of the decentralized approach clarify as the correlation of the local solution procedure is effected by the number of machines of the flow shop and the coordination mechanism is effected by the number of the jobs to be scheduled.     

Dynamic schedule execution in an agent based holonic manufacturing system

The present paper deals with the negotiation based task allocation to the resources for preparing dynamic scheduling in an agent based holonic control framework. The scheduling priority is developed by Multi Objective Optimization on the basis of Ratio Analysis (MOORA) technique under Fuzzy Multi Criteria Decision Making (FMCDM) environment considering several attributes. The well-known Contract Net Protocol (CNP) is followed for the purpose of task allocation by negotiation and cooperation, where message based communication is accomplished by eXtensible markup language (XML) using J2EE. Different Document Type Definitions (DTDs) are developed for intended applications. Necessary modifications in the scheduling arising out of changes in the volume-mix are made by a distributed cooperative problem-solving algorithm to meet the demand without violating the deadline. The algorithm is implemented using HTML code in front-end with Java Server Page (JSP) through Apache Tomcat 6.02 server. It is advocated that the cooperation based teamwork coupled with higher flexibility and agility is the key to success to remain unperturbed and provide reasonably good solution in the face of disturbances and stands superior to its hierarchical counterpart.     

Intelligent distributed and supervised flow control methodology for production systems

This paper deals with the development of an intelligent distributed and supervised control approach for high-volume production systems, in which the flow of parts can be approximated by a continuous (fluid) model. The proposed approach is based on the decomposition of the production system into elementary modules in order to reduce the control design computational complexity. In this context, a two levels control structure is proposed. At the local level, a surplus-based principle is adopted to regulate the production flow for each module according to the distributed structure. The proposed control methodology decides how to adjust the production rate in order to avoid system overloading and eliminate machine starvation or blocking. In this context, the local control law is synthesized by using the Takagi–Sugeno fuzzy systems. At the high level, a supervisory controller is designed to improve the overall system performances. A supervisor provides an additive component for each local controller when the overall system performances deviate from their acceptable domains (degraded mode). This is done by combining both local and global information into a unified formalism by using aggregation operators and according to fuzzy interval representation of the desired objectives. Finally, the feasibility of the proposed methodology is validated with simulation examples.