A Web-based on-line Monitoring and Diagnosis System for Machinery of Hot Strip Mill

全文

(1)A Web-based on-line Monitoring and Diagnosis System for Machinery of Hot Strip Mill Hung-Chieh Yeh^, Chung-Nan Lee^, Song-Jau Tsai*, Chuan-Wen Chiang^, Ming-Shen Jen^, Chang-Tsun Lin*, Cheng-Yu Wei*, Shaw-Ching Chang#, Der-Lin Wang#, Kun-Hong Shieh#, and Sheng-Yang Lin* ^Department of Computer Science and Engineering, National Sun Yat-Sen University, *Steel and Aluminum Research and Development Department, China Steel Corporation, #South Area Information Division of Institute for Information Industry Kaohsiung, Taiwan As modern industrial facilities move toward high. Abstract A multi-agent based on-line monitoring and diagnosis system (MAMDS). is developed for. speed and automation, the on-line monitoring and diagnosis become important. Since a production line. distributed measuring, monitoring and diagnosis. facility. may. break. down. shutdown. will. unexpectedly,. the. machines of a hot strip mill in the China Steel. unscheduled. cause. and. Corporation. MAMDS takes the advantage of. deteriorate the quality ofproducts. The traditional. Web-based properties, it releases the restrictions of. time-based maintenance strategy does not take the. temporal and regional isolation and provides a. practical machine condition into account and only. platform-independent user interface. Through the. performs the maintenance practice at a fixed time. system, users can easily monitor and diagnose the. interval from the maintenance personnel. Therefore,. machines on remote production lines through its. accidental breakdowns unavoidably happen, when the. graphical user interfaces, such as display of machine. chosen time interval is too long; otherwise, it leads to. status, malfunction alarming, trend chart, waveform,. over-maintenance, when the chosen time interval is. and spectrum. It uses collaborative agents to reduce. too short [1].. losses. the cost of development and enhance the feature of. In order to increase industrial automation,. reuse. The system provides a real-time and dynamical. industrial monitoring and diagnosis system based on. operation circumstance and can meet the needs of. computer technology gains popularity [2-4]. In. different levels of users. Consequently, it is of great. general, monitoring the machines or supervision is. help to run machines smoothly, lower the cost of. carried out by measuring some data from some. personnel and enhance the ability of market. machines. competition.. development of on-line monitoring and diagnosis. Keywords – monitoring and diagnosis, web-based. system is prompted by the needs to increase the. system, multi-agents, plant automation. availability of machines and to protect machines from. in. the. plant. production. lines.. The. problems. Furthermore, maintenance personnels can 1.. Introduction. access the information that is provided from a.

(2) machine monitoring system to check whether these. Furthermore, they can work with each other for an. machines are running normally.. objective requested by a user. In this paper we use a. As the Internet is in widespread use, many. collaborative software agent to achieve the objective. enterprises take advantage of the Internet to extend. of MAMDS. By collaborative agents, the modification. services. regional. of MAMDS becomes simple and easy. According to. restrictions[5,6]. Therefore, the trend has brought. the objectives of the users’ requests, some designated. about changes in the manufacturing procedure. The. agents will be able to handle the requested.. without. temporal. and. gradually. In order to release the restriction of temporal and. developed for carrying out remote control and. regional isolation, MAMDS is developed as a. information retrieval on machines to be monitored. In. Web-based system too. Hence, some problems such as. general, those systems are built as traditional. exchanging information, sending/receiving messages,. client/server architecture. But such architecture has. and uploading/downing files through a standard web. some restrictions on the use and development of the. browser need to be taken care of. Furthermore, the. system, for example, a user has to often install. system has been developed over the years. It needs to. updated software on the operating platform often.. exchange and interact information among agents. Consequently, when the number of clients grows. implemented in different programming languages.. distributed. measurement. system. is. and. The remainder of the paper is organized as follows.. compatibility of application softwares on different. Section 2 presents the architecture of MAMDS.. operating platforms become tedious.. Section 3 describes the analysis and design of. gradually,. the. management. of. versions. In recent years, system developers of large. MAMDS. The issues of implementation are described. applications adopt agent technology; it simplifies. in Section 4. Demonstrations illustrating the operating. functions of the system by the cooperation of agents. production line of a hot strip mill are given in Section. [3,7-9]. Those agents in their executing environment. 5. Section 6 concludes the paper.. can work asynchronously and autonomously [10].. to be developed in Java. Web server. developed in COM. MA : Monitoring agent DA : Diagnosis agent IEA : Information Exchange agent. BA IEA operating client. UA SQL server DAS. operating client. TA. DA MA Plant Network. operating client UA : Userinterface agent TA : Tracking agent BA : Broker agent. measurement bench. Fig.1 The system architecture of MAMDS. 2. DB.

(3) 2.. setting data from each user.. The Architecture of MAMDS In order to fulfill users’ requests and requirements,. . Broker agent (BA): This agent acts as a broker for. MAMDS must complete each request and requirement. all other agents at different tiers. The main. by the autonomous, cooperative, and coordinated. responsibility of this agent is to send requests and. agents. Agents in MAMDS can be implemented in. receive process results between the application tier. different programming languages, for example, Java. and the data tier. In addition, this agent coordinates. or COM (Component Object Model). As illustrated in. all requests to solve the synchronous accessing. Fig.1, the MAMDS architecture consists of agents,. problem.. web server, database server, and measurement bench. . User interface agent (UA): In order to interact with. for operating machines of a production line. The. users smoothly, this agent provides sequences of. agents in MAMDS can be divided into three tiers: the. manipulation graphic user interfaces. This agent. application tier, the middle tier, and the data tier as. constructs user interfaces in accordance with the. shown in Fig. 2. In the following paragraphs, the tasks. model-view-controller (MVC) pattern to help to. and functions of each agent will be first described,. display and to manage visualization information.. followed by a discussion on the relationships of. Besides, it is responsible for sending a user’s. agents.. requests to servers.. . Monitoring agent (MA): It is capable of performing measurement. . on. monitored. machines. of. Tracking agent (TA): In order to respond to the. a. status and measurement information of monitored. production line in an independent way or accepting. points at any time, this agent must update the. a measurement command at any time. It performs. up-to-date. measurement independently according to different. application tier, it must track communication. measured points at different measurement modes.. between a client and a server and notify UI agent. Those measurement modes are either at constant. when visualization information is updated. Besides,. time interval or at some triggered condition. The. in order to send the up-to-date information to. measurement mode can be configured by a user.. current on-line clients, this agent also tracks. Diagnosis agent (DA): It is responsible for. communication conditions of current clients.. accepting the measurement data from MA. And. . . . information. continuously.. At. the. Information Exchange agent (IEA): This agent is. then, it compares the up-to-date measurement data. responsible. with some defined threshold values and operations. information into Java-based agents. Therefore,. to generate diagnosis results that consists of statuses,. agents implemented in Java can retrieve the same. and preliminary diagnosis messages. The defined. information provided by other agents implemented. threshold values can be modified and stored into a. in COM. Besides, this agent must ensure message. database.. passing correctly and smoothly among agents. Data acquirement and storage agent (DAS): This. implemented in Java and COM, so that the. agent is employed to store information into a. measurement data can be. database.. The. measurement. information data. consists. provided. from. of. raw MA,. post-processing analysis data provided by DA, and. 3. for. wrapping. heterogeneous. received and updated..

(4) Moreover, MAMDS operates and provides. tier to allow a user to manipulate monitoring. monitoring and diagnosis services at a plant network.. information collected from machines of a production. It consists of database server, measurement bench, and. line.. Application tier. sub sele ct Browser me. Data tier. Middle tier. UA. DAS MA. IEA MA. DB. BA. h c n e B t n e m e r u s a e M. DA MA TA. TA MA. MA MA. Fig.2 The relationships of agents in MAMDS a Web server. Database is responsible for storing. 2.2 Measurement Bench. measurement and diagnosis data and providing. The core of MAMDS is a measurement bench that. historical measurement data for pre-diagnosis. Besides,. takes all measurement information corresponding to. Web server can provide transmission of web pages. all measurement points on the machines of a. and applets to achieve files uploading/downloading. production line. The measurement bench connects to. and execute applets. Measurement bench provides. all monitored machines and collects measurement. measurement data during runtime of machines of a. information continuously during the operating time of. production line.. a production line. MAMDS measurement bench of a. Figure 2 shows the relationships of agents in. hot strip mill in the China Steel Corporation (CSC). MAMDS during user operating runtime. During. uses. runtime the monitoring agent, the diagnosis agent, the. (DIOCs)[1]. data. the. designed by CSC. MA of MAMDS is made up of the. information exchange agent, and the tracking agent. DIOCs. MA measures the signals coming from. are operating in the data tier to collect measurement. various sensors and transmits the digitized signals. information and wait for users’ requests. The broker. through the plant network to DA and DAS. The. agent is running at the middle tier to act as an. measurement bench consisting of a number of DIOCs. information exchange broker.. and hardware units is shown in Fig.3.. acquirement. and. the. storage. agent,. The user interface. agent and the tracking agent operate at the application. 4. the. Distributed which are. Input/Output data. Controllers. collection devices.

(5) Plat Network. DIOC. DIOC. DIOC. DIOC MA. Measurement Bench. Fig.3 The measurement bench for DIOC is a data collection device that is installed in the vicinity of the machines to be monitored. Its main. record. various. components. in. systematization. procedures. We follow the analysis and construction concepts. functions include: (1) Providing various kinds of signal process. of UML to analyze and design the on-line monitoring. modules for preprocessing of the measured. and diagnosis system. The use case model in UML is. signals. used to represent scenarios of system’s functions.. (2) Diagnosing of the signal property to judge whether the signals are short-circuited or open.. 3.1 The Use Case Model The use case in UML specifies actions for a. (3) Accepting commands to execute measurement.. system to interact with external actors including users. (4) Digitizing measured signals.. and the system. It describes the main functions 3.. including the system, high-level system architecture,. Analysis and Design The Unified Modeling Language (UML), which is. proposed by Rational Software Cooperation, Grady. external actors, and then interaction relationships. The use case model for MAMDS is shown in Fig.4.. Booch, James Rumbaugh, and Ivar Jacobson [11] is. The levels of authority of the system for external. widely accepted for the representation of developing. actors (users) are classified into normal monitoring. enterprises or large-locale?? software systems. It. personnel, local operator, and system administrator.. allows users to designate, represent, construct, and. MAMDS offers functions, such as system monitoring,. Multi-Agent based Monitoring and Diagnosis System. <<actor>> Measurement System. system monitoring. normal monitor ing personnel. <<actor>> Database Management System. system setting local operator system management <<uses>> identifier confirmation. system administer. Fig.4 The use case model for MAMDS 5. <<actor>> File System.

(6) system setting, system management, and identity. (2) Login checking: It determines a user’s validity. confirmation. Identity confirmation is responsible for. using the levels of authority. The levels of. verifying the identification of each user and for. authority classify the levels of users, from which. determining the levels of authority of the user. Then, it. it. decides what kinds of functions and data a user can. operations by unauthorized users that might lead. access. System monitoring provides a real-time and. to system and security problems or unnecessary. current operating status. Since some functions are. information provided by normal users. When a. only for acquiring information, it can be used for all. user completes login phase, UA must notify TA to. users. System setting includes measuring methods,. record this UA information in the data tier via BA. measurement time interval, and companion basis of. in order to display the UA updated information.. diagnosis about a measurement point. In addition,. MAMDS can transmit and. system setting provides modifying current monitored. measurement information of measurement points. points, such as adding and removing. An administrator. via TA in the data tier.. protects. the. system. from. inappropriate. notify current. can edit the users’ data and modify a user’s level of. (3) Data browsing: UA presents monitoring and. authority through system management function. The. diagnosis information in terms of waveform,. data. database. spectrum, and trend chart. UA can receive the. management function and file system function to store. messages about visualization information that is. the information of the production line, users of the. updated from TA at the application tier. It helps to. system, and other system data settings.. display the real-time measurement information to. storage. repository. is. used. by. a user by the cooperation of UA and TA. (4) File uploading and downloading: In order to. 3.2 Operational Phase Analysis MAMDS mainly uses operational phase to achieve. download the collected data from measurement. monitoring and diagnosis objectives. Operational. devices by operators, MAMDS permits uploading. phase includes monitoring and diagnosis, login. files. checking,. file. measurement results. At the same time, operators. downloading, measurement mode configuration, and. can download files that contain the historical. information updating. Functions in an operational. measurement data. This phase is completed by. phase are as follows:. the cooperation of BA and DAS.. data. browsing,. file. uploading,. or. downloading. files. that. contain. (1) Monitoring and diagnosis: MA and DA are. (5) Measurement mode configuration: In order to. activated unless all machines stop operating. MA. retrieve measurement data from monitored points. must. be. on machines of a production line, MAMDS. monitored either at constant time interval mode or. allows to configure the measurement mode that. at triggered condition mode. In addition, MA also. consists of measurement time interval and. can take measurement on behalf of a user’s. condition. After UA receives configuration. request. When MA takes measurement data, it. requests information from a user, it transmits the. transmits measurement data to DA for diagnosis.. configuration requests information to DAS via. MA transmits raw measurement data to DAS.. BA.. continuously. measure. points. to. And DA also sends the diagnosis results to DAS.. 6. (6) Information updating: MAMDS mainly executes.

(7) monitoring and diagnosis. At the same time, TA at. Therefore, MAMDS provides a platform-independent. the data tier is notified what measurement point. monitoring and diagnosis graphical user’s interface. has measurement information to be updated, and. by the characteristics of Java and function of IEA.. whether MAMDS has on-line users. Finally, TA. With information technologies that grow gradually,. transmits the up-to-date information to the. and protocol and programming languages become. front-end via BA.. more diverse, APIs of Java alone may not be enough to satisfy a wide variety of applications. For this. 4.. reason, Java provides JNI to communicate among. Issues of Implementation This section describes the implementation of. other. programming. languages.. In. order. to. MAMDS for machines. MAMDS has been developed. communicate among agents implemented in Java and. over several years. During the period of development,. COM, IEA must be responsible for information. different languages have been used for different. exchange and receiving all events for the updating. agents. For example, UA, TA, BA and IEA in. measurement information of monitored machines.. MAMDS are implemented in Java language. MA,. IEA uses the techniques provided in [10,11] to. DAS, and DA are implemented in COM. Java. handle different data type exchanges between Java. language provides several characteristics, such as. and COM to communicate and collect measurement. platform-independent as “write once, run anywhere”,. information of monitored machines among agents.. powerful network ability and distributed computing,. Second, IEA must receive all events for updating. and integration. Besides, the class files of Java. measurement information of monitored machines. program can move through the Internet easily and run. validly to protect from measurement information loss.. through any standard web browser.. However, MA is implemented in COM and fires. Inside MAMDS it has to exchange and interact. window’s. events. for. updating. information. of. information among agents either implemented in. monitored machines to a hide window. In other words,. COM or in Java. To achieve the requirements and. MA notifies updating information via a running. objectives of the system, many implementation issues,. background window. Therefore, IEA is responsible for. such as information exchange, security, and data. communicating with this background window to build. consistency must be solved. Details of each issue are. a communication bridge between MA and IEA. The. discussed in the following subsections.. steps for IEA to receive all events for updating measurement information from MA are as follows: Step 1: At the initiation, IEA and MA must. 4.1 Information Exchange In order to retrieve current and historical. activate. MA is responsible for collecting. information of monitored machines from MA and. the. measurement. information. DAS implemented in COM, IEA of MAMDS is. monitored. machines. responsible for handling information heterogeneity. However,. IEA builds. among agents. Since data type and message passing. window to receive all events.. of. continuously. a. background. are different between Java and COM, it may cause. Step 2: When MA completes a measurement task,. some problems. IEA uses Java Native Interface (JNI). MA fires an updating measurement event. to. to this background window.. implement. information. exchange. function.. 7.

(8) Step 3: After this background window receives an updating. measurement. event,. file and reducing the cost of development.. this. background window finds current running. 4.3 Data Consistency. Java Virtual Machine (JVM) and retrieves. Measurement data are stored in a database and. JNI via JVM. Afterwards, this background. accessed via DAS implemented in COM, which are. window. accessed through pointers. It is quite simple to. can. send. an. updating. maintain those common data through these pointers in. measurement message to IEA. Step 4: IEA receives an updating measurement. COM. During exchanging and updating of data, the. message and then retrieves the updated. system. only. needs. to. update. the. up-to-date. measurement information by the updating. information once. However, Java is designed for. measurement message.. simplicity in the initial stage, so it discards all pointer references. Therefore, it is hard to directly access data through pointers for agents implemented in Java. In. 4.2 Security MAMDS must be accessed from a standard web. order to maintain raw relations of data that is accessed. browser. However, a standard web browser has to. through functions of pointer, we use Java objects to. assume that an applet is not trustable in order to. represent original data object in COM. These relations. protect a host from downloading and executing an. among common data objects accessed through. unsafe Java applet. It restrains the use of a web. pointers in COM must be handled. We have to build the Java objects that correspond. browser in the following ways [12]. Accessing file system through a web browser. to COM’s data models, and translate data from. Executing a native code. COM’s data types to Java’s data types by Java Native. Object A IDispatch. COM object A. Object C IDispatch. IDispatch. COM object B. COM object C. Object B Object C Java. Fig.5 COM objects transform to Java objects Reading certain system properties. Interface (JNI). However, in order to avoid data. Receiving incoming socket connections. duplication as illustrated in Fig.5, we must construct. Based on the restraints mentioned above, the use of general applet to satisfy our requirements is. the relation of Java objects corresponding to the relation of objects in COM through pointers.. impossible. In order to satisfy the requirements of. In this situation, MAMDS must update all objects. uploading or downloading files of measurement data,. which are related to this exchanged object for. BA in MAMDS is implemented in the Java Servlet.. transforming COM objects to Java objects through. Relying. accomplish. JNI directly. Besides, data duplication may cause. uploading/downloading measurement data in a binary. more usage of capacities. In order to solve the above. on. Servlet. can. 8.

(9) IDispatch. COM object A. Object A IDispatch. IDispatch. COM object B. COM object C. Object B. Key 1 Key 2 : Key of Object : C : : Key n Hash Table Java. Object C. Fig.6 An improved data structure, which transforms a COM problems and to keep the relation of objects through pointers in COM, we reduce the usage of capacity by. 5.. Demonstration of MAMDS. rebuilding the relations of data and modifying the way. In this section, we demonstrate MAMDS of a hot. of data exchange between Java and COM. We store. strip mill currently used on the China Steel. every data as Java’s object once and establish. Corporation. Figure7 shows a main screen snapshot of. relational reference to a data hash table structure,. MAMDS. The colored rectangle in the left top corner. which uses ‘id’s in COM data types as a key to. of the screen represents the current conditions of the. retrieve the related object through the established hash. machines present in the whole production line; green. table. An improved data structure of Fig. 5 is shown in. stands for normal, yellow for alarm and red for danger.. Fig.6.. In order to attract the user’s attention, the colored. Fig.7 The snapshot of main screen of MAMDS for hot strip mill 9.

(10) rectangle keeps on blinking unless all unconfirmed. information of all measurement points in a table, as. alarms are checked. The hierarchical tree for this. shown in Fig.8. The information of each measurement. production line is displayed on the left-hand side of. point consists of name, the latest measured data. Fig.7. The hierarchical tree shows the workflow of the. (measured data and time, measured parameters,. production line and each measurement points of. measurement value, etc.), preliminary diagnosis, and. facilities. Each hierarchical node contains one icon. the types of measured data of the channel. The color. with different colors representing current status of. in the status column of each row represents the. measured points or production line. Therefore, it. condition of the latest measurement data. The. provides concise message for measurement data of. “yellow” or “red” colors represent an alarm or a. this monitored and measured points. The popup table. danger message, respectively. In order to trace alarm. lists the measurement information for an arbitrary. or danger measured signals for later use; these. node, once it is pressed. On the right-hand side of. measured. Fig.7, a clear image of the production line is shown.. automatically. The other measured signals with. When the hierarchy node representing a measurement. normal machine condition can also be saved manually.. point stops measuring, a solid line is displayed at the. When the operating time of production line. middle of a hierarchy node. At the right hand side, this. increases, the system can analyze the records of. system uses transparent buttons on a picture or an. measurement data of each measurement point and can. image to represent the workflow structure or facilities’. transform them into useful statistical data. These. locations of the production line. These pictures or. statistical data help users to observe the trend of each. images correspond to the tree node on the left hand. measurement point and the operating condition of. side. In addition, it synchronizes with the tree on the. facilities and to provide forecast information in. left hand side.. accordance with this trend Information, which is. In order to understand the current status of each. signals. stored. in. the. represented into a chart as illustrated in Fig.9.. monitored and measured point, the system lists the. Fig.8 Display of the latest measurement data. 10. are. database.

(11) Fig.9 Trend chart of a measurement point time, it provides users the operating trend of each. 6. Conclusions In this paper, we have constructed a Web-based. machine and allows operators to take precautions to. cooperative multi-agent system -- MAMDS to. prevent machines from becoming deteriorated. As a. enhance industrial automation.. It may release. result, it avoids machine damages and problems of. restrictions such as geographical isolation, and. product quality. Consequently, this system can assure. operating time. It enhances flexibility for satisfying. smooth operation of the facilities, reduction of. the changeable and multiform requirements and. damage of accidental breakdown, and maintenance of. reducing the troubles of updating the application. product quality at the desired level.. software in client sites. The system integrates software modules developed at different phases, as a result, it. Acknowledgements. also enhances the reuse of industrial software and then. This research was supported by South Area. reduces the cost of development by the collaborative. Information Division of the Institute for Information. agents. MAMDS offers the users a better operating. Industry and the Automation & Instrumentation. environment for monitoring machines of a hot strip. System Development Section, Steel & Aluminum. mill. In addition, the system assists users to perform. R&D Department of China Steel Corporation.. maintenance and solve harmful problems of machines. This system can provide real-time information of. References. monitored machines accurately and response to machine’s condition during the operating period. By. [1] S. J. Tsai, C. T. Lin, J. J. Jeng, C. Y. Wei, K. M.. the historical data recorded at every measurement. Chang, C. C. Chang, C. H. Ko, and Y. C. Chiang,. 11.

(12) “An On-line Monitoring and Diagnostic System. [9] G. P. Azevedo, B. Feijó, and M. Costa, “Control. for Machinery of CSC No.2 Hot Strip Mill,”. Centers Evolve with Agent Technology,” IEEE. China Steel Technical Report, No. 12, pp.116-126,. Computer Applications in Power, pp. 48-53, July. 1998.. 2000.. [2] B. Byman, T. Yarborough, R. S. V. Carolefeld, and. [10] http://users.rcn.com/danadler/javacom/. J. V. Gorp, “Using Distributed Power Quality. [11] http://java.sun.som. Monitoring. System. [12] J. Hunter, JavaTM Servlet Programming, O’Reilly,. Management,” IEEE Transactions on Industry. ISBN 1-56592-391-X, First Edition, October. Applications, Vol. 36, NO. 5, September /. 1998. for. Better. Electrical. October 2000.. [13] D. B. Langeand M. Oshima, Programming and. [3] H. Wangand C. Wang, “Intelligent Agents in the. Deploying Java Mobil Agents with Aglets,. Nuclear Industry,” IEEE Computer, Vol. 30, Issue. Addison-Wesley,. 11, pp. 28-31, November 1997.. printing, November 1998.. ISBN. 0-201-32582-9,. 2nd. [4] D. Lin, D. H. Zhu, F. Q. Li, and K. X. Tan, “A. [14] M. Fowler and K. Scott, UML Distilled: a brief. Distributed On-line Monitoring and Diagnosis. guide to the standard object modeling language,. System of Power Equipment,” Proceedings of the. 2nd Edition, Addison Wesley Longman, Inc. ISBN. 6th International Conference on Properties and. 957-566-755-7. Applications of Dielectric Materials, Vol. 2, 2000 [5] T. Snadholm and Q. Huai, “Nomad: Mobile Agent System for an Internal-Based Auction House,” IEEE Internet Computing, pp.80-86, April 2000. [6] G. Froehlich, H. J. Hoover, W. Liew, and P. G.. [15] S. A. Albir, UML in a nutshell, O’Reilly, ISBN 957-8247-16-8 , March 2000 [16] Z. Cui, B. Odgers, and M. Schroeder, “An In-Service System,”. Agent. Monitoring. Proceedings. of. and. the. 11. Analysis th. IEEE. Sorenson, “Application Framework Issues when. International Conference on Tools with Artificial. Evolving Business Applications For Electronic. Intelligence, pp.237-244, 1999.. Commerce,” Information Systems, Vol. 24, No. 6, pp. 457-473, August 1999.. Oriented. [7] T. Wittig, N.R. Jenningd, and E.H. Mamdani, “Archon. Framework. for. Intelligent. Co-Operation,” Intelligent Systems Eng., pp. 168-179, Autumn 1994.. Approach. Environments,”. Modeling. Proceedings of the 6. of th. Real-time. Control,”. Internet Conference on. Emerging Technologies and Factory Automation, pp. 387-390, 1997. [18] R. Schoop and R. Neubert, “Agent-Oriented. [8] G. P. Azevedo, B. Feijó, and M. Costa, “An Agent-Based. [17] H. Chebeane and F. Echalier, “Reactive Object. to. Proceedings. EMS of. Material Flow Control System Based on DCOM,”. in. Open. Proceedings of the 3rd IEEE Internal Symposium. the. IEEE. on. PowerTech, Budapest, August 1999.. Object-Oriented. Real-Time. Computing, pp.342-345, 2000.. 12. Distributed.

(13)