《外文翻译分布式机床的设计.doc》由会员分享，可在线阅读，更多相关《外文翻译分布式机床的设计.doc（15页珍藏版）》请在三一办公上搜索。
1、 南京理工大学毕业设计(论文)外文资料翻译系部： 机械工程系 专 业： 机械工程及自动化 姓 名： 学 号： 外文出处： Design of the Distributed Architecture of a Machine-tool 附 件： 1.外文资料翻译译文；2.外文原文。指导教师评语： 签名： 年 月 日分布式机床的设计FIP现场总线的用途Daping SONG, Thierry DIVOUX，费朗西斯勒帕热自动化中心研究所的Nancy摘要：本文中我们基于FIP现场总线上提出了一种分布式控制系统。它将取代传统的CNC(计算机数字控制装置)用于机床上。该系统是由一套以微处理机为基础的模
2、块(PC机、运动控制器、I/O接口) 利用FLP实时网络相互联接的。这主要是使每个模块智能化以提高整个系统的灵活性和容错能力。每个模块都是一个分控系统，用于实现自己的分控任务，其中有些模块用于运动控制，另一些模块用于传感器评价和执行器调节。FIP决定了这些模块之间的通讯（信息交流和同步），同时执行任务分配以及设备布局分布。我们讨论一些分布标准并描述实验的执行。1.引言近几年，一直对分布式体系结构进行了许多研究。分布式体系结构在系统集成上发挥主要作用。在机床控制域，目前CNC技术有它内在的缺点。将几根固定数量的轴容入CIM环境中是非常费时，灵活和不易的。超大规模集成电路微处理器技术和通信网络的迅
3、速发展使分布式控制成为可能。虽然逐步扩展没有完全替代硬件更换但分布式控制系统的性能，模块化，完整性和可靠性正在提高。它为替代控制系统架构提供了一个很好的前景。本文致力于对分布式机床结构的研究。它建立在智能设备与通信相联系的基础上。分布式机床的特点是分布式任务和分布式数据，且具有独特的控制方法。它是结合标准设备和FIP系统总线设计而成，通过实验证明该系统具有可执行性，在实验中该系统控制了复合轴系，成功执行坐标之间的关系同时也反应了对传感器值的变化。该论文结构如下：第2部分描述了机床控制系统构架。第3部分简要介绍了FIP 现场总线。第4部分概述了我们实验的实施。最后，我们在第5部分总结了一些一般性
4、意见和今后的研究前景。2.机床控制系统架构该机床控制系统是一个实时多任务系统。其功能结构如图1所示。它包括三种单元：用户接口/ 监控单元/规划单元，伺服单元，传感器/制动器单位。这个系统的主要功能是用来控制工件的加工。它包括两个不同的和相关的任务： 为了确保轨迹的准确性和对机床移动部件的速度控制 为了调查定位（跟踪）过程的正确执行，环境变化的影响与指定操作的执行或机床机件的运动同样重要。例如：工具开关，冷却，润滑等。 CAM的制造日期工作规划/编程（轨迹，工具的选择，其他加工参数采集）基本替换计算传感器/制动器环境自动安全监测加工轴伺服控制系统 机床构件图1：架构功能按时间顺序，这项任务也可分
5、为两个步骤：规划控制程序规划和执行控制程序。第一步，机床机件没有直接的方向，只有运动和被指定执行的操作，这是“数据采集和预处理”的一步。虽然在第二步，是控制的有效执行。值得指出的是：在第二步由于多任务的性质，并行处理是可行的。3.FIP 现场总线FIP系统被用来满足分布式机床上实时通信的需要。在这一节中，我们简要地解释一下FIP系统的技术性能。FIP（工厂仪表协议）是网络系统用于传感器的驱动器和控制设备如PLCS，CNCS或机器人控制器之间的信息交换。FIP系统的结构采用所谓的密封性以减少OSI模型(物理层,数据链路层和应用层)，这种结构使实时通信和常规的信息沟通之间有明显的区别。在数据链路层
6、中，相关服务一方面与其他传递信息服务可变转让。在应用层中，我们可区分MPS服务（制造周期/非制造性规范），它采用来自数据链路层的信息设备所支持的MMS设备。FIP支持两个传输媒体：屏蔽双绞线和光纤。FIP允许各种各样的布局，最长部分可达500米，至少有4个部分被中继器代替。3种比特率被确定为：31.25k.比特/秒，1兆位/秒和2.5兆位/秒。FIP介质访问控制是集中的。所有转让都由the Bus Arbiter控制，时间安排转移必须遵守时间要求。变量和信息之间的传递可采用定期配置或根据站的要求来转让，而在我们的应用中，FIP只采用可变转让。FIP采用生产者和消费者的模样来产生可变交流。变量对
7、于生产者和消费者而言，是被确定的一个独特的识别标志，一套制作和消费变量可以集结在一个站，但是这些识别标志不涉及任何物理地址站。图2显示了变化信息。 广播的BA标识符 认识的人 P和一些消费者生产者发出的日期 P 所有消费者接受的数据 C 图2：FIP系统的MAC图象首先，the Bus Arbiter 广播持有可变的标示符，所有的节点接受帧并检查变数是生产还是消费产生的或不给予影响。第三步：作为生产者的站必须响应包含数据的帧。第四步：获取消费者的变化价值并存储。当更新产生时，消费者和生产者便形成了。FIP有两种类型的数据交流：周期性和非周期性的。在这两种情况下，汇率发生情况如上图（图2）。在第
8、一种情况下，the Bus Arbiter根据从配置要求价值相应的标识定期转移。第二种情况下，the Bus Arbiter 可根据现有的带宽产生转让请求信号。在我们的应用中，实时的限制是非常严格的。为了使机床遵守给定的轨迹，轴的控制必须同步。这就要求和网络连接的控制节点应该同时接受开始命令，因此网络必须播出命令。为了确保相同的瞬时命令能同时被几个接受器接受，稳定的传输是非常必要的。因此，一些传感器例如运动控制传感器就应该要求定期调查限位开关以使网络能定期无重大延误的传输数据。一句话，像分布式机床的操作，像数据广播的要求，时间和空间的一致性，定期传输不能满足任何一般用途的网络。然而，实时网络例
9、如FIP就是数据一种好的解决方法。4.实验实施如图3所示，我们的应用目标是要实现一个分布式的两轴机床控制系统。它由以下设备分布在FIP总线的四个节点上。节点1：微机（i80486 微处理器）。它作为运营商终端。节点2.3：两个相同的节点。每个均由微机（i80486）配备了运动控制器(克莱斯勒PCIOO + 克莱斯勒三菱商事100)。节点4：一个带有传感器/制动器作为辅助业务的可编程控制器（低温100）。网络：FIP和1比特/秒的双绞线介质间的选择。软件架构的执行系统是基于概念的多层次分布式控制。它有三种层次结构，其中第二和第三层次可实现分配。它包括以下层次：分析层：控制任务的执行选择它被映射到
10、微机的提供用户界面的节点上。用来处理计划收购和储存，不同业务模式（手动和自动模式）的交换，起始点和终点以及其它各节点之间的计算和发送。惯例层：确定某一任务的控制算法它被映射到2种其他微机上（节点2和节点3）。这两种微机具有根据给定的参数和命令（轨迹类型，速度，加速度等）进行基本位移计算（插补）的功能。每个轴的插补算法是软件设计的困难之一，因为轴控制分布后，每个中间坐标轴的计算是独立的。正确的算法设计可保证这些轴的连贯性。工艺层：执行控制它包含两种运动：运动控制器和可编程控制器。这些设备执行伺服系统运动控制，处理加工件的举行/紧缩政策，传感器的评定和驱动器的调节使工具切换任务和监控系统更安全。为
11、了验证拟议的架构是否与时间限制和网络能力相适应，预期流量的估计是必要的。主要有两种性质的信息交流： 命令从中央决定站（节点）传到其它站。 统计信息由站（节点）与站之间产生。例如:在我们的实验平台上,一些变数分布如下：FIP系统 节点1 节点2 节点4 节点3 FIP系统 FIP系统 运动控制 运动控制 传感器和制动器 X轴 Y轴 图3:硬件执行5.结论本文中为满足CIM的要求，我们的研究通过实验实施进一步达到审定。我们现在正致力于用来证明符合执行实时限制的经营架构的仿真和性能分析的工作。我们的目标不仅是一个试样样机，更是研究设计、优化的分布式系统理论方案的发展。Design of the Di
12、stributed Architecture of a Machine-toolUsing FIP FieldbusDaping SONG, Thierry DIVOUX, Francis LEPAGECentre de Recherche en Automatique de NancyUniversite de Nancy I, BP239, 54506 Vandoeuvre-les-Nancy cedex, FranceAbstract: In this paper we propose a distributed control system based on FIP fieldbus.
13、 It is applied to machine-tool as a replacement for the traditional CNC (Computerized Numerical Controller). The system is composed of a set of microprocessor-based modules (PCs, motion controllers, I/OS, . .) interconnected by FLP real-time network. The main idea is to enable each module to be inte
14、lligent, improving thus the flexibility and the fault tolerant capability of the whole system. Each module being a sub-control system, accomplishes its own control task, some of them for motion control and others for evaluating sensors and regulating actuators. The communication (information exchang
15、es and synchronization) among these modules is ensured by FLP. This system allows both task distribution as well as equtpment topological distribution. We discuss some distribution criteria and describe an experimental implementation.1. IntroductionDistributed system architecture has been the subjec
16、t of many research activities in recent years. It plays a major role in systems integration. In the machine-tool control domain, present CNC technology has its inherent shortcomings. It is centralized, limited to a fixed number of axis time-consuming, inflexible and difficult to be integrated in CIM
17、 environment. The rapid development of VLSI microprocessor technology and communication network enables the distributed control to be considered. Distributed control systems present the advantage of improving performance, modularity, integrity and reliability while allowing incremental expansion wit
18、hout complete hardware replacement. It offers a promising alternative to control system architecture.This paper is dedicated to study a distributed machine-tool architecture. It is based on intelligent devices interconnected on communication link. It is characterized by distributed tasks and distrib
19、uted data, but with unique control access system. It is designed by using standard devices and FIP fieldbus and verified by a experimental implementation, in which the system controls a multi-axis machine to successfully execute a coordinated motion as well as to respond to sensors values changes.Th
20、e paper is organized as follows. In section 2, the machine-tool control system architecture is described. Section 2 gives a brief description of FIP and Section 3 outlines our experimental implementation. We conclude in section 4 with some general remarks and future research perspectives.2. Machine-
21、tool control system architectureThe machine-tool control system is a real-time and multitask system. Its classical functional architecture is shown in Fig.1. It consists of three units: user interface/supervisiou/programming unit, servo unit, and sensors/actuators unit. The main mission of this syst
22、em is to control workpart machining. It includes two different and related tasks aspects: to ensure the precise trajectory and speed control of the mobile organs of machine-tool.to survey the correct execution of this positioning (tracking) process, to react on environment changes as well as to perf
23、orm the specified operations or actions upon machine-tool mechanics. such as tool switching, cooling, lubricating, etc. Fig. 1 Functional architectureChronologically, this mission is also divided into two steps: control program planning and control program executiug. In the first step, there is no d
24、irect action on machine-tool multitask nature: “data acquisition and preprocessing” step. While in the second step, the control is effectively executed. It is worth to note that in the second step, the parallelization is possible due to the mechanics, only the motions as well as the operations to be
25、 performed are specified. This is the “data acquisition and preprocessing” step. While in the second step, the control is effectively executed. It is worth to note that in the second step, the parallelization is possible due to themultitask nature.3.FIP fieldbusTo meet the real-time communication ne
26、ed in our distributed machine-tool, FIP is adopted. In this section, we briefly explain the main technical properties of FIP.FIP (Factory Instrumentation Protocol) is an industrial network designed for the exchange of information between sensors, actuators and control devices such as PLCs, CNCs or r
27、obotcontrollers. The architecture of FIP follows the so-calkd reduced OS1 model (Physical layer, Data link layer and Application layer). This architecture makes a clear distinction between real-time communication and conventional message communication. At Data Link layer, there are services associat
28、ed to variable transfers on the one hand and conventional messaging services on the other hand. At Application layer, we distinguish the MPS (Manufacturing Periodic/aperiodic Specification) services which use variable transfers of Data Link layer from the MMS services which are supported by the mess
29、aging services of the Data Link layer.FlP supports two transmission media: shielded twisted pair and optical fiber. It allows for a wide variety of topologies. The maximum length of a segment is 500 m ;and at most 4 segmentsare authorized with repeaters. Three bit rates have been defined: 31.25 K.bi
30、t/s, 1 Mbit/s and 2.5 Mbit/s.FIP medium access control is centralized. All transfers are under control of the Bus Arbiter that schedules transfers to comply with timing requirements. Transfers of variables and messages may take place periodically according to system configuration or aperiodicalIy un
31、der request from any station. In our application, only variable transfer of FIP is used.For variable exchanges, FIP uses the producer-consumer model. Variables are identified by a unique identifier known from the producer and the consumers. A set of produced and consumed variables can be regrouped i
32、n one station, but the identifier is not related to any physical address of stations. Fig. 2 shows the broadcast of a variable.Fig2 Principle of MAC protocol of FIPFirst, the Bus Arbiter broadcasts a frame that holds the identifier of the variable. All nodes receive the frame and check whether they
33、are producer or consumer of the variable or not concerned. In a third step, the station that recognizes itself as the producer replies with a response frame that contains the data. In a fourth step, all the consumers of this variable capture the value and store it. The consumers and the producer are
34、 formed when the update takes place. FlP defines two types of data exchanges: periodic and aperiodic. In both cases, the exchange takes place as indicated above (Fig. 6). In the first case, the Bus Arbiter knows from the configuration that it has to request periodically the transfer of the value cor
35、responding to an identifier. In the second case, transfer requests are signaled to the Bus Arbiter that will serve them according to the available bandwidth.For our application, the real-time constraints are very stringent. To make the machine-tool to follow an accurate trajectory, the control of th
36、e axis must be synchronized. This requires that the control nodes connected by a network should simultaneously receive the starting order, so the network should be able to broadcast orders. To ensure that an order of the same instant is received by several receivers, a spacec onsistency statue is al
37、so necessary. For responsiveness reason, some sensors like movement-limit switches should be polled periodically requiring that the network be able to transmit periodic data without important delays.In one word, for an application like distributed machine-tool, the requirements like broadcast of dat
38、a , the time and space consistencies, the periodic transmission can not be met by any general-purpose networks, a real-time network like FIP is then a good solution.4. Experimental implementationAs shown in Fig. 3, our application is aiming to realize a distributed two-axis machine-tool control syst
39、em. It is composed of the following devices distributed on four nodes over FIP fieldbus:node1: a microcomputer (i80486 microprocessor). It is used as operator terminal,node 2. 3: two identical nodes. Each consists of a microcomputer (i80486) equipped with a motion controller (DCX PClOO+DCX MC 100).n
40、ode 4: a PLC (LT 100) with sensors/actuators for auxiliary work: FIP with 1Mbits/s over twisted pair medium is chosen.The software architecture of the implemented system is based on the concept of multilayered distributed control. It has a three-level hierarchy and the distribution is realizes at th
41、e second and the third levels. It consists the following layers:Analysis layer: performs selection of the control tasksIt is mapped on to the microcomputer of node 1 which provides an user interface. It deals with the program acquisition and storage, switches the different operational modes (manual
42、and automatic modes), computes and sends the start and arrival points coordinates as well as other orders to each other node respectively.Rule layer: determines the control algorithms for a given task.It is mapped on to the two other microcomputers (node 2 and 3) which function is elementary displac
43、ements calculation (by interpolation) according to the given parameters and orders ( trajectory type, speed, acceleration,. etc.).One of the software design difficulties is the interpolation algorithms for each axis. Because after axis control distribution, the calculation of the intermediate coordi
44、nates of each axis becomes independent, the coherence of these axis should be ensured by correct algorithm design.Process layer: executes the control.It includes the two motion controllers and the PLC. These devices executes servo systemmotion control, handles the workpart holding/tightening, tools
45、switch tasks and monitors system safety by evaluation of sensors and regulation of actuators.In order to verify if the proposed architecture is suitable with time constraints and network capacity, in is necessary to estimate the expected traffic.There are mainly two natures of information exchanges:
46、orders from central decision station (node 1) to other stations.state information produced by the stations ( node1) to other stations.For example concerning our experimental platform, we have delined some variables distributed as following: Fig3 Hardware implementation5. Conclusionln this paper, we investigated a distributed machine-tool architecture in order to meet CIM requirements. Our rese
