《外文翻译数控机床的核心及动力测量值的几何误差.doc》由会员分享，可在线阅读，更多相关《外文翻译数控机床的核心及动力测量值的几何误差.doc（10页珍藏版）》请在三一办公上搜索。
1、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： Proceedings of SPIE Vol. 4222 (2000) 0277-786X! 附 件： 1.外文资料译文；2.外文原文。 指导教师评语： 签名： 年 月 日附件1：外文资料翻译译文数控机床的核心及动力测量值的几何误差摘 要 在这篇论文中，介绍了一个用于测量数控机床上所有动力学方面几何误差的系统，并且给出了一些实验结果。这些实验结果表明了一些静态与动态误差特性间的重要差异。通过关于在动态信号运动或暂停时的分析，会发现数控机床上一些错误的方法。另外，用这个系统任何一种形态的
2、轮廓误差可以被直接地测量而不需要用到一个球型金属块或者别的什么装置，它可以提供一个简单而又实用的估计数控机床轮廓误差的方法。关键词：机床用具度量衡 几何误差 动态测量1序言 在过去的几年中，机床的校准对于机床的制造方和使用方来说变得愈加重要，每个国家都有它承认的标准去评估数控机床的性能13。然而，在这些标准中几乎所有的几何误差都是在静态或类似静态的情形下被测量出的。换句话说，机器的轴被移动到特定的位置，停止转动的时候被测量记录的。这个过程明显很耗费时间，也可能很耗费劳力。现今很多如惠普5529A激光测量系统和雷尼绍激光干涉系统的现代化测量系统提高了校准效率。这些校准包括自动收集信息，自动调调整
3、校准部分程序和机械补偿参数。这些工具可以简化校准过程，费较小的劳力。但是他们也不能克服只能在静态情况下测量的限制。很长的测量周期加购买测量器械的花费使得整个校准过程异常昂贵。如果机械很大而且可能出现误差的地方又多的话将会花费更多。众所周知，一台三轴机床有21处可能出现几何误差的地方需要测量。另外，当数控机床在工作状态加工零件时，机床的静态几何误差并不是表现出来的那么正确。机床是由机械部分、电部分和数字化部分等组成，每个组成部分都在一定的条件下起作用，比如速度，加速度，摩擦力，牵引力，能量变化等。所有因素都影响机床的正常工作状态，并且机床的动力传输都假定在静态状态下。因此，有必要评估机床的动态性
4、能并且去研究精确地测量机床动态几何误差2动态测量系统 为了评估机床的动态性能，一个比较研究被正式纳入从两个基本调查中得出的结果。首先，机床的静态 几何误差由惠普激光干涉仪测量。其次，同一台机床的动态几何误差由house-built数据采集系统同时图1 机床的动态几何测量工作示意图 采集，高转速，基于时间的测量，来比较编码器读数与激光干涉仪读数。图1显示了机床的动态几何测量误差的相关机构4。在这一动态测量系统中，使用Zygo Axiom 2/20激光干涉仪，因为它可以允许数控机床的工作台或轴的移动速度达到1 .8m/min ，它是惠普激光系统的5倍速度。同时，house-built数据收集和处理
5、系统被使用。它主要由一个Ariel DSP的接口卡与全局总线+数字接口和l6M的随机存储器，house-built控制电子，和一个使用C和Visual Basic软件的PC接口组成。在机床动态测量的实验中，模型Monarch 45数控铣床加工中心被用作对象的测试。在开始测量前，当Monarch数控铣床加工中心通电并处于一个静态的情况下，整个数据采集系统开始收集数据。测量出的背景噪声代表了测量系统本身的误差和主轴的振动。许多实验表明，当机床已适当调整过后，最大误差小于0.5微米。3 .实验结果和讨论3.1静态和动态测量值的对比 静态测量，使用惠普5528激光干涉仪，三套的双向测量是采取为每个线性
6、轴间隔50毫米。这些通过所有测量均在同一点的数据作为静态测量的结果并为参照来和动态测量的结果作比较。对于动态测量，选择不同的原料来找出是否一些动态几何误差与工作台的移动速度有关系。在下面的动态实验中，在工作台以一个稳定的速度移动后收集数据，以便消除工作台加速度的影响。事实上，图2 动态与静态线性位移测量值的比较 加速度会产生巨大的动态误差，尤其是当有一个更大的Abbe偏移和一个更大的加速度。图通常表明一些显着性差异静态和动态测量误差的线性位移。图2表明了一些静态和动态测量误差直线位移的差异。3.2噪声测量 一个动态和静态测量最明显的区别是动态测量比静态测量有更大的随机变化，尤其是在图2所示测量
7、直线位移误差。这些随机变化可称为测量噪声。噪音是由数控机床上许多不同种类的误差引起的，如滚珠丝杠和螺母节距误差，编码器读数误差，工作台倾斜引起的Abbe误差，激光测量系统误差，主轴振动等。它可被视为一个反映了机床动态特性的参数。在我们的实验中，在这台机床上以不同运动速度，从100mm/mim至4000mm/mim，做同样的实验，这超过了获得一个高定位精度的正常速度。通过所有这些实验，我们发现测量噪声与机床的进给速度几乎没有任何关系。通过FFT分析，如图3所示，我们发现测量噪声的空间频率约2.84毫米，这个现象可以进一步验证图3a是图2的一部分。它可以得出结论， 滚珠丝杠或螺母可能是测量噪声主要
8、来源。 a .动态测量噪声 b.测量噪声的FFT分析 图3 动态测量噪声和FFT分析3.3跳动误差 当工作台以一定的速度沿着x轴从0毫米位置至100毫米的位置， 停止移动一到两秒钟，然后回到0毫米的位置。以不同的速度重复几次这个过程，我们反复观察到约2微米的跳动误差。从这些实验中，也观测到跳动误差与工作台的移动速度几乎没有任何关系。这个误差是工作台初始角度重新调整的动力引起的。当工作台前后移动时，角度和直线测量的结果也显示一个在水平方向上的跳动误差。正因为导轨之间存在间隙，特别是在水平方向，动力方向引起工作台倾向相反的方向倾斜，产生Abbe跳动误差。 3.4轮廓误差 图4. 用这个系统得到的轮
9、廓误差 众所周知，循环试验提供了一个快速和高效的方式来测量机床沿圆形轮廓的轮廓精度。圆形轮廓提供一个最好的检查轮廓的表现，当机床沿圆形轨迹加工时每根轴上都有一个正弦加速度，速度和位置的变化。因此，在所有的评估数控机床的标准中，圆形轮廓测试是一个关键组成部分。球型金属块被广泛使用于这项测试。还有圆形挡板，十字网格编码器等其他设备和工具。在这里，我们使用我们的发展动态测量系统，用x和y编码器读数作为时间函数，来获得任何形状的轮廓误差。3.5数控机床的不完整运动当工作台是移动到一个理想的位置，例如，不作任何停顿地移动到x轴100mm位置然后返回，实际上工作台并没有达到100mm点。实验还表明，工作台
10、的移动速度越来越高，实际位置和理想的位置之间的偏移量也随之变得越来越大。停顿几毫秒时间的控制模式可使工作台移动到理想的位置。4.结束语使用商用激光干涉仪系统和house-built数据收集和处理系统，成功地检测出了机床的静态和动态几何误差。实验结果显示了一些静态和动态误差特性的重大差异。随着对机床误差而导致的测量噪声改进，其他一些误差被观测出来，比如因为数控机床轴来回移动而产生的跳动误差，因此有必要对数控机床的动态特性进行测量和评估。附件2：外文原文：Quick and dynamic measurements of geometric errors of CNC machinesABSTRA
11、CTIn this paper, a system that is used to measure dynamically all kinds of geometric errors of CNC machines is introduced, and some experiment results are given. The experiment results showed some significant differences between the static and dynamic error characteristics. Through analyses of the d
12、ynamic signals in both time and space fields, some error resources of the CNC machines can be found. In addition, any shape of contouring errors can be directly measured by this system without using a ball bar or other devices, which provides a simple and practical way to evaluate contouring errors
13、of CNC machines. Key Words： Machine Tool Metrology, Geometric Errors, and Dynamic Measurement1. INTRODUCTIONCalibration of machine tools has become increasingly important for both machine tool builders and users over the past few years, and each country has its own standard to evaluate the performan
14、ce of CNC machines 13 However, in these standards almost all the geometric errors are measured when CNC machines are in a static state or a quasi-static state. That is, the machine axis is moved to target positions, stopped and a measurement recorded. This process is obviously very time-consuming, a
15、nd may be labor-intensive. Nowadays, many modern measurement systems, such as HP 5529A laser measurement system and Renishaw laser interferometer system, provide features to improve the efficiency of calibration. These include automatic data collection facilities, and automatic generation and transf
16、er of the calibration part program and machine compensation parameters. These facilities can streamline the calibration process and make it less laborious, but they do not overcome the inherent time-intensive nature of the static calibration method. Long measurement periods, together with the relati
17、vely high capital cost of the measurement equipment and machine, can make the whole calibration process costly. The problem is obviously compounded if the machine is large and many error components are to be measured. It is well known that for a three-axis machine tool, there are 21 potential geomet
18、ric errors that could be measured. In addition, parts are machined when CNC machines are in a dynamic state, measuring static geometric errors of machine tools does not totally represent the performance of them. As machine tools are a synthesis of mechanical, electrical, digital components etc., eac
19、h of these components reacts in a time-dependent way, such that, velocities, accelerations, frictional forces, drive forces, power source variations etc. All these influence the active state of the machine and the dynamic performance of the machine is hypothesized to be sufficiently distinct from th
20、e static In Process Control and Inspection for Industry, Shulian Zhang, Wei Gao, Editors, Proceedings of SPIE Vol. 4222 (2000) 0277-786X!OO/$1 5.00performance. So it is necessary to evaluate the dynamic performance of machine tools and to study the methods of measuring the dynamic geometric errors o
21、f machine tools for precision applications.2. DYNAMIC MEASUREMENT SYSTEM In order to evaluate the dynamic performance of machine tools, a comparative study is formalized incorporating the results from two basic inquiries. First, static geometric errors of the machine tool are ascertained using a Hew
22、lett Packard laser interferometer. Second, dynamic geometric errors of the same machine tool are measured by using a house-built data acquisition system capable of simultaneous, high-speed, time-based measurements in order to compare encoder readings with laser interferometer readings. Fig.1 shows t
23、he setup for dynamically measuring the geometric errors of machine tools4. In this dynamic measurement system, a Zygo Axiom 2/20 laser interferometer is used as it can permit the table or the axis of a CNC machine to move at a speed of 1 .8m/min, which is 5 times faster than a HP laser system. At th
24、e same time, a house-built data collecting and processing system is used. It mainly consists of an Ariel DSP interface card with a GlobalBus+digital interface and l6Mwords of RAM, house-built control electronics, and a developed PC interface using C and Visual Basic software. In the experiment of dy
25、namic machine tool measurement, the model Monarch 45 CNC milling center is used as the object of testing. Before starting measurement, the whole data acquisition system starts to collect data with the Monarch CNC milling machine center table being in a static status and with power on. The measuremen
26、t background noise represents the errors of the measurement system itself and the vibration of the spindle. Many experiments showed that the maximum error is less than half a micrometer when the Z driver has been adjusted properly.3. EXPERIMENT RESULTS AND DISCUSSIONS3.1 Comparison between static an
27、d dynamic measurementsFor the static measurements, three sets of bi-directional measurements are taken for each linear axis at an interval of 50mm by using the HP 5528 A laser interferometer. The data through averaging all measurements at the same point is used as the result for the static measureme
28、nt and as a reference to compare with the dynamic measurement results. For the dynamic measurements, different feed rates are chosen to find out whether or not some dynamic geometric errors have a relationship with the speed of the moving table. In the following dynamic experiments, the data are col
29、lected after the table moves at a stable speed to eliminate the influence of the acceleration of the moving table. In fact, acceleration can produce great dynamic errors for CNC machines, especially when there is a greater Abbe offset and a larger acceleration. Fig.2 typically shows some significant
30、 differences between static and dynamic errors for measuring the linear displacement. 3.2 Measurement Noise One of the most obvious differences between dynamic and static measurement is that dynamic measurement has much larger random variations than static measurement, especially in the measurement
31、of linear displacement errors shown in Fig.2. These random variations may be called measurement noise. The noise is produced by many different kinds of errors in CNC machines, such as, the pitch errors of the ballscrew and the nut, encoder reading errors, Abbe error variations caused by the table ti
32、lt, laser measurement system errors, vibration of the spindle etc. It may be considered as one of parameters that reflect the dynamic characteristics of machine tools. In our case, the same kind of experiment is done with different moving speed ranging largely from 100mm/min to 4000 mm/min that exce
33、eds the normal speed of getting a good position accuracy in this machine. Through all these experiments, it is found that measurement noise nearly has nothing to do with the feed rate of the machine. Through a FFT analysis, it is found that measurement noise has a spatial frequency of about 2.84mm s
34、hown in Fig.3b, and this phenomena can be further verified in Fig.3a that is one part of Fig.2. It can be concluded that the ballscrew or the nut is likely the main source of the measurement noise.3.3 Jump ErrorsWhen the table is moving along the x-axis at a certain speed from 0mm position to 100mm
35、position, it stops for one or two seconds and then is traveling back to 0mm position. This procedure with different moving speeds is repeated several times and an obvious jump error of about two micrometers is observed repeatedly. From these experiments, it is also observed that the jump error has n
36、early nothing to do with the speed of the moving table. This error is mainly caused by the initial table angular readjustment induced by dynamic forces. The results of both angular and straightness measurements also show obviously an angular jump in horizontal direction at the return point when the
37、table moves forth and back. As there is a gap between the slides, especially in horizontal direction, the direction of the dynamic force induces a table tilt in the opposite direction and produces an Abbe error jump.3.4 Contouring errorsIt is well known that circular tests provide a rapid and effici
38、ent way of measuring a machine tools contouring accuracy along a circular contour. Circular contours provide one of the best checks for contouring performance in that as a machine is traversing with multiple axes along a circular trajectory each axis goes through sinusoidal acceleration, velocity, a
39、nd position changes. So in all standards for evaluating CNC machines, circular contour test is a key part. A Ball bar is widely used for this purpose. Other devices and instruments include circular masks, cross grid encoders. Here, we use our developed dynamic measurement system to get any shape of
40、contour errors by using the direct x and y encoder readings as a function of time. Fig.4 is an example of the circular contouring error. 3.5 The imperfect movement of CNC machinesWhen the table is driven to a desired position, for example, to the 100mm position along the x direction and then back wi
41、thout a dwell time, the table actually does not reach the point of 100mm. The experiments also showed that offset between the actual position and the ideal position get bigger when the speed of moving table is getting higher. A control mode of dwell time with a few mil-seconds can make the table be
42、in the desired position. 4. CONCLUSIONSStatic and dynamic machine tool geometric errors were performed using a commercially available laser interferometer system and a house-built data collecting and processing system. The experiment results showed some significant differences between the static and
43、 dynamic error characteristics. Along with the expected increase in measurement noise, which results from many error resources of machine tools, some other errors like jump error from back and forth movements of CNC machine axis are observed, which makes it necessary to measure and evaluate the dynamic characteristics of machine tools.
