Extrusion Asia Edition 2-2018

36 测量技术 Extrusion Asia Edition 2/2018 相同条件下测量值的分散情况，也就是测量仪器自身的测量 值噪声特性。 仅用数值表示“可重复性“是不充分的。可能发生的是：一个供 应商标出单一值的标准偏差，而另一家供应商则依据一系列 的平均值计算这一数值。 ”绝对精度“（也称为”正确性“）和”可重复性“（也称为”精度”） 的定义通常用图形显示，见图1.A a. ）不能重复，绝对不精确 b. ）可重复，绝对精确 c. ）可重复，绝对不精确 d. ）不能重复，（平均起来）绝对精确 测量仪器的“测量速率“是每秒钟产生的测量值的数量。这是一 个更为重要的比较标准。”越多“被认为是”越好“。但是，作为 The colloquial meaning of “accuracy” is the total of all measur- ing errors. However, for the evaluation of a measuring device, it has to be differentiated: “absolute accuracy” means the com- parison of a mean measuring value with a certified standard va- lue. “Repeatability” is defined as the scattering of the measur- ing values under the same conditions and, therefore, a charac- teristic of the measuring value noise of the device itself. The sole specification of only a numerical value for “repeatabi- lity” is not sufficient. It might be that one supplier indicates the standard deviation of single values, whereas another calculates those based on a sequence of averaged values. A common visualization of the definitions “absolute accuracy” (also called “correctness”) and “repeatability” (also called “pre- cision”) is shown in . The “measurement rate” of a measuring device is the number of measurement values generated per second. This is a further important comparison criterion where “more” is seen as “bet- ter”. For an objective comparison, however, the knowledge of the interdependence between measurement rate and absolute accuracy and repeatability of a single measurement is crucial. It may be the case that a measuring device with a higher mea- surement rate, but lower single value precision is less suitable for controlling or characterization of a process than a device with a lower measuring rate but higher single value precision. For example, this is the case when a long averaging time is ne- cessary due to a lower single value precision. Then, there is a risk that actual product variations, which occur within this ave- raging time are levelled out while really present. In the worst case, the specifications might even be violated without this being signaled by the measuring device. The following example of a temperature profile taken over a week in September 2000 shows to which extent averaging of a measuring value can influence the perception ( ). The displayed “real value” results from single measurements taken in ten minute intervals. Averaging over a period of one hour only smoothes the extreme values. When averaging the varying temperature for more than 12 hours, the changes in temperature are displayed lower than they actually are. Further- more, if the mean value is generated over an entire day, the in- formation about the daily temperature variations is completely lost. A device that needs the latter averaging depth will not be suitable for a process where an alarm has to be raised or an ad- justment has to be made depending on the temperature range. A practical example taken from the hose and tube extrusion process is the diameter measurement based on the shadow projection method with rotating mirrors. Often high measuring rates are indicated, which result from the rotation rate multi- plied by the number of mirrors’ facets (Zanoni, 1973), (Voss- berg, 1981). The specification of accuracy, however, usually is based on mean values of up to one second due to a relatively poor single value precision. This has various reasons: Each single measurement is done with a different mirror facet. 图2：温度曲线的例子 （蓝色：实际值，红色：1小时平均值，黄色：每日平均值） Picture 2: Temperature profile as example 图3：SIKORA直径测量头用于衍射分析的线性传感器技术 Picture 3: Line sensor technology for diffraction analysis in a SIKORA diameter gauge head