Fits~ Overview –Options-1

你也可以指定原始采样周期的特定重采样倍数对数据进行重采样。例如，若工作区采样周期为15秒，而在这一拟合中你想要一分钟的采样周期，则可将重采样倍数设为4。重采样可通过在AIDAPro引擎中使用抗锯齿Butterworth重采样算法实现。
而如果输入变量和输出变量的采样周期存在差异，重采样也是非常重要的。例如，如果你的温度和压力数据的采样周期为1分钟，而分析仪的质量测量采样周期为10分钟。你想要将温度和压力数据作为输入变量，质量数据作为输出变量来拟合一个模型。当你研究分析仪输出数据趋势时，由于分析仪的采样和保持机制，你将看到每10分钟出现一个测量峰值，以及一条水平线。水平线并不能反映质量的实际动态过程，你只有将每10分钟的输出数据和每10分钟的输入数据进行对应处理。在这种情况下，你需要将重采样倍数因子设置为等于分析仪采样周期（10分钟）除以输入测量（1分钟）的采样周期，以保持输入和输出数据关系的一致性。

数据段窗口字段
Resample Multiple 以工作区周期的倍数周期对数据进行重新采样。
Use（使用） 选中需使用的数据段。
Name（名称） 工作区数据段的名称。
Description（描述） 工作区数据段的描述。
Start Time（起始时间） 数据段的起始时间。
Stop Time（终止时间） 数据段的结束时间。
选项
可以为拟合更改和设置的某些选项。
适用于整个Fit的选项：
•Closed-Loop Data（闭环数据）：此复选框告诉AIDAPro正在使用的数据是闭环收集的。 这意味着拟合独立变量（输入）可能不是完全独立的，且可能存在一定的相关性。 若选中此选项，AIDAPro引擎将在拟合时使用ARX（自回归外生）方法。 ARX方法试图根据在自变量和输出间干扰的每个输出的未建模部分寻找模型。一般地，如果你有闭环数据（闭环可以包括在阶跃测试期间操作员的恒定干预！），这种方法预期将得到更准确的模型。默认情况下，当选中闭环数据复选框时，drifting flag（漂移标签）将关闭。在FIR Fit选项卡窗口中，settling time（设定时间）也设置为45分钟。

---

原文：
You can also specify that the data be resampled at a particular resample multiple of the original sampling period. For example, if the workspace sampling period is 15 seconds, and you want a sample period of a minute for this fit, use a resample multiple of 4. Resampling is implemented using an anti-aliasing Butterworth resampling algorithm in the AIDAPro engine.
Resampling is also important when you have discrepancies in the sampling period of your input variables and your output variables. For example, you have temperature and pressure data with sampling period of 1 minute and quality measurement from an analyzer that has a sampling period of 10 minutes. You want to fit a model between the temperature and pressure data as input variables and the quality as an output variable. When you trend the analyzer output data, you will see a measurement peak of every 10 minutes and a flat line in between as a result of the sample and hold mechanism of the analyzer. The flat lines do not resemble the actual dynamics of the quality, you only have output data every 10 minutes that correspond to the input data of every 10 minutes. In that case, you will need to set the resample multiple factor equal to the analyzer sampling period (10 minutes) divided over the sampling period of your input measurements (1 minute) to maintain the consistency of your input and output data relationships.
Segments Window Fields
Resample Multiple Resample data at a period that is a multiple of the workspace period.
UseSelect the segment to be used
**Name **Name of the workspace segments.
**Description **Description of the workspace segments.
Start Time Starting time of the segment.
Stop Time Ending time of the segment.
Options
Certain options can be set and changed for the Fit.
Options applicable to the entire Fit:
•Closed-Loop Data: This check-box tells AIDAPro that the data being used was collected in the closed loop. This means the fit independents(inputs) may not be entirely independent and some correlation may exist. If this flag is checked, the AIDAPro engine uses the ARX (Auto-Regressive exogenous ) approach during fitting. The ARX approach tries to find a pattern in the unmodeled portion of each output during between the independents and the output due to disturbances. In general, if you have closed-loop data (closed-loop could include constant operator intervention during a step test too!) this approach is expected to result in more accurate models. By default, the drifting flag is turned off when the Closed-Loop Data check box is checked. The settling time is set to 45 minutes as well in the FIR Fit Tab window.

---

2016.11.2