AIDA-Glossary of Terms 专业术语-1

译文：
AIDA：先进辨识和数据分析.
Alias： 工作区内分配给位号引用的标识名.
Aliasing：离散数据采样过程中丢失或损坏的信息.
APC：先进过程控制.
ARX：外因输入模型自回归.
Bad Slice：坏区 对数据窗口中的数据指定为BAD且不用于技术计算.
Bias：偏差 在统计方面，这指的是一个统计变量的预期值和得到的值之间的偏差.
Compaction：简化 通过一组较少的有代表性的系数逼近阶跃响应（或 FIR）系数。是实现简化模型的一种方式。
Confidence Interval ：置信区间 置信区间给出了某个值可能的（根据某种概率）包含预估参数的范围。间隔的宽度预示了某个预估参数的不确定性。
Constraints： 参数限制。
Continuous Time Models：连续时间模型 在连续时间内用数学公式来表示模型。
Correlation：相关度 测量两个变量之间线性相关的程度。在自相关中是基于某个变量和该变量在一段时间后（滞后）的测量关系。在互相关中是基于两个变量在一段时间（滞后）的测量关系。
CV ：被控变量。
Data Segment：数据段 数据窗口或周期。
Dead-time：死区时间 当一个独立变量改变后，其对应的因变量没有发生响应的时间。
Deterministic：确定性 系统或数学表达式部分所包含的系统行为确定的规则所引起的因果关系和可预测行为。随机性的反义词。
Differencing：差分 是一项删除数据某些趋势的数学技术，通常用于可视化改善斜坡模型的质量。差分通过一个新的时间序列代替原有的时间序列，其中每个数据点需要减去其前一个数据。在数学上，时间序列y通过下列方程转化为一个新的时间序列z : zi = yi – yi-1 。注意在实施差分时，时间序列缩减了一个数据点。
Discrete Time Models：离散时间模型 在离散时间内用数学公式来表示模型。
Disturbance：干扰 可控制之外影响过程的输入。干扰分可测量和不可测量。
Drift：漂移 数据趋势线（时间序列）倾向于不停留在平均值周围。当某个信号不平稳时其会显示漂移。

---

原文：
AIDA Advanced Identification and Data Analysis.
Alias A nick-name assigned to a tag for referencing within the workspace.
Aliasing Loss or corruption of information during discrete data sampling.
APC Advanced Process Control.
ARX Auto-Regressive with eXogenous input model.
Bad Slice Data corresponding to a window of data designated as BAD and not to be used for technical computations.
Bias In the statistical context, this refers to the difference between the expected value and the obtained value of a statistical variable.
Compaction Approximation of step-response (or FIR) coefficients by a fewer representative set of coefficients. A way to achieve model parsimony.
Confidence Interval A confidence interval gives a range of values likely (with some probability) to include an estimated parameter. The width of the interval provides an idea of the uncertainty about a parameter estimate.
Constraints Limit on parameters.
Continuous Time Models Models represented by mathematical equations in continuous time.
Correlation A measure of the degree to which two variables are linearly related. In auto-correlation the relationship measure is between a variable and itself at a certain distance of time (lags) away. In cross-correlation the relationship measure is between a variable and a second variable at a certain distance of time (lags) away.
CV Controlled variable.
Data Segment Period or window of data.
Dead-time Duration of time for which an independent change has no effect on a dependent variable.
Deterministic Part of the system or mathematical expression comprised of definite rules of system behavior leading to cause-and-effect relationships and predictable behavior. Opposite of stochastic.
Differencing A mathematical technique for removing trends from data which often improves the visualization of the quality of ramp models. Differencing a time series replaces the original series with a new time series where each data point is subtracted from its predecessor. Mathematically, a time series y is transformed to the differenced time series z by the following equation: zi = yi – yi-1 Note that when applying differencing, the time series is shortened by one data point.
Discrete Time Models Models represented by mathematical equations in discrete time.
Disturbance An input affecting a process outside of control. Disturbances can be measured or unmeasured.
Drift The tendency of a data trend (time-series) to not stay around a mean value. When a signal is non-stationary it shows a drift.

---

2016/4/14