A2.2 Control Terminology and Functionality-2

压实点 - Exasmoc只在某些预定点计算控制器动作，而不是将来的每一个时间点。这些预定点在控制器的离线设计期间被指定，称为压实点。必须为输入（控制动作，即Exasmoc MV）和输出（CV's）指定压缩点，并且可以选择不同的压实点。压实点用于减少控制动作的计算。
**响应模型 **- 这是由Exasmoc控制器通过一系列操作计算得出的实际过程动态模型。Exasmoc控制器通过参考此模型来预测控制动作。
换句话说，响应模型将MV（即设定值或直接的阀位）的变化与POV（即中间和被控变量）建立相关。
可测量干扰模型 -假设该扰动将来保持不变的前提下，根据历史测量干扰（例如DV的进料速率变化）计算的模型。
它将前馈控制补充到响应模型中预测。
**不可测量的干扰模型（或“观察者”模型） **- 当存在不可测量干扰的影响时，这种模型是有用的。该模型包括 -

• 随机/噪声模型
• 中间变量干扰
• 用于集成过程的模型组件

不可预测的POV变化通过采用预测（基于卡尔曼滤波）来为响应模型提供另一个补充控制。
如果MV和DV不改变动作，那么可测量干扰模型不会预测任何控制; 因此，不可测量的干扰模型是唯一可以预测对意外情况控制的模型。
这可以通过使用快速响应的中间变量来实现，该变量提供被控变量的未来行为的早期指示，可以被包含在控制器中（作为不可测量的干扰模型），以允许更加鲁棒和及时的控制动作。
所有三个模型都在SMOC Graphical Model Builder的窗口中构建。
以下图示说明了上述术语和功能。

Fig. A 2.2 A generic Depropaniser process flow scheme with Exasmoc controller 图A 2.2一个具有Exasmoc控制器的通用Depropaniser（脱丙烷塔）工艺流程图

CV - 丙烷和丁烷质量
MV - FC的回流和再沸器设定值
DV - LPG进料速率
POV - CV和中间变量（中间变量是顶部温度和风机叶片）

Fig. A 2.3 MV, Model & POV (Intermediate & Controlled variables) 图A 2.3 MV，Model＆POV（中间和被控变量）

Fig. A 2.4 Exasmoc models – A simplified overview 图A2.4 Exasmoc模型 - 简要概述

---

原文：
**Compaction point **- Exasmoc calculates the controller action not at every point in the future time, but only at some predefined points. These predefined points are specified during the off-line design of the controller and called compaction points. Compaction points must be specified for the inputs (control actions i.e. Exasmoc MV’s) and the outputs (CV’s,) and can be selected differently. Compaction points are used to reduce the computations of the control actions.
**Action Model **– This is the actual dynamic model of the process calculated from the series of manipulations of the Exasmoc controller. Exasmoc Controller predicts control action by referring to this model.
In other words Action Model relates changes in MV’s (i.e. setpoints or direct valve positions) to POV’s (i.e. Intermediate and Controlled Variables).
**Measured Disturbance Model **– This model is calculated from the measured disturbance from the past (e.g. feed rate change which is actual DV) assuming that this disturbance will remain constant in future.
It supplements feed forward control to the predictions of the Action Model.
**Unmeasured Disturbance Model (or“Observer”Model) **–This type of model is useful when effects of those disturbances that cannot be measured are present. This model includes –
-Stochastic/Noise Model
-Intermediate Variables Disturbances
-Model component for integrating process
Unexpected variations in POV’s are used to adopt predictions (based on Kalman filtering) to provide another supplementary control to Action Model.
If MV and DV do not change Action and Measured Disturbance model will not predict any control; therefore it is the only unmeasured disturbance model, which predict control action to unexpected circumstance.
This can be achieved by making use of fast responding intermediate variables, that provide an early indication of future behavior of the Controlled Variables, can be included in controller (as unmeasured disturbance model) to allow more robust and timely control action.
All three models are constructed in same window of SMOC Graphical Model Builder.
Following figures provide diagrammatic explanation of above terminology and functionality.
CV's - Propane & Butane Quality
MV's - Reboiler & Reflux Setpoints of FC
DV - LPG feed rate
POV's - CV's and Intermediate Variables (Intermediate Variables are Top Temperature & Fan Pitch)

---

2017.3.26