案例2：反应器（加氢装置）质量控制-2

通常情况下，此过程的SMOCPro控制器将控制动作写入两个PID控制器的设定值（炉的TC和急冷的FC）。为了使SMOCPro识别炉出口温度变化以及反应器瞬时效应的变化，在SMOCPro模型中添加温度控制PV作为中间变量。

Figure -3 SMOCPro control.
图3： SMOCPro控制
该过程的模型为：

Figure -4 Reactor model with PID block.
图4：反应器模型PID功能块
在这里炉出口温度控制器是明确建模的。鉴于SMOCPro中的模型仅表示MVs对CVs的影响（针对预测目的），对于相同流程的等效模型是：

Figure -5 Equivalent Reactor model with closed loop dynamics.
图5：闭环动态等效反应器模型
炉温的SP和PV的传递函数是温度控制回路闭环动态行为。在ClosedDyn_FOT块左上角的蓝色三角形表示用于PID/串级回路动态模型的设置。
在两种情况下，PID控制器将消化温度PV的不可测量扰动。否则SMOCPro将通过改变PID设定点的方式尝试消化干扰，因为这样PID将补偿温度PV的变化。这将导致PID设定点不必要的动作，以及整体质量控制性能的恶化。
接下来的主题将说明在SMOCPro中如何实施这样一个控制器设计。控制器模型中基于PID模块选择的两个选项都已显示。只有进行设计相关点时才是详细的。
两个框架都有几种常见的设计参数。这些选项包括：
 控制周期：60秒；
 默认压缩点；
 操作变量权重；
名称 阻尼系数 惩罚因子
FOT SP 0.1 0.901124
Quench F 0.1 0.400500
 被控变量权重；
名称 偏差 惩罚因子
Temperature 0.1 10
Quality 1 1
选项1
我们考虑的第一个选项是PID回路被明确建模的情况下。经过模型编译SMOCPro自动生成如下不可测量干扰：

这时候将开环阀位动态变化、PID设置准确地与这些控制器实施的DCS对准是没有必要的。在SMOCPro中只有一种PID块是可用的，因此在某些情况下在DCS和SMOCPro之间进行一些转换是必要的；然而，SMOCPro的PID模块必须在大多数情况下都满足使用。在本例中的开环传递函数是从其它例子拷贝的，通过调谐P和I参数以获得满意的闭环响应。

---

原文：
Normally, the control actions of a SMOCPro controller on this process are sent to the setpoints of the two PID controllers (TC at furnace and quench FC). To have SMOCPro recognize the variations in furnace outlet temperature and their transient effects on the reactor, include the temperature controller PV in the SMOCPro model as an intermediate variable.
The model for the process is:
Here the furnace outlet temperature controller was modeled explicitly. Because a model in SMOCPro only represents the effects of the MVs on the CVs (for prediction purposes), an equivalent model for the same process is:
The transfer function between the furnace temperature SP and its PV is the closed-loop dynamic behavior of the temperature controller loop. The Blue triangle in the top left corner of the ClosedDyn_FOT block denotes the setting for PID/Cascade loop dynamic model.
In both cases, the PID controller rejects the unmeasured disturbance on the temperature PV. Otherwise SMOCPro tries to reject the disturbances by changing the PID setpoint while the PID is compensating for the variations in temperature PV. This leads to unnecessary moves on the PID setpoint and a deterioration of the overall quality control performance.
The next topics illustrate how such a control design is implemented in SMOCPro. Two options are shown, based on the choice of PID model in the controller model. Only the relevant points in the design are detailed.
There are several design parameters common to the two frameworks. These options are:
 Control Period: 60 seconds
 Default compaction points.
 Manipulated Variable Weights
Name Damping factor Penalty
FOT SP 0.1 0.901124
Quench F 0.1 0.400500
 Controlled Variable Weights
Name Deviation Penalty
Temperature 0.1 10
Quality 1 1
Option 1
The first option we consider is the case when the PID loop is explicitly is modeled. After model compilation SMOCPro automatically generates the unmeasured disturbances as follows
It is not necessary that the open loop valve dynamics and the PID settings be aligned exactly with those of the controller implemented in the DCS. There is only one type of PID block available in SMOCPro and in some cases some conversions are necessary between the DCS and SMOCPro; however, the PID block within SMOCPro should satisfy most of the cases. In this example the open-loop transfer function was picked from another example and the P and I parameters are tuned to get a satisfactory closed-loop response.

---

2016.5.15