从模型图中可以看出，顶部和底部两个温度都是影响其相应质量的中间变量。SMOCPro不控制这些温度；相反，通过使用观察模块，它预测温度扰动对产品质量的影响。由于操作手段和所需要控制的质量之间存在着显著的delay，这是非常重要的。还需要注意的是塔盘装载量不会影响任何其它输出。
在该模型中进料流量（Feed Flow）是一个测量干扰。它的影响绕过了底部温度，被直接送入底部质量。根据流程类型，底部质量变化是底部温度变化的反映。然而，因为进料流量（Feed Flow）对底温没有显著的影响，上述模型是不矛盾的。测量干扰背后的最终目的是从进料变化预见底部质量的变化。
7、 输入模型后，关闭GMB以返回到SMOCPro主窗口。
8、 到编译（Compile）窗口并单击Compile。
这个程序编译了模型，并为模型生成不可测扰动，以解释可能的模型误差。
9、 回到GMB查看更新后的模型。
该模型现在反映了模型误差的潜在原因，如未测量干扰（默认假设集成为白噪声），现在已经被添加到每个流程输出测量。需要注意的是模型可以将这些干扰由一个输出传递到另一个，由温度传递到质量。
控制器设计（Controller Design）
若要访问控制器设计模块，请单击树节点上的Controller（控制器）。
10、 在Controller Selection（控制器选择）窗口中，单击New创建新的控制器设计。
11、 选择控制器后，打开了Sub-Controller Scope Definition（子控制器范围定义）窗口的默认设计：在同一动态子控制器中所有的过程变量都被配对。
12、 在树中选择Sub_Debutanizer节点并接受默认的概要参数，使动态子控制器具有总控制器的执行周期。
13、 在Controlled Variables选项中，取消顶温和底温选项，因为它们不是控制目标的一部分。它们只是作为中间变量观察过程。
注意MVs也能被设成被控变量。
14、 在Weights选项中，设定以下权重以强调顶部质量的控制：
原文：
As can be seen from the model diagram, both the Top and Bottom temperatures are intermediate variables leading to their respective qualities. SMOCPro does not control these temperatures; rather, through the use of the observation module, it anticipates the effect of temperature disturbances on the qualities. This is important because significant delays exist between the means of actions and the qualities to be controlled. Note also that the tray loading does not affect any other outputs.
In this model the Feed Flow is a measured disturbance. Its effect is directly fed into the bottom quality, bypassing the Bottom temperature. According to the type of process, the variations in Bottom quality are a reflection of variations in the Bottom temperature. However there is no contradiction in the model above in case the Feed Flow has no significant effect on the Bottom temperature. The ultimate objective behind the measured disturbance is to anticipate the variations in Bottom quality from Feed changes.
Once the model is entered, close the GMB to return to the main SMOCPro window.
Go to the Compile window and click on Compile.
This procedure compiles the model and creates the unmeasured disturbances for the model, explaining possible modeling errors.Go back to GMB to view the updated model.
The model now reflects the potential source of modeling errors, for example, unmeasured disturbances (assumed to be integrated white noises by default), have now been added for each process output measurement. Note that the model lets these disturbances propagate from one output to another, from temperatures to qualities.
Controller Design
To access the controller design module, click on Controller on the tree nodes.In the Controller Selection window, click New to create a new control design.
After the controller is selected, the Sub-Controller Scope Definition window opens with a default design: all the process variables are paired in the same dynamic sub-controller.
Select the Sub_Debutanizer node in the tree and accept the default Summary parameters, so that the dynamic sub-controller has the overall controller execution period.
In the Controlled Variables tab, uncheck the Top and Bottom temperatures, as they are not part of the control objective. They are only used as intermediate variables for the observation process.
Note that MVs can also be set as controlled variables.In the Weights tab, set the following weights to emphasize control of the Top quality:
2016.5.14
