案例4：三个储罐串联的液位控制-3

Figure 20. Simulation results for the case where MV_1 is removed (Left: all CV at Prio 1, Right: CV Prio 1,2,3). 图20：MV_1移除时的仿真结果（左侧：所有的CV优先级等于1，右侧：CV优先级为1,2,3）。

如下所示，接下来我们探索当MV_2处于“Local”时的仿真结果。此图有三个重要的地方需要指出：

1. 注意时间戳中当控制器切换到控制模式（第5步）至当干扰信号第一次从原点移除（第20步）。这里我们注意到在图的左侧SMOCPro不能控制CV_1和CV_2至其设定值。这种情况下，所有的CVs都具有相同的优先级，鉴于SMOCPro只能通过MV_1控制2个CVs至设定值，它不能同时满足所有的控制目标，因此基于CV权重其放弃了所有的CVs，这与该仿真是相同的。这导致CV_1和CV_2偏离设定点的差值相同。现在将你的注意力转移到图右侧的相同时间戳处。这里我们已经分别指定CV优先级为1,2,3。 SMOCPro使用此信息来计算MV_1的动作计划以控制更高优先级的CV，在本例中是CV_1,并且放弃CV_2。第3个CV是由MV_3控制的，，因此在这里不受影响。
   2.接下来我们来关注从干扰进入系统时（第20步）到MV_2重新置回控制问题时（第400步）的时间戳。在此期间我们可以看到图左侧都有类似的缺陷，因为所有的CVs都具有相同的优先级， SMOCpro不知道区分CV的优先顺序，因此只能以同时放弃CV_1和CV_2并将CV_3保持在设定点而结束。图的右侧显示了当斜坡CV优先级已经错开时的控制器行为。注意SMOCPro如何控制CV_1和CV_3到设定点，并且放弃CV_2。
   3.最后，在400步后SMOCPro能够重新操作MV_2，因此使用所有这3个手段把所有CV带回设定值，而不用顾忌CV优先级。

Figure 21. Simulation results for the case where MV_2 is removed (Left: all CVs Prio 1, Right: CV Prio 1,2,3). 图21：MV_2被移走时的仿真结果（左：所有CV的优先级都为1，右：CV优先级分别为1,2,3）

如下所示，接下来我们观察最后一种情况的仿真结果。该方案研究当MV_3处于“Local”状态时的控制行为。然后如前面的情况一样，让我们来分析相同的三个时间戳：

---

原文：
Next we explore the simulation results for the case when MV_2 is on “Local,” shown below. There are three important items to point out in this figure:

1. Focus on the timeframe from when the controller is switched to control (step 5) to when the disturbance signal first moves away from its origin (step 20). Here we notice that on the figure on the left SMOCPro has not been able to control CV_1 and CV_2 to their setpoints. In this case all CVs are at the same priority and since SMOCPro only has MV_1 available to control 2 CVs to setpoint it cannot meet both control objectives and gives up on both CVs based on the CV weights, which for this simulation are the same. This results in the deviation from setpoint being equal for both CV_1 and CV_2. Now shift your attention to the same timeframe on the right side of the figure. Here we have specified the CV priorities to be 1,2,3 respectively. SMOCPro uses this information to plan moves for MV_1 to control the higher priority CV, in this case CV_1 and give up on CV_2. The third CV is controlled with MV_3 and as such is not affected here.
2. Next, focus on the timeframe from when the disturbance enters the system (step 20) to when MV_2 is brought back into the control problem (step 400). In this period on the left side we see a similar deficiency in that because all the CVs are at the same priority SMOCPro does not know which CV to prioritize and ends up giving up on both CV_1 and CV_2 while maintaining CV_3 at setpoint. The right side of the figure shows the controller behavior when the ramp CV priorities have been staggered. Notice how SMOCPro controls CV_1 and CV_3 to setpoint and only gives up on CV_2.
3. Lastly, after step 400 SMOCPro is able to manipulate MV_2 again and uses all three handles to bring all CVs to their setpoints, regardless of the choice of CV priorities.
   Next we look at the simulation results for the last case which are shown below. This scenario considers the control behavior for the case when MV_3 is on “Local.” Again, let us analyze the same three timeframes as in the previous case:

---

2016.5.28