仿真
我们考虑了4个具有不同Impulse Factors（脉冲因子）的仿真情景。该过程的操作条件如下：
容器液位初始值在其设定值50（Setrange High设定高限 = Setrange Low设定低限= 50)。
MV初始值也为50，最大值为100，最小值为0，最大动作幅度为2。
Inlet Flow 进料流量(DV)起始于操作点0。
为了生成可重复测量噪声，我们设定“Measurement_Noise” DV的“Noise Seed = 1”和“Noise std dev = 1.0”。为了由SMOCPro将这一测量量实现为真实不可见噪声，我们需要将“Measurement_Noise” DV的“Disconnected（断开连接）”标签设定为“True”。
控制器在Standby （挂起）模式下开始，并在第5步时切换到control（控制）。在第30步时我们给inlet flow（进料流量）引入一10个单位的斜坡干扰，在第80步时我们将inlet flow（进料流量）设定回初始值。在第140步时我们将液位设定值由50下调到40（Setrange Low设定低限 = Setrange High 设定高限= 40)。
我们运行199步仿真，并观察SMOCPro计算的将储罐液位带回设定点的动作计划。所考虑的4个仿真场景除了液位POV的Impulse Factors（脉冲因子）不同，其余条件都相同。其值分别为0,0.5,0.9和0.99。
*Figure 11. Level control for a SISO example with varying POV impulse factors.
图11：SISO例子具有不同POV impulse factors（脉冲因子）的液位控制*
上图显示了具有4个不同液位POV Impulse Factors（脉冲因子）的仿真情景运行结果。可以明显地看出不同的Impulse Factors（脉冲因子）将影响控制器性能。在所有附图中，我们可以看到顶部是传感器噪声（绿色），往下是可测量干扰（紫色），再往下是蓝色的MV，最下面是红色的CV。需要强调的三个重要方面是：
左上象限为脉冲因子设定为0时的仿真结果。如指南文件中所讨论的一样，这种情况将导致SMOCPro突出所有的不可测干扰，因此斜坡CV将把所有的不可测干扰整合进未来。最终的结果是，由于移动目标计算的缘故，SMOCPro层面将过度计算MV动作。这可以通过蓝色虚线所示的MV目标值的快速变化看出。
提高脉冲因子将使得MV动作更为平滑，但可能导致对一个“真正的”扰动存在响应迟缓。
最后，由于脉冲因子的“指数”性质，从图的下半部分可以看出，IP=0.9和IP=0.99有明显的不同。请注意蓝色的MV目标值在IP=0.99时是稳定的，但在IP=0.9时依然有略微锯齿状。
原文：
***Simulation ***
We consider four simulation scenarios with different Impulse Factors. The operating conditions for the process are the following:
Initially the level in the vessel is at its setpoint of 50 (Setrange High = Setrange Low = 50).
The MV also starts at 50 with a Maximum value of 100, Minimum value of 0 and Maximum Move Size of 2.
The Inlet Flow (DV) starts at an operating point of 0.
To generate repeatable measurement noise we set “Noise Seed = 1” and “Noise std dev = 1.0” for the “Measurement_Noise” DV. To implement this measurement as a true unseen noise by SMOCPro we must set the “Disconnected” flag for the “Measurement_Noise” DV to “True.”
The controller starts off in Standby mode and is switched to control at step 5. At step 30 we introduce a ramp disturbance of 10 units into the inlet flow and at step 80 we bring back the Inlet Flow to its starting point. At step 140 we lower the setpoint in the level from 50 to 40 (Setrange Low = Setrange High = 40).
We run the simulation for 199 steps and observe the planned moves that SMOCPro calculates to bring the vessel level back to setpoint. The four simulation scenarios under consideration are all identical with the only difference being the different Impulse Factors on the Level POV. These values are: 0, 0.5, 0.9 and 0.99.
The figure above shows the results of running the simulation scenario with the four different Impulse Factors for the Level POV. It is clearly seen that having different Impulse Factors affects the performance of the controller. On all figures, we see the sensor noise (green) at the top, followed by the measured disturbance (purple), the MV is shown next in blue and the CV is shown at the bottom in red. Three important aspects that should be highlighted are:
The top left quadrant shows the simulation results of setting the impulse factor of 0. As discussed in the guidelines document, this case results in SMOCPro projecting any unmeasured disturbance so that the ramp CV will assume that the unmeasured disturbance must be integrated wholly into the future. The end result is that for the case with levels SMOCPro calculates excessive MV movement due to moving target calculations. This can be seen by the quickly changing MV targets depicted by the dashed blue lines.
Increasing impulse factor smoothes out MV movement but may result in sluggish response in the presence of a “real” disturbance.
Lastly, due to the “exponential” nature of the impulse factor, there is a non-trivial difference between IP=0.9 and IP=0.99 as can be seen in the bottom half of the figure. Notice how the target values for the MV in blue are steady in the 0.99 case but still are slightly jagged for the 0.9 case.
2016.5.21
网友评论