PCTP

案例3:储罐液位控制-MV Horizon(范围)和POV Im

2016-05-22  本文已影响31人  橡果

干扰抑制
现在我们把注意力转向控制器整定。在这个例子中,目标是调谐控制器鲁棒性以抵消不可测干扰对液位的影响。此外,我们向仿真中引入噪声并与先前讨论的选型作比较。
过程模型

*Figure12. Tank vessel control model with unmeasured inlet flow.* *图*12**:含不可测进料流量的油罐容器控制模型**
在控制器模型中,存在着两个可能的干扰来源:Inlet Flow(进料流量)是不可测量干扰,Measurement Noise(可测量噪声)将被用于直接向液位测量中注入干扰。通常情况下,以下两种类型的干扰将输入到系统:
 类白噪声干扰可能是由于出料流量执行器的噪声(难以预测的);
 阶梯状干扰,可能表示注入相关的进料流量(可能停留片刻)。
对某个控制器,从长远来看最好忽视第一种类型,并要尽量保持对第二种类型的可预见性。这项考虑被嵌入在一个叫Impulse Factor(脉冲因子,数值在0.0~1.0范围内)的调谐参数中。这个参数代表了在POV估算中当前预测误差可能在控制时域中消失的部分所占的百分比。
控制器设计
控制器搭建时将储罐液位定义为CV。此外,如下图所示,我们将Vessel Level SR(储罐液位SR)定义为一个Elaborated(复杂的) CV。这个额外的CV允许我们将储罐液位的Setrange(设定范围)定义一个较高的优先级,而为设定点定义较低的优先级。

接下来,我们制定了下列move suppression(动作抑制)和被控变量权重。


最后,我们对快速控制动作感兴趣,因此如下图所示,默认的MV时域100 × dT将被修改为5 × dT。记住进行这项修改时要使用Calculator(计算器)功能。


仿真
在这个仿真中我们将关注先前讨论的两种类型的干扰,随机噪声和斜坡干扰将被注入储罐中。我们用两种方法实现这些干扰,以展示给用户向不可测扰动变量(UNM)和测量变量注入干扰的差异。
我们要运行的第一组仿真,是将干扰直接注入到Inlet Flow(进料流量:不可测干扰)变量中。对这种情况,General仿真情景选项卡包含下列设置:

原文:
Disturbance Rejection
Now we turn our attention to controller tuning. In this example, the goal is to tune the controller for robustness against unmeasured disturbance rejections on the Level. In addition, we compare the previously discussed options of introducing noise into the simulation.
Process Model
In the controller model, there are two possible sources for disturbances: the Inlet Flow is an unmeasured disturbance and the Measurement Noise will be used to inject disturbances directly into the Level measurement. Typically, two types of disturbances enter the system:
 A white noise-like disturbance possibly due to noise on the Outlet Flow actuator (Hardly predictable), or
 A step-like disturbance that may represent coherent Inlet Flow injection (Likely to stay for a while).
From a controller point of view, it may be best to disregard the first type in the long run and to try to keep a certain predictability of the second type. This consideration is embedded in the tuning parameter called the Impulse Factor (with values between 0.0 and 1.0). This parameter represents the percentage of the current prediction error in the POV estimation that is likely to vanish along the control horizon.
Controller Design
The controller is build with the Vessel Level defined as a CV. In addition, we define Vessel Level SR as an Elaborated CV as shown below. This extra CV allows us to define a high priority Setrange with a low priority setpoint for the Vessel Level.
Next, we specify the following move suppression and controlled variable weights.
Lastly, we are interested in fast control action therefore the default MV horizon of 100 × dT gets modified to 5 × dT as shown below. Remember to use the Calculator feature when making this change.
Simulation
We now present simulations where the two types of disturbances previously discussed, namely random noise and a ramp disturbance, are injected into the vessel. We implement these disturbances in two ways to show the user the difference between injecting the disturbances into the unmeasured disturbance variable (UNM) and into the measurement directly.
The first set of simulations that we run are with the disturbances being injected directly into the Inlet Flow (unmeasured disturbance) variable. For this case the General simulation scenario tab contains the following settings:


2016.5.22

上一篇下一篇

猜你喜欢

热点阅读