Switching between several controllers, depending on SP error.

[jblc]

I'm curious about your preferred way to analyze and develop a hybrid feedback control system, where the controller is actually made up of a few unconnected controllers that you switch between. The loop is a standard feedback loop: SP --> SPerror --> controller --> plant --> PV, and PV is fed back to get SPerror.

As an example, say you control Valve 1 a certain way if the SP error is small, and control Valve 2 a certain different way if the SP error is larger. Assume each valve has a different risetime and dead-time.

The controller in this case is a hybrid control made up of two controllers that control two plants (valves), and is nonlinear in nature.

How do you go about analyzing performance for such a system? Clearly it's east to analyze each loop independently (the Valve 1 control+plant loop, and the Valve 2 control+plant loop), but that doesn't characterize both together...

Independent loop analysis is especially not useful if trying to model the system with an input disturbance, since analyzing each loop independently wouldn't capture how the actual hybrid system would performs with both plants.

 

[waross]

Have you looked at split ranging?
The greater the setpoint error, the greater the controller output.
Pneumatic actuators are available that act on only part of the range.
EG: with a 3to 15 psi output, actuators are available that will go from zero to full range with an input pressure of:
3 to 9 psi.
6 to 12 psi
or 9 to 15 psi.
The first actuator will act as the controller output increases from 3 to 9 psi,
the second actuator will act as the pressure rises from 6 to 12 psi
and the third actuator will act as the pressure rises from 9 to 15 psi.



Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[controlnovice]

Typically this is not done with two controllers.

Per waross - Try Split Range with one controller: When controller is between 0 - 50%, it send the output to 0 - 100% of the small valve. When the controller is between 50-100%, it send the output to 0 - 100% of the large valve.

Another option with two controllers:
The controller for the small valve is controlled using the measured value only....like a single loop controller.
The controller for the large valve is controlled off the position of the small valve. So...if the small valve is open more than 80%, the large valve will start to open. If the small valve is less than 20% open, the large valve will start to close. If the small valve is between 20-80%, the large valve doesn't move.

With each of these, the Cv of the valves needs to be considered where they 'cross over'.

______________________________________________________________________________
This is normally the space where people post something insightful.

 

[djs]

What I have seen is that the controller tuning values are changed on the fly. Something you might want to look into. Switching controllers may be difficult due to making the transition bumpless.

 

[sreid]

Today, complicated and non-linear control systems are simulated.