MATLAB: PID Tune: Plant cannot be linearized. Why

linearizationlinearizedpidplantSimulink Control Design

Hello everyone,

I'm using simulink for simulating a cooling system. When I try to use the tune button on the PID it opens the tuning window and say:

Plant cannot be linearized.

Why is this happening?

The PID is doing is job. I've attached files.

Thank you very much for the help,

Norman

Best Answer

The plant is linearized, but the linearization at t=0 is a gain of zero. Most likely this happens due to initial conditions in the model. You can ask the PID Tuner to linearize the model at a simulation snapshot of say t/2, when everything comes to steady state.

When you do that, the PID Tuner successfully linearizes the model and designs a PID controller.

Here are the snapshots of the steps using R2014a.

Related Solutions

MATLAB: Linearization of a simulink model and pid controller tuning

This is very similar to the sitaution shown in this example . If you have 14a and all needed tools (Simulink Control Design and System Identification Toolbox) you can follow the workflow shown in the example. I tried it and everything works just fine.

A couple of points:

When you specify a step signal to inject to the plant, you need to do it in a smart way, which means you can't just keep default options. If you run the simulation with your existing gains, you will see that controller request (output of PID Controller block stays at around 0.007. Therefore, when you specify a step signal to inject, specify an offset to be 0.007 and step magnitude to be 0.001 as shown below.

Now run the simulation, apply the results, then fit the default first order model. You can see it is not perfect, you can play with higher order models, but I was able to get a decent controller even with first order plant model.

Tune how fast you want your controller to be using the sliders, here is my design (solid line) vs your existing gains (dashed line). Your existing design seems pretty good already.

apply the tuned gains, run the simulation. Error signal looks pretty good.

MATLAB: How to measure PID parameters from PID tuning of SISOtool

Export the compensator you designed, C, to MATLAB workspace. Then in MATLAB type

C=pid(C)

This will convert the compensator from a transfer function to a PID object. You can then access the gains as shown below.

Best Answer

Related Solutions

MATLAB: Linearization of a simulink model and pid controller tuning

MATLAB: How to measure PID parameters from PID tuning of SISOtool

Related Question