how can I design a controller for a plant that cannot be linearized

race lounge

[Raj]

I think you should see this first:

https://in.mathworks.com/matlabcentral/answers/6200-tutorial-how-to-ask-a-question-on-answers-and-get-a-fast-answer

Nevertheless, Since you have not shared any details about your system, its difficult to say anything. Most of the plants have atleast a few trim points about which you should be able to linearize the system and design a linear controller for that trim point. Otherwise I see no option but to go for non linear control techniques like say fuzzy control.

Answer 1

[Jesús Zambrano]

Hi Ali,

Even if the dynamics equations are fine, some nonlinear systems are difficult or impossible to linearize. A possible way is to use data to estimate a model of the system.

The workflow is as follows:

• In the PID Tuner tab, select Identify New Plant.
• The Plant Identification tab will open. Select Get I/O Data, then Simulate Data so to generate identification data by simulating the model.
• It will open the Simulate I/O Data tab, where you specify the input data (for example a step function) used for the plant identification.
• Click Run Simulation. Then, click Apply and Close.
• Then, the Plant Identification tab will open showing the results of the plant identification. Once in there you can modify the response by moving or adding factors (for example delay, zero, integrator).
• Select Apply. It will create a new variable in the Data Browser section of the PID Tuner.
• Go to the PID Tuner tab, select Step Plot: reference tracking plot. Here the PID is using the new identified plant. You can still improve the system response by moving the sliders.
• Click Update Block and simulate the model with the new PID parameters.

Hope this can help to solve your question.