🛠️ Help with PID Tuning and System Linearization for Thermostat Project

Hello MATLAB Community,

I'm currently engaged in a thermostat project focused on household heating, which involves integrating a PID controller. Within my Simulink model, I've implemented the PID alongside components such as a water-heated radiator equipped with a two-way valve, and a room model.

The PID controller operates by taking the error (temperature setpoint minus the air temperature in the room) as input and generating a percentage for the opening of the two-way valve, thereby controlling the radiator's output.

At the outset, the room temperature is set to 17°C, while my desired setpoint is 20°C. My primary objective is to fine-tune the PID controller for optimal performance. However, I encountered an issue when attempting to utilize the tuner, as it indicated that the plant cannot be linearized.

To address this, I explored the option of linearizing the system using the "re-linearize closed loop" feature, focusing on a specific operating point (snapshot time). Initially, I chose to linearize the system at steady-state, specifically at 17°C just before transitioning to my setpoint of 20°C. Subsequently, I performed another linearization at the 20°C steady state, which resulted in the tuner suggesting significantly improved default PID gains compared to the linearization performed at 17°C.

I seek guidance on whether my approach to linearizing the system is correct. Should I prioritize linearization at the 17°C steady state or the 20°C steady state? Additionally, I welcome any recommendations or insights regarding PID tuning strategies for this application.

Thank you for your time and assistance!

Best regards !

[Ram Prasanth]

Hello @Jeffry, When you linearize a non linear system. try to look in to different setpoints like one stepoint on the lowest level, one on mid level and the other one on high level. in your case may be you can choose 15, 20 and 25.

Then compare the outstup of the step response given and compare the system order and see the changes between the setpoints. Sometimes you might get first order system with delay on the lowest level and 2nd order system on the higher level. in such case it is advisable to use the 2nd order system and this should linearize the system on the tough coniditions too and provide expected contrôle.

I hope my explanation is clear.

Also there is another type of controlled which is IMC. you can look in to it. it might be useful for the non linear systems.

[Jeffry]

Hello, Ok thanks I will do that :) !