How to determine if the control signal I got in tuning my PI controllers for my SS model is valid?

How to determine if the control signal I got in tuning my PI controllers for my SS model is valid?

Answer 1

[Birdman]

Well, I have checked your model. You defined your control signal as Joule, and it almost rises from zero to its peak value in no time. This is not realistic since you arranged your Kp coefficient higher than 700. In simulation, it may give you a fine result, but in reality, since the control signal applied on the system is mostly voltage, you can not increase your Kp coefficient this much. Therefore, to make your control process more realistic, I suggest you to either put a saturation element in front of your control signal or redesign your controller. Note that mostly the energy given to the system is generated by an electrical source, which means you have to control the voltage level coming out of your driver circuit.

[Mr. NailGuy]

Hi cvklpstunc, I actually reduced my Kp to 500 so I got an input signal of around 5x10^4 which is actually large enough too. But looking on the B-matrix of the model, will this not require such amount of energy to reach a steady state before 120s with no overshoot at all? Thanks

[Mr. NailGuy]

I actually tried so many combinations and also placed a saturation inside the PID block to 8000 but I cant reach the steady state value of 90 before 120 seconds. I also did decoupling before the PI controllers but the outputs were all the same. Do you have any suggestions how to modify these? Would appreciate any.

[Birdman]

Actually, increasing Kp this much should have brought overshoot to the system but since it does not bring any overshoot, you do not have to worry about overshoot issue. Decreasing Kp will eventually increase the settling time. Just want to know, what generates your control signal in reality?

[Mr. NailGuy]

The control signal is the heating power required in the bake plate that would be used to then heat the wafer in one of the processes in semiconductor manufacturing. I am actually not sure if its in watts or in Joules as it was not specified.

[Birdman]

I understand it, the energy source will be electrical, therefore it means that your voltage source has to have a capacity to meet your controller requirements, otherwise it would be not possible with this parameters to control the temperature.

[Mr. NailGuy]

Hi cvklpstunc, the thing is I don't why the response is like this? I tuned it for so many times and it don't reduce to reasonable levels. I guess, I will just account this to the model provided which I am not in control? The B-matrix have a very low values tho. Thanks

[Birdman]

Then you have to check your model once again.

[Mr. NailGuy]

Hi cvklpstunc, I can't change it as it was just provided to me :) The only thing I can do is use controllers to adjust its response. Thanks

[Birdman]

Ok, if there's anything, let me know again.

[Mr. NailGuy]

Hi cvklpstunc. I guess I need to stick with this as the deadline is tomorrow. But if in case you tried it and come up with a better tuning with lower values for the control signal, I will be glad if you can pass that to me :) Although, I already adjusted my proportional gains to around 500 and that is by far I think my best tuning. Thanks