Does a LQG/LQR controller work well in the industry?

Question

A PID-controller are often used in the industry because it's very easy and fast to set up. But there are som limitations.

LQG/LQR are not used in the industry because it requires a good mathematical model. But the are a lot of benefits about LQR/LQG.

LQG/LQR is also called Optimal Control and LQG have no guaranteed stability margins, but LQR does. LQG have a kalman filter which reduce noise and can estimate all states by using one mesurement.

So my question for you are:

How good is a LQG/LQR in the industry?

Is LQR/LQG to none-robust for the industry?

Why are not LQR/LQG used?

Answer 1

first of all I can only tell you why LQR/LQG is not so far used in industrial application. Actually you answered your question by yourself. LQG is nothing else than a minimum variance controller based on the optimal estimation of the Kalman-Filter algortihm for the states. Due to model uncertainties in the system identification for modeling and the tuning parameters which has to be determined with brute force application the Kalman Filter can optimize with these information. So with wrong adjustments you won't get a good estimation for the controller performance.