About stationary kalman predictor


Thread Starter


In the process control, we may get involved with the estimation of system states through state observer. One of the problems, i.e. to assign observer poles in an optimal way, is also to minimize the state estimation error. And the solution can be derived through the stationary kalman predictor by solving an ARE (algebraic Riccatic equation) based on the well-established model. But, does this conclusion still stand if the system matrices (i.e. A, B, C) are identified through the available process measurements? As far as I'm concerned, the LQR theories or other optimal theories may degrade with the bad model because of the noise. Thanks in advance.