Re. Skelton et G. Shi, ITERATIVE IDENTIFICATION AND CONTROL USING A WEIGHTED Q-MARKOV COVER WITH MEASUREMENT NOISE, Signal processing, 52(2), 1996, pp. 217-234
An identifiability result determines whether there exists a linear tim
e-invariant model capable of matching certain input/output data observ
ed from experiments with measurement noise. If no match is possible, a
method is presented to find the closest linear model to the data. A w
eighted q-Markov COVER method is introduced for identification with me
asurement noise. These results are used to develop an iterative closed
-loop identification/control design algorithm.