Out-put over-sampling based identification of noisy closed-loop system

The identifiability of closed-loop system was restricted by either reference signal or controller structure in ordinary approaches. In order to resolve the limitations of earlier works a new identification method for a linear discrete-time noisy closed-loop system is proposed based on an output over-sampling scheme. Because the over-sampled outputs contain significant information about the system structure, even if the ordinary identifiability conditions are not satisfied, a close-loop system can be identified without utilizing any external test signal or controller switching. For the three noise conditions of white noise, colored noise and plant noise, general least squares method(GLS), prediction error method(PEM) and assistance variables method(AVM) based on out-put over-sampling are advanced. Compared with conventional algorithms, these methods have higher estimation speed and estimation accuracy. Numerical simulation shows that the parameters difference between real system and the estimation is lower than 1%, 2% and 5% separately when using the three identification methods of GLS, PEM and AVM.