抄録
This paper addresses the problem of identifying errors-in-variables models, where the input measurement is corrupted by white noise whereas the output measurement is corrupted by colored noise. The Koopmans-Levin method is one of possible methods for identifying errors-in-variables models. However, it requires a priori knowledge of the measurement noise covariances. To achieve the consistent estimation without a priori knowledge of the input noise variance and the colored output noise covariances, the methods presented in this paper use the idea that only the bias induced by the colored output noise can be removed using instrumental variable technique. Then the parameter estimation problem can be solved as the generalized eigenvalue problem. Moreover, bias compensation principle based algorithms are derived. The results of simulated example indicate that proposed methods provide good parameter estimates.
