Characteristic Parameters Identification of Characteristic Models of Linear Time Invariant Systems
线性定常系统特征模型的特征参量辨识

For the linear time invariant(LTI) systems, it is shown that the characteristic parameters of characteristic model are condensed by the system information of high order LTI form, therefore some tracking algorithms, which are used to deal with the slowly time varying parameters that are unrelated with system states,are not suitable in this case. This paper establishes the connection between the characteristic parameters of characteristic model for LTI systems and subspace method, and presents a composite identification algorithm to estimate these parameters. Furthermore, it is proved that when the sample number for subspace identification is sufficiently large and the time for state estimation is sufficiently long, the error between the estimated values and the true values of characteristic parameters can be sufficiently small. A simulation example of six order singleinput single output(SISO) model is considered, and the proposed method is compared with the projecting forgetting factor recursive least square(FFRLS) algorithm. The simulation results show that the proposed method in the paper has more advantage than FFRLS.