Identification of linear time-varying systems using a modified least-squares algorithm (Q1571084)

The paper presents a modification of the standard least-squares method aimed for use in the identification of continuous-time linear time-varying systems. The modification consists in the inclusion of extra terms in the parameter estimates and the covariance matrix update laws. A series of properties of the obtained estimates, of the covariance matrix and of the identification error are established, essential from the viewpoint of adaptive control.