Adaptive control for linear slowly time-varying systems using direct least-squares estimation (Q5926173)

The proposed adaptive scheme relies on a least-squares identification algorithm, where the minimized criterion includes an exponential forgetting factor. It is shown that the estimates and their change ratio are bounded. In the case of no persistent excitation, the estimates can be modified without losing their properties. The procedure does not require explicit knowledge on the noise bound or the region where the true parameters lie. Meaningful properties of the estimates are obtained when bounds on the noise and parameter variations are small. A hybrid control strategy ensures the closed-loop pole placement and the boundedness of the state-space variables.