Simple adaptive algorithm for system with short sample time

Recursive identification has played a major role since 1960 at which time the first self-adaptive control system was proposed. The paper describes a simplified estimator which utilizes a covariance resetting scheme. The estimator structure is similar to the well known least square algorithm which has fast initial convergence rate, but the gain correction algorithm reduces fast when the covariance matrix elements get very small after few iterations. The covariance resetting technique proposed in the paper revitalizes the gain algorithm giving the capability of tracking the time varying parameters. Some convergence analysis is also given