Identification and control of non-stationary time delayed systems

This paper presents a study on identification and control of continuous systems including a time-delay. A new recursive identification method is proposed based on the using of a finite impulse response filter. Stochastic simulations as well as a criterion analysis of the identification algorithm show that the recursive estimator converges towards a single optimum. Next, the recursive identification algorithm is associated to the synthesis of a self-tuning controller. This adaptive control algorithm is validated on a physical system. The results show that the required performances in closed loop are respected when the time-delay varies in a large range. Copyright (C) 2000 IFAC.