Adaptive identification for time-varying continuous-time systems is considered in this paper. Although the convergence speed of least square adaptive law is extremely rapid, its performance becomes deteriorate remarkably when it is applied to time-varying systems. In order to overcome its defect while making full use of its advantage, an adaptive identifying scheme using least square algorithm with intermittent reset is proposed. When parameters to be identified vary as an exactly quadratic function of time, the identification problem can be reduced to the one with constant parameters by introducing two additional parameters for each time-varying parameter. The adaptive identifying scheme using least square algorithm is applied over a prescribed short period by assuming that parameters vary quadratically with time. At the end of the period, the matrix Γ(t) is reset to Γ(0) of large value, while the initial values for the estimates are chosen so as the estimated value and its derivative continue at the connecting instant. The scheme is cofirmed to be of satisfactory tracking performance through simulation studies. Besides, the least square algorithm for contiuous-time systems is developed in this paper.
