Adaptive control of linear time-varying plants (Q1092857)

The paper deals with the model reference adaptive control (MRAC) of a class of linear time-varying (LTV) continuous-time single-input single- output plants. The parameters of the plant depend on time in a smooth and bounded fashion and, for each fixed time, they satisfy the usual assumptions for MRAC of linear time-invariant (LTI) systems. The reference model in turn consists of an LTI system. The structure of the controller employed is the one used for the LTI case in \textit{K. S. Narendra} and \textit{L. S. Valavani} [IEEE Trans. Aut. Control AC-23, 570- 583 (1978; Zbl 0389.93025)]. It is first shown that, if the parameters are sufficiently slowly varying, it is possible to find a controller with the given structure such that the closed loop input-output (I/O) operator of the plant closely approximates the I/O operator of the LTI reference model. To adjust the controller parameters it is then used the robust adaptive law (RAL) given by the authors [ibid., AC-31, 1033-1043 (1986)]. This requires, as additional a priori information, a lower bound on the stability margin of the inverse plant for each fixed time. It is shown that such RAL guarantees the boundedness of all the signals in the closed loop for any bounded initial condition. Furthermore, the tracking error converges to a residual bounded set and reduces to zero asymptotically as the plant parameters become constant.