A new approach to model reference adaptive control

Here we propose a new approach to model reference adaptive control which yields a linear periodic controller and which tolerates rapidly varying parameters. Rather than estimating the plant or compensator parameters, instead we estimate what the control signal would be if the plant parameters were known and the ´correct LTI compensator´ were applied; we are able to do so in a linear fashion. Our assumptions are reasonably natural extensions of the classical ones to the time-varying setting, although we add a compactness requirement. We prove that we can obtain arbitrarily good tracking with a nice transient response.