Investigations into model-reference adaptive control systems

The paper considers the behaviour of model-reference adaptive control systems when the system is very simple. Theoretical analysis of the behaviour of the adaptive loop is followed by analogue-computer studies. A second-order system with a second-order model and then a third-order system with a second-order model are considered. Effects of various inputs are investigated, including a ramp, square wave and white noise. The results demonstrate that the model-reference adaptive loop is capable of very fast response. It is also shown that, where the system and the model are of a different order, the frequency spectrum of the input has to be carefully considered to avoid large errors. It is also shown that one effect of the presence of noise may be to keep the loop parameters at or around their correct value in the absence of demand signal.