Model reference adaptive control for systems having non-square transfer functions

This paper describes algorithms for model reference adaptive control of systems having non-square transfer functions. The motivation for this study was a problem related to the control of an artificial heart having 3 outputs and 2 inputs. The paper treats the case where the number of inputs is greater than or less than the number of outputs. The adaptive control algorithms developed here are shown to be globally convergent provided a subset of the parameters is projected into a suitably defined convex region. This is a natural generalization of procedures used in the single-input single-output case.