Multivariable adaptive actuator compensation with partially known high frequency gain matrix

To design an adaptive control scheme for a multivariable plant with unknown actuator nonlinearities at the input of a linear dynamics, it is important to use less a priori knowledge of the high frequency gain matrix of the linear dynamics. To solve this issue, this paper presents an adaptive control scheme, using only partial knowledge of the high frequency gain matrix. Such an adaptive control scheme employs a multivariable adaptive inverse to cancel the actuator nonlinearities and leads to stable parameter adaptation. It is also shown that the required high frequency gain matrix knowledge for an adaptive inverse design is also necessary if a quadratic gradient algorithm is used for adaptation