A Globally Stable Multivariable Adaptive Controller with Reduced Prior Knowledge

A globally stable direct adaptive scheme for multivariable stably invertible systems is proposed. The relevant prior knowledge required to implement the algorithm reduces to the powers of the diagonal terms of the interactor and its largest overall power. This result significantly weakens the assumptions required to adaptively stabilize multivariable systems. To achieve the claimed result, a modified process reparametrization and estimation scheme are proposed. The main technical device to prove stability is the sector stability theorem [10].