Neural network output feedback passivation of MIMO uncertain systems: The case of unmodeled dynamics

A robust adaptive output feedback neural network switching controller is designed, capable of rendering an affine in the control uncertain MIMO nonlinear system strictly passive, in the presence of unmodeled dynamics. Subsequently, a simple output feedback is employed to stabilize the system without using any measurement or estimation of the unmodeled state. Under an Unboundedness Observability property assumed for the unmodeled dynamics, uniform ultimate boundedness of both the system's output and of the nominal system's state is guaranteed, along with boundedness of all other closed loop signals.