Adaptive output feedback control of nonlinear systems using neural networks

An adaptive output feedback controller design procedure for uncertain nonlinear systems is developed which avoids the use of state estimation. To achieve this goal three separate problems are addressed independently: controller design, derivation of parameter update laws and approximate mapping of an unknown dynamic function from its input/output history. To handle the uncertainty, the controller, in the form of a dynamic compensator, is augmented by a single hidden layer (SHL) neural network that adjusts on-line for unknown nonlinearities. The parameter update laws for a SHL neural network are derived from stability analysis. Simulations illustrate the theoretical results.