ROBUST ADAPTIVE CONTROL USING NEURAL NETWORKS AND PROJECTION

By using dynamic neural networks, we present a novel robust adaptive approach for a class of nonlinear system. We assume the controlled plant is a "black-box", only input-output are measurable. The dynamic neural networks are used to identify the plant on-line. Then a linearization controller is designed based on the neuro identifier. Since the approximation capability of the neural networks is limited, another robust compensator is addressed. Four different compensation approaches are discussed in this paper.