Adaptive observer design for nonlinear systems with nonparametric uncertainties
非参数不确定性非线性系统的自适应观测器设计

Based on neural networks, we design an adaptive observer for a class of systems with nonparametric uncertainties, improving the traditional methods which fail in designing the adaptive observers for such a class of systems. First, a neural network is used to approximate the uncertainties; and then, a filtered transformation of the system is performed by using the vector of the basic functions of the neural network. Based on the transformed system, the observer for the original system is constructed. An estimation of the observation error is also given. The result of the paper shows that the observation error value can be made arbitrarily small by reducing the error in the approximate uncertainties and properly choosing the design parameters of the observer.