针对一类仿射非线性动态系统，提出一种基于神经网络非线性观测器的鲁棒故障检测与隔离的新方法．采用ＲＢＦ神经网络逼近观测器系统中的非线性项，提高了状态估计的精度，证明了状态估计误差稳定且渐近收敛到零；同时提出了一种新的网络权值调整指标方法，提高了神经网络故障分类器的泛化能力，从而保证该方法对被监测系统的建模误差和外部扰动具有良好的鲁棒性 Aiming at a class of affine nonlinear dynamic systems, a new method of robust fault detection and isolation based on neural network nonlinear observer is proposed. The RBF neural network is used to approximate the nonlinear term in the observer system, which improves the accuracy of the state estimation. It is proved that the state estimation error is stable and asymptotically converges to zero. At the same time, a new method of network weight adjustment index is proposed, The generalization ability of neural network fault classifier, so as to ensure the robustness of the proposed method to modeling errors and external disturbances of the monitored system