论文部分内容阅读
针对一类仿射非线性动态系统，提出了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．该方法采用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，并从理论上证明了状态估计误差稳定且渐近收敛到零；另一方面引入神经网络分类器进行故障的模式识别，通过在神经网络输入端加入噪声项来进行训练，提高神经网络的泛化逼近能力，从而保证对被监测系统的建模误差和外部扰动具有良好的鲁棒性．最后，利用本文方法针对某型歼击机结构故障进行仿真验证，仿真结果表明本文方法是有效的．
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. This method uses neural network to approximate the nonlinear term in the observer system and improves the accuracy of the state estimation. It is proved theoretically that the state estimation error is stable and asymptotically converges to zero. On the other hand, the neural network classifier is introduced to perform faulty Pattern recognition is carried out by adding noise term at the input of neural network to improve the generalization and approximation ability of neural network so as to ensure good robustness to modeling errors and external disturbances of the monitored system. Finally, the method of this paper is used to verify the structural failure of a fighter plane. The simulation results show that the proposed method is effective.