【摘要】 针对一般非线性系统的故障检测 ,从工程应用的角度提出了一种基于自适应模糊输出观测器的非线性系统鲁棒故障检测方法。该方法以自适应模糊系统构造未知非线性模型的输出观测器 ,在充分考虑外加噪声干扰和系统误差的情况下 ,通过对一般反向传播学习算法进行改进 ,提出采用鲁棒反向传播学习算法调整观测器参数以辨识系统输出 ,再结合阈值故障检测策略检测系统故障。为保证算法具有较快的收敛速度 ,本文给出了根据模糊规则确定算法初始参数的选择方法并证明了算法的收敛性。仿真结果表明 ,对一般非线性系统故障检测 ,该方法具有有效性和实时性 ,以及对噪声干扰和系统误差的鲁棒性更多还原

【Abstract】 Aiming at the fault detection and its actual application in general nonlinear systems, a robust fault detection scheme based on adaptive -fuzzy output observer is proposed. The adaptive-fuzzy systems are designed as an output observer in nonlinear systems to identify its output, and fault detec tion based on the observer is achieved by the threshold logic. When modeling err ors and disturbances exist in unknown nonlinear systems, a robust back propagati on(RBP) adjusting algorithm is presented by an improving ordinary BP algorithm. The constringency of RBP algorithm is proved, and the initialization based on fu zzy rules is selected to guarantee that the constringent speed of the algorithm is faster. Finally, simulation results demonstrate that the fault detection is e ffective and robust.更多还原