中文摘要：

讨论了利用滑模观测器实现不确定系统的在线故障重构问题．首先应用奇异值分解，对系统进行规范化处理，设计了鲁棒滑模观测器的LMI算法，并通过非线性介入使设计观测器对系统不确定性具有鲁棒性及跟踪系统状态的收敛性；然后根据滑模观测器设计方法，利用等价输出误差介入、H∞约束优化原理以及加入输出滤波器方法，提出了执行器故障和传感器故障在线重构算法，直接获取故障信息；最后，通过实例给出滑模观测器估计状态以及重构执行器故障的仿真结果，并验证所提方法的有效性．

英文摘要：

This paper considers the problem of fault reconstruction for uncertain dynamical systems by using the sliding-mode observer. First, the system is processed by the canonical transformation using the singular value decomposition （SVD）. The linear matrix inequality （LMI） method of the robust sliding-mode observer is designed, and the nonlinear injection is applied in the observers in order to make the observer have robustness for uncertainties and convergence for tracking states. Then, based on the proposed method of the sliding-mode observer, the actuator fault and sensor fault reconstruction algorithms are developed by using the equivalence output error injection, the optimization concept and the output filter approach in order to directly obtain fault information. Finally, the numerical simulation results of sliding-mode observer estimation states and reconstruction actuators are presented to validate the effectiveness of the proposed method.