一类伴有不匹配未知干扰的线性系统的故障检测
Fault detection for linear systems with mismatched unknown inputs

摘要 针对伴有不匹配未知干扰的线性时不变系统，本文将降维观测器的设计方法和未知干扰建模技术相结合，设计出对未知干扰解耦的故障检测观测器.首先，将部分匹配未知干扰从未知干扰中分离出来，剩余的外部干扰部分即为不匹配未知干扰；然后，依据先验知识对不匹配未知干扰代数建模，构建增广系统，使得秩条件得以满足；最后，针对增广系统设计故障检测观测器,使得残差对未知干扰鲁棒，对故障敏感.通过对一个机器人子系统实例进行仿真，仿真结果表明本文所提出的方法是有效的.
Abstract：For linear time-invariant systems with mismatched unknown inputs, we present an approach to design Fault Detection Observer(FDO) through using the unknown disturbances modelling technique. The unknown disturbances may be decomposed into matched unknown disturbances and mismatched unknown disturbances. Under the assumption that mismatched unknown disturbances can be viewed as a fictitious autonomous dynamical system, an augmented system is obtained by combining the original system model with unknown disturbances model such that the rank condition is satisfied. Then a reduced-order FDO is constructed and the residual is generated which is robust to unknown disturbances while being sensitive to the faults. The proposed method is simulated to an example of a robot sub-system and the stimulation results demonstrate its effectiveness.