论文部分内容阅读
通过推进系统外部可测的姿态信息，利用强跟踪滤波器和多模预报估计器，研究了一类非线性系统常值阶跃型故障和间歇性故障的检测与分离问题。对关机失效，利用强跟踪滤波器获得故障效应的估计，检测逻辑采用改进的贝叶斯分类方法，依据推力器失效效应的方向性进行故障单元的分离；对开机失效，采用多模预报，根据相应残差向量范数的比较进行故障的即时无阈值判别，同时完成了故障后系统状态的重构问题。以一类大型航天器的反作用姿态控制系统为例，对其中出现的推力器失效进行了故障识别仿真研究，表明所提方法是有效的。
By advancing the detectable attitude information outside the system and using the strong tracking filter and the multimode prediction estimator, the problem of detection and separation of constant-value step-type faults and intermittent faults in a class of nonlinear systems is studied. For the shutdown failure, a strong tracking filter is used to obtain the estimation of the fault effect. The detection logic uses an improved Bayesian classification method to separate the faulty units according to the directionality of the thrusters failure effect. The corresponding residual vector norm comparison of fault instantaneous without threshold discrimination, at the same time to complete the reconstruction of the system state after the fault problem. Taking a kind of large-scale spacecraft’s reaction attitude control system as an example, the fault diagnosis of the thrusters in the spacecraft was simulated and the results show that the proposed method is effective.