中文摘要：

对于卫星姿态控制系统，提出一种基于PD型学习观测器（Learning observer，LO）的系统故障重构方法。在P型LO的学习算法基础上引入测量输出估计误差的微分项，设计了一种PD型LO，估计卫星姿态角速度和姿态角的同时，快速精确重构卫星执行机构故障。给出了所提观测器的稳定性条件，并基于线性矩阵不等式技术提出一种系统化PD型LO设计方法。进一步，将所提PD型LO设计扩展用于卫星姿态敏感器故障的快速重构。最后，将所提方法应用于微小卫星推力器故障重构和陀螺故障重构，仿真结果校验了所提方法的有效性。

英文摘要：

The problem of PD-type Learning Observer （LO） -based fault reconstruction is addressed for satellite attitude control systems in this paper. By introducing differential term of measurement output estimation error into learning updating algorithm of the existing P-type LO, a novel PD-type LO is designed such that it can not only estimate satellite attitude angular velocities and attitude angles simultaneously, but also reconstruct actuator faults fast and accurately. Then, stability conditions of the proposed PD-type LO are explicitly provided, and a systematic PD-type LO design method is given based on linear matrix inequality technique. Further, the proposed PD-type LO design is extended for reconstructing attitude sensor faults in satellite. At last, the proposed approaches are applied to reconstruct thruster faults and gyroscope faults respectively in microsatellites, and simulation results validate the effectiveness of the proposed fault reconstruction approaches.