Abstract: The traditional methods for Unmanned Aerial Vehicles（UAV） Flight Control System（FCS） sensor fault detection and diagnosis are based on the accurate analytical FCS model, which is obtained hardly. A UAV-PCA algorithm is presented in this paper, which combines variance sensitive adaptive threshold and the direction of fault feather to complete the UAV flight control system sensor fault detection and diagnosis. The algorithm doesn’t need any system analytical models, and resolves the problems, such as the false alarms in the transient state and the false diagnoses. These problems occur when traditional PCA method is used for FCS fault detection and diagnosis. The simulation results show that the UAV-PCA method has less false alarm in the transient state and more correct diagnoses than the traditional PCA methods.

Key words: Unmanned Aerial Vehicles（UAV）, fault detection and diagnosis, Principal Component Analysis（PCA）, flight control system, variance sensitive adaptive threshold, direction of faults

摘要： 无人机（UAV）飞控系统传感器故障检测和诊断常采用基于解析模型的方法，但飞行控制系统（FCS）的精确数学模型往往获取困难。针对此问题，提出了一种UAV-PCA算法。该算法在传统主成分分析（PCA）方法的基础上结合方差敏感自适应阈值的故障检测方法和基于特征方向的故障诊断方法，实现UAV飞控系统传感器的故障检测和诊断。算法不需要系统的数学模型，解决了应用传统PCA方法进行FCS故障检测与诊断时易出现暂态过程虚警和误诊断的问题。仿真结果证明该算法可以快速准确地检测传感器故障，而且可以有效地降低暂态过程虚警和提高诊断结果准确度。

关键词: 无人机, 故障检测与诊断, 主成分分析, 方差敏感自适应阈值, 故障特征方向法