基于小波与GBDT的无人机传感器故障诊断

相对于有人飞行器,确保无人机传感器的正常工作更为重要.针对无人机传感器的故障诊断,提出了一种将小波特征提取与梯度提升决策树(GBDT)算法相结合的故障诊断方法.采用基于多层小波包分解的特征提取方法,将小波包分解系数与频带能量熵组合构成特征向量,相比单一的能量特征提取方法,有效提升了故障的可分性.采用梯度提升的策略对弱分类器进行迭代优化和线性组合,构成强分类器,使故障分类精度得到显著提高.仿真结果表明,该方法能有效进行特征提取和故障类型识别,且有较高的诊断精度和较强的泛化能力.

Fault Diagnosis of UAV Sensors Based on Wavelet and GDBT

Compared with manned aircraft,it is more important to ensure the normal work of UAV sensors.In order to solve the problem of the fault diagnosis of UAV sensors,a fault diagnosis method of the sensors on UAV which combines wavelet and gradient boosting decision tree (GBDT) technique is presented.Initially,the feature of original output signals is extracted by multi-level wavelet package decomposing,the wavelet packet decomposition coefficient and frequency band energy entropy are combined to form a feature vector,which effectively improves the separability of the fault compared with single energy feature extraction methods.Subsequently,the gradient boosting strategy is used to optimize and linearly combine the weak classifier by iteration to form the strong classifier,the fault classification accuracy is significantly improved.The simulation results show that the proposed method can identify fault types effectively,and improve the diagnostic accuracy and generalization ability.