为提升无人机飞行稳定性与安全性,基于小波阈值去噪、遗传算法和 BP 神经网络构建两种故障检测方法,并将其应用于无人机飞行过程中的传感器监测数据的故障检测识别中.结果表明:融合遗传算法+BP 神经网络检测方法的收敛速度显著快于融合小波阈值去噪+BP 神经网络检测法,且前者融合算法的平均识别正确率也略高于后者;两种算法对于突变故障或者正常状态识别效果较好,但对于慢变故障的识别效果较差.由于融合遗传算法+BP神经网络检测方法具有更快的检测速度和检测精度,故优先推荐此方法应用到无人机飞行过程故障检测识别中.
Research on Fault Detection and Identification Method of UAV Flight Process
To improve the stability and safety of drone flight,two fault detection methods were constructed based on wavelet threshold denoising,genetic algorithm,and BP neural network,and applied to the fault detection and recognition of sensor monitoring data during drone flight.The results show that the convergence speed of the fusion genetic algorithm+BP neural network detection method is significantly faster than that of the fusion wavelet threshold denoising+BP neural network detection method,and the average recognition accuracy of the former fusion algorithm is also slightly higher than the latter.The two algorithms have good recognition performance for abrupt fault or normal states,but poor recognition performance for slow changing faults.Due to the faster detection speed and accuracy of the fusion genetic algorithm+BP neural network detection method,it is recommended to prioritize its application in fault detection and recognition during drone flight.
NETL
NSTL
万方数据
