摘要
小型无人机传感器的工作状态直接影响着飞行的安全性和稳定性。因小型无人机特殊的工作环境,其故障诊断的难度较大。因此,为了提高小型无人机传感器故障诊断的精确度和适用性,提出一种基于改进粒子滤波算法的小型无人机故障诊断方法。所提出算法结合遗传算法对粒子滤波算法进行了改进,并把传感器故障诊断问题视为多维复合假设检验问题。此外,利用序贯概率比检验法分析粒子滤波估计值与传感器输出值的残差,从而诊断对应的传感器是否发生故障。一旦传感器出现故障,使用广义最大似然法则来辨识具体的故障类型。仿真结果表明:该方法能完成小型无人机的多种故障类型识别,且具有较高的诊断精度。
The working state of the small UAV sensor directly affects the safety and stability of the flight.However,the special working environment of the small drone makes the fault diagnosis more difficult.Therefore,in order to improve the accuracy and applicability of small UAV sensor fault diagnosis,a small particleless fault diagnosis method based on improved particle filter algorithm is proposed.The algorithm is combined with genetic algorithm to improve the particle filter algorithm,and the sensor fault diagnosis problem is regarded as the multi-dimensional compound hypothesis test problem.In addition,the residual probability ratio test is used to analyze the residual of the particle filter estimate and the sensor output value,thereby diagnosing whether the corresponding sensor is faulty.Once the sensor fails,the generalized maximum likelihood rule is used to identify the specific fault type.The simulation results show that the method can effectively accomplish the identification of multiple fault types of small unmanned aerial vehicles and has high diagnostic accuracy.
作者
刘彦超
刘慧文
高薇
李凤银
LIU Yanchao;LIU Huiwen;GAO Wei;LI Fengyin(Automation School,Baotou Light Industry Vocational Technical College,Baotou 014030,China;Electric Power College,Inner Mongolia University of Technology,Hohhot 010051,China;Software College,Yunnan University,Kunming 650091,China;School of Information Science and Engineering,Qufu Normal University,Rizhao 276826,China)
出处
《重庆理工大学学报(自然科学)》
CAS
北大核心
2019年第10期174-179,共6页
Journal of Chongqing University of Technology:Natural Science
基金
内蒙古自治区自然科学基金面上项目(2016MS0504)
内蒙古自治区教育厅课题资助项目(NJZY17530,NJZY11271)
关键词
小型无人机
传感器故障诊断
粒子滤波算法
遗传算法
序贯概率比检验
small unmanned aerial vehicle
sensor fault diagnosis
particle filter algorithm
genetic algorithm
sequential probability ratio test