引用本文:杨春,张磊,郭健,陈庆伟.采用双状态传播卡方检验和模糊自适应滤波的容错组合导航算法[J].控制理论与应用,2016,33(4):500~511.[点击复制]
YANG Chun,ZHANG Lei,GUO Jian,CHEN Qing-wei.Fault-tolerant integrated navigation algorithm using chi-square test with two state propagators and fuzzy adaptive filter[J].Control Theory and Technology,2016,33(4):500~511.[点击复制]
采用双状态传播卡方检验和模糊自适应滤波的容错组合导航算法
Fault-tolerant integrated navigation algorithm using chi-square test with two state propagators and fuzzy adaptive filter
摘要点击 4256  全文点击 1362  投稿时间：2015-04-23  修订日期：2015-10-31
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2016.50328
  2016,33(4):500-511
中文关键词  组合导航  联邦卡尔曼滤波  故障检测  故障诊断  双状态传播卡方检验  模糊自适应
英文关键词  integrated navigation  federal Kalman filter  fault detection  fault diagnosis  two state propagator chi-square test (TSPCST)  fuzzy adaptive
基金项目  国家自然科学基金项目(61074023)资助.
作者单位E-mail
杨春* 南京理工大学 自动化学院 yangguang326@126.com 
张磊 南京理工大学 自动化学院  
郭健 南京理工大学 自动化学院  
陈庆伟 南京理工大学 自动化学院  
中文摘要
      太阳能高空长航时无人机导航系统中, 捷联惯导/北斗2/全球卫星导航/星光导航(SINS/BD2/GPS/CNS)是一 种可用的组合方案. 针对常规容错组合导航算法故障检测类型单一, 故障时滤波精度下降的问题, 提出一种采用双 状态卡方检验(TSPCST)和模糊自适应滤波(FAF)的容错组合导航算法. 为了同时检测多种故障, 将TSPCST应用于 联邦滤波结构中; 为了防止故障数据污染系统, 利用FAF输出的高精度导航信息, 对双状态传播器定期交替校正; 进 一步, FAF运用TSPCST检测得到的故障信息变量, 定义量测子系统模糊有效域, 将检测阈值模糊化, 以弥补常规固 定检测阈值算法难以选取阈值的不足; 最后, 通过计算信息分配因子, 自适应处理多种故障数据. 仿真结果表明, 该 容错组合导航算法性能优于常规固定检测阈值算法.
英文摘要
      SINS/BD2/GPS/CNS is a usable integrated scheme which is suitable for the navigation system of the highaltitude long-endurance solar-powered UAV. The normal fault-tolerant integrated navigation algorithm can only detect the single fault and the precision deteriorates in faulty operations. To solve these problems, a fault-tolerant integrated navigation algorithm is proposed using chi-square test with two state propagators (TSPCST) and the fuzzy adaptive filter (FAF). First, with the application of the TSPCST in the structure of the federal Kalman filter, the algorithm can detect multiple types of measurement faults at the same time. Then, these two state propagators are alternatively reset in a fixed period with the high precision navigation information provided by the FAF. By this method, the risk of using a contaminated data in the system is avoided. Furthermore, the FAF utilizes the fault context variables obtained by the TSPCST to define the fuzzy validity domains of each subsystem. By this method, the detection threshold is blurred to make up the drawback that it is difficult for the conventional algorithm with the fixed detection threshold to choose the threshold. Finally, the algorithm can process the failures adaptively through calculating the information distribution factor. Simulation results show that the performance of this algorithm is better than the conventional ones with the fixed detection threshold.