基于主成分分析的车辆组合导航系统故障检测与隔离算法

Principal Component Analysis Based Fault Detection and Isolation Algorithm for Integrated Vehicle Navigation System

故障检测与隔离是保证组合导航系统容错性能的重要措施.本文针对车辆导航系统的可靠性及安全性需求,分析了组合导航系统的传感器故障特性,建立了包含三个数据通路的车辆组合导航系统容错设计结构,提出了一种基于主成分分析的故障检测与隔离算法.该算法利用数据的相关关系,采用正常条件下的导航历史数据建立统计PCA模型,通过一定的统计量控制限检验新的导航数据样本相对于模型的背离程度实现故障检测,并根据不同数据通路的故障检测结果动态调整导航子系统的有效性因数,从而隔离故障造成的影响.仿真结果表明,该算法实现简单,能够有效实现对组合导航系统故障的检测与隔离.

Fault detection and isolation is crucial to support the fault-tolerance of integrated navigation sys- tem. Based on the requirements of reliability and safety for vehicle navigation, the characteristics of naviga- tion sensors are analyzed, the structure of integrated navigation system is designed with three data channels, and a fault detection and isolation algorithm is proposed based on principal component analysis for integrated vehicle navigation. In this method, considering the correlation of the navigation data, statistical principal component analysis （PCA） models are developed with historical navigation data under normal conditions, the fault detection is realized by verifying the deviation of new navigation data from the PCA models with con- trol limits of certain statistics. The detection results of different data channels can be used for the dynamical adjusting of efficiency factor in navigation subsystem by prior strategies, and then fault can be isolated effi- ciently. The simulations results illustrate that the proposed algorithm can realize the fault detection and isola- tion efficiently with simple implementation.