基于抗差估计和改进AIME的缓变故障检测方法

为了解决经典缓变故障检测法——自主完好性监测外推法(autonomous integrity monitoring extrapolation, AIME)在组合导航中检测故障延迟时间较长和不能准确判定故障结束时刻的问题, 提出了基于抗差估计和改进AIME的缓变故障检测方法。该方法采用标准t分布和IGG-Ⅲ (Institute of Geodesy & Geophysics Ⅲ)方案设计自适应增益矩阵以缓解卡尔曼滤波故障跟踪的影响。同时, 结合AIME故障检测状态, 提出由外推法和残差卡方检验(residual chi-square test method, RCTM)故障检测统计量构成的r_A/R统计量概念, 然后在AIME检测到缓变故障的状态下利用样本分位数原理对r_A/R序列进行异常值检测, 从而判断缓变故障结束时刻。仿真结果表明, 在检测缓变故障时, 所提方法可明显缩短故障检测延迟时间, 并能够准确判定故障结束时刻。

Approach for detection of slowly growing fault based on robust estimation and improved AIME

An improved slowly growing fault detection method which is based on robust estimation and modified autonomous integrity monitoring extrapolation (AIME) is developed in this paper for the purpose of solving the problems that conventional AIME method for integrated navigation system has a big detection delay and it can not judge the ending time of the fault. The method mitigates the influence of fault tracking of Kalman filter by designing an adaptive gain matrix based on standard t-distribution and IGG-Ⅲ (Institute of Geodesy & Geophysics Ⅲ) scheme and proposes the concept of r_A/R statistics, which is composed of fault detection state of AIME and residual chi-square test method (RCTM) fault detection statistics. Then, the outlier of r_A/R sequence is detected through using sample quantile principle, so as to determine the ending time of the slowly growing fault when AIME method has detected the existence of the fault. The simulation result shows that the proposed method can reduce the delay time of fault detection significantly as well as accurately determine the ending time of fault when a slowly growing fault occurs on the integrated system.