卫星多故障探测和识别的独立分量分析法

鉴于独立分量对异常值具有较强的敏感性,提出了基于独立分量分析(ICA)的伪距多变量时间序列异常值探测算法,并且利用契比雪夫不等式给出了异常值探测的阈值,引入时间序列干预模型估计了潜在故障扰动的大小,根据3σ准则确定出故障星的位置。根据RAIM的实时性要求,采用滑动窗口的思想对上述批处理探测算法进行改造,本文提出了一种卫星多故障在线探测和识别的新算法,并给出了新RAIM算法的实施流程。利用5个iGMAS北斗监测站的民用观测数据对新算法进行验证,试验分析结果表明,新算法对于卫星多故障的实时处理具有较好的效果,且其故障正确探测率优于以往的RANCO方法。

Multiple Satellite Faults Detection and Identification Based on the Independent Component Analysis

Considering that the independent component is sensitive to outliers, we propose an algorithm for faults detection in multivariate pseudorange time series based on independent component analysis (ICA).The threshold for outlier detection is determined through the Chebyshev inequality.Then we introduce the interventional model of time series to estimate the magnitudes of the potential satellite faults, and finally the satellite faults are identified based on the 3σ principle.In order to meet the real time requirement of receiver autonomous integrity monitoring (RAIM), a sliding window is used to transform the fault detection algorithm of the batch process into a real time one.Furthermore, a new algorithm for on line detection and identification of multiple faults is designed, and then the implementation process of the new RAIM algorithm is given.We validate the new algorithm by the civil data from 5 iGMAS monitoring stations of BeiDou in China.Examples illustrate that the new algorithm is effective in handling multiple satellite faults in real time, and the correct detection probability of faults is higher than that of the existed RANCO algorithm.