INS/GNSS/VO组合导航系统复合型异常检测与容错算法

Compound anomaly detection and fault tolerance algorithm for INS/GNSS/VO integrated navigation system

INS/GNSS/VO导航系统的异常输出常由外部扰动引起,表现为幅值较小的突变信号或者增长速率较慢的斜坡信号。为了增强异常检测算法的检测能力,提高组合导航系统在外部干扰情况下的导航精度,提出一种INS/GNSS/VO组合导航系统复合型异常检测与容错算法。在非全局融合阶段使用状态卡方检测法对量测信息进行异常检测与隔离;在全局融合阶段,构建检测统计量的归一化阈值比,作为量测噪声方差阵的权重系数,对子滤波器进行加权量测更新。街道场景中的离线测试结果表明,所提算法的导航误差均值与标准差相对于其他的异常检测与容错算法分别减小了9.6%与2.7%,提高了组合导航系统的异常检测能力和容错性能。

Anomaly outputs of INS/GNSS/VO navigation system are often caused by external disturbances,which are abrupt signals with small amplitude or slope signals with slow growth rate.In order to enhance the detection ability of anomaly detection algorithm and improve the navigation accuracy of integrated navigation system in the case of external disturbances,a compound anomaly detection and fault tolerance algorithm for INS/GNSS/VO integrated navigation system is proposed.In the non-global fusion stage,the state Chi-square detection method is used to detect and isolate the anomaly observations.In the global fusion stage,the normalized threshold ratio of detection statistics is constructed as the weight coefficient of measurement noise variance matrix,and the weighted measurement update is carried out on the subfilter.The offline test results in the street scenes show that the navigation error mean and standard deviation of the proposed algorithm are reduced by 9.6%and 2.7%respectively compared with other anomaly detection and fault tolerance algorithms,which improves the anomaly detection ability and fault tolerance performance of the integrated navigation system.