基于奇异值分解的接收机自主完好性监测算法

基于最小二乘法残差的接收机自主完好性监测（receiver autonomous integrity monitoring，RAIM）算法本质是一种基于伪距残差矢量的一致性监测算法，但由于残差矢量中各分量具有一定的关联性，掩饰了某些重要的不一致性信息。为了消除这种关联性，提出了一种基于奇异值分解的接收机自主完好性监测方法。在方法中利用奇异值分解对伪距观测矩阵中的观测系数矩阵进行分解，获得奇异值空间矢量和奇异值空间矩阵。基于奇异值空间矢量构造能够直接反映故障卫星偏差信息的检验统计量，从而可以简便地进行粗差监测，更好地满足完好性监测需求。鉴于实际中完好性故障包含运控系统故障、导航系统故障、信号传播异常以及地面接收处理故障等多类因素，以脉冲型和阶跃型两种故障方式进行基于奇异值分解的RAIM故障检测与识别，并开展仿真分析研究。结果表明，提出的方法能够正确检测、识别故障卫星，在特定参数下能够达到很好的故障识别率，即当误警概率设置为1×10-5/h、引入阶跃故障误差为25 m时，算法能够实现98.8%的故障识别率。

Receiver autonomous integrity monitoring algorithm based on singular value decomposition

The receiver autonomous integrity monitoring (RAIM) algorithm based on least squares residuals is essentially a consistency monitoring algorithm based on pseudo-range residual vector. However, due to the correlation of each component in the residual vector, some important inconsistent information is covered up. In order to eliminate this correlation, a receiver autonomous integrity monitoring method based on singular value decomposition is proposed.In this method, the singular value decomposition is used to decompose the observation coefficient matrix in the pseudo-range observation matrix to obtain singular value space vector and singular value space matrix. Based on the singular value space vector, the test statistics that could directly reflect the deviation information of the fault satellite could be constructed, so that the gross error could be easily monitored and the requirements of integrity monitoring could be better met. In view of the fact that the integrity fault includes operation control system fault, navigation system fault, signal propagation abnormalities and ground receiving and processing fault, the RAIM faults detection and identification based on singular value decomposition are carried out in two fault modes of pulse and step fault, and the simulation analysis and researched are performed. The results show that the method proposed can correctly detect and identify faulty satellites, and achieve a good fault identification rate under specific parameters, that is, when the false alarm probability is set to 1×10-5/h and the step fault error introduced is 25 m, the algorithm can achieve a fault identification rate of 98.8%.