多路径误差是北斗导航定位系统高精度动态监测的主要误差源.针对北斗导航定位系统多路径误差的特性,结合广义特征值盲源分离方法的优势,提出一种基于参考信号的广义特征值盲源分离算法来削弱多路径效应的影响.首先将前一天的原始坐标残差序列通过奇异谱分析方法进行去噪,其结果作为初始参考信号;然后将当天的原始坐标残差序列进行经验模式分解方法分解,分解得到的IMF分量作为虚拟观测数据,利用广义特征值盲源分离算法获取当天多路径误差信号;最后,利用仿真数据和连续10天的实际观测数据进行试验分析,结果表明利用该方法建立的多路径误差改正模型能有效地了削弱多路径的影响,北、东、天三个方向精度分别提高了78.8%、35.3%、90.1%.提出的模型在一定程度上解决了固定多路径模型随着时间推移重复性减小且有效性降低的问题.
Multipath error is the main error source of Beidou navigation satellite system (BDS) in high precision dynamic detection. According to the BDS multipath error characteristic and combined with the advantages of the generalized eigen-value decomposition blind source separation method, a method for mitigating the effects of multipath effects is proposed based on the generalized eigen-value blind source separation algorithm of the reference signal. Firstly, the original coordinate residual sequence from the previous day is denoised by the singular spectrum analysis (SSA) method, and the result is taken as the initial reference signal. Afterwards the empirical mode decomposition (EMD) of the original coordinate residuals of that day is carried out, and the Intrinsic Mode Function component is decomposed as the virtual observation data. Finally, the generalized eigen-value blind source separation algorithm is used to obtain the multi-path error signal. The simulation results and the analysis of the actual 10-day observation data show that the multipath error correction model established by this method can effectively eliminate the influence of multipath, and the precision of N, E and U directions are improved by 78.8%, 35.3%, and 90.1%, respectively. The proposed model solves the problem that the repeatability and effectiveness of fixed multipath model decreases with time.