Offline outlier identification for dynamic measurements in underwater geomagnetism navigation

Underwater geomagnetism-aided navigation now is advanced as a leading subject in navigation research field, for its merits of sensing-passivity, distribution-extensity and etc. But yet geomagnetic fieldpsilas unsteadiness may cause positioning resultspsila divergence after correlation matching, so the dynamic measurementspsila gross errors make this project an issue. After analyzing the specialties of underwater vehiclespsila slow sailing and marine magnetometerspsila quick sampling, this paper deduces that real-surveyed total intensity sequence can be considered as nonstationary stochastic process. Based on the procedure of inertial navigation system (INS) rectified by matching algorithm intermittently, combination outlier identification algorithm in offline mode (COIO) is advanced. COIO includes segmentation fault-tolerant gross-error identification method (SFTGI) for the sections before matching sequencepsilas end point and wavelet multiscale analysis method for checking the dubious outliers picked out above. Then one-step prediction judgment (OSPJ) is assumed for end point. Then comparison with Least Square (LS) algorithms and simulation experiments show that, outliers can be detected from dynamic surveying sequence more effectively by COIO. The divergence problem of correlation matching algorithmpsilas result can be avoided, and the efficiency of INSpsilas rectification also can be improved.