Intelligent recognition of ship motion state based on multi-source heterogeneous features of atomic marine gravimeter

The atomic marine gravimeter (AMG) represents the emerging technology and a new trend in dynamic measurement of ocean absolute gravity. It is often a complex integrated instrument that synthesizes multiple subsystems and modules. Beyond its primary function of measuring ocean absolute gravity, the AMG's various components contain rich information about the equipment's operating environment and carrier platform. Based on this, a novel recognition method of ship motion state based on multi-source heterogeneous (MSH) features of atomic marine gravimeter (AMG) was proposed in this study. In this method, 3-axis accelerometer features, 2-axis inclinometer features, and the numbers of total atoms and 3-states atoms were collected, the system's feature characterization ability is enhanced through dimensional transformation, mutual combination, and compensation processes. Hence, the efficient and the accurate identification of the ship motion state was achieved, and the recognition accuracy rate reached 98.9%. Compared with the conventional single- source (SS) feature (3-axis accelerometer, 2-axis inclinometer, the numbers of total atoms and 3-states atoms) methods, their accuracies were increased by 69.2%, 51%, 69.5%, 65.4%, 54.6%, respectively. This research provides valuable insights for AMG data mining, multi-element information integration, and real-time ship motion state assessment. More importantly, it is the first to combine atomic gravity measurement with ship kinematics, achieving a cross-domain integration of two seemingly unrelated technologies and confirming their remarkable performance.