부산대학교 도서관

The global navigation satellite systems (GNSS)/inertial navigation system (INS) integrated navigation system is widely used in the unmanned aerial vehicle (UAV) positioning. However, GNSS signals may be interrupted in complex environments such as signal shielding and occlusion, and the positioning accuracy will rapidly decrease under the independent INS system. To improve the positioning accuracy of UAV during GNSS outages, an error compensation method of GNSS/INS integrated navigation system assisted by long short-term memory (LSTM) neural network integrating attention mechanism (AT-LSTM) is proposed. When GNSS signals are available, the error compensation model between specific force, angular rate, INS attitude, and GNSS position increment is established. When GNSS signals are unavailable, the error compensation model outputs pseudo-GNSS signals to compensate for the integrated navigation system and suppress the divergence of positioning errors. The simulation experiment and actual experiment verify the performance of the error compensation method, and the experimental results show that the positioning accuracy of the AT-LSTM method is improved by more than 90% than that of INS within 60 s of GNSS outages, so the method can effectively improve the positioning accuracy of UAV during GNSS outages.