A Fast Yaw Alignment Algorithm of Seaplane Based on LSTM Network

Initial alignment must be carried out before the formal operation of the airborne strapdown inertial navigation system (SINS) on seaplane. The complex wind and wave conditions on water surface can cause angular swaying and large linear motion disturbance, which make it tough for seaplane to realize rapid and accurate self-alignment. When the inertial-frame self-alignment algorithm is applied to alignment, the horizontal misalignment angles can quickly converge with high accuracy, while the yaw misalignment angle is slow to converge. Compared with other neural network models, LSTM(Long Short-term Memory) networks are universally used in long-term sequence prediction problems. Thus, an LSTM network is introduced in this paper to train inertial sensor data, and then the trained network is adopted for yaw alignment. Simulation results indicate that the proposed algorithm in this paper has high alignment accuracy and speed, and maintains good robustness to the changes of motion parameters.