Research into the high-precision marine integrated navigation method using INS and star sensors based on time series forecasting BPNN

The integrated navigation method based on Inertial Navigation System (INS) and star sensor is a fully autonomous navigation technology. The algorithm employs navigation information from a star sensor as a benchmark to perform periodic recalibration of INS divergent position error. However, when Autonomous Underwater Vehicle (AUV) floats to the surface of the ocean, it displays a rocking motion, due to waves and other factors. This motion results in the star map acquired by star sensor appears trailing phenomenon. Thus, the navigation information obtained from star sensor cannot be used as benchmark information. To solve this problem, an integrated navigation method based on INS and star sensor using the Adaptive Differential Evolution Back Propagation Neural Network (ADE-BPNN) to predict carrier attitude is proposed in this paper. The ADE-BPNN time series forecasting technology enables the use of data describing the integrated navigation attitude over a period of time to predict the attitude of the carrier when a star map appears the trailing phenomenon. The effectiveness of this approach was demonstrated by simulation and experimental study. The results show that this approach can estimate the navigation errors of star sensor and INS. Therefore, the INS/star sensor integrated navigation accuracy is improved.