A fast intelligent algorithm based positioning error suppression method for strap-down inertial navigation system

• Published in: IEEE sensors journal Vol. 23; no. 5; p. 1
• Main Authors: Hongyang, He, Bao, Li, Feng, Zha, Hongqiong, Tang
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.03.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Damping; Divergence; Earth; Field tests; Gyroscopes; Inertial coordinates; Inertial navigation; Navigation; Navigation systems; Particle swarm optimization; positioning error suppression; Sensors; SINS; Training
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2023.3237365

It is a difficult problem to suppress navigation errors of strap-down inertial navigation system (SINS) when external observation is rejected or discontinuous. With the help of intelligent algorithms, the device errors of SINS can be searched and compensated when the vehicle has no significant linear motion, so as to limit the positioning error of INS to a certain extent in case of external observation rejection. However, the inertial navigation system contains long period error terms. In order to suppress the navigation errors, long time data needs to be observed and then trained by intelligent algorithms, which limits the usability of the correlation error suppression algorithms in practice. In view of this problem, this paper proposes a fast intelligent algorithm (FIA) based positioning error suppression method for inertial navigation systems. By virtually extending the inertial navigation update interval, the divergence law of the longitude error could be changed. On this basis, a fast training algorithm based on particle swarm optimization is designed. The contributions of the work presented here are twofold. First, the mathematical relationship between the positioning error and the device error of SINS with virtually extended update interval is analyzed. Secondly, compared with the traditional methods, the amount of data required for training is greatly reduced. Therefore, the device errors of INS could be searched and compensated with short term observation data. And then, the long-term positioning accuracy of INS can be improved efficiently. Field tests are performed, which validate the efficacy of the proposed method.