Quasi-Newton Cubature Kalman Filitering Method Based on BFGS Correction Formula for Attitude Determination Applied to Underwater Glider

Abstract

The performance of Micro-Electro-Mechanical System sensors is degraded seriously and errors will become much larger in the complex underwater environment, the measurements become unobservable particularly when the magnetic sensor is disturbed by external magnetic interference, resulting in the navigation information determined erroneously. In order to deal with the problem mentioned, the paper proposes a novel Quasi-Newton CKF method based on BFGS Correction Formula. This method makes better use of the superiority of the anti-interference ability and the computation efficiency to obtain navigation information with high accuracy for a long time of gliding. Meanwhile, BFGS Correction Formula, which is used in Quasi-Newton method, adopts the inaccurate linear search to improve the computational efficiency further.