AUV cooperative navigation method of maximum correntropy adaptive cubature particle filter

Abstract

The invention discloses an AUV cooperative navigation method of maximum correntropy adaptive cubature particle filter (MCACPF). The MCACPF is used for completing the state estimation problem in the AUV cooperative navigation process. During AUV cooperative navigation, a state equation and a measurement equation of the AUV cooperative navigation are reconstructed into a nonlinear recursive model, and then maximum correntropy cubature Kalman filter (MCCKF) is adopted in the framework of cubature particle filter (CPF) to generate an importance probability density function required in particle filter (PF), generated particles are then re-sampled by using a Kullback-Leibler distance (KLD) resampling method, and finally the estimation of an AUV state is obtained according to the algorithm flow of the CPF so as to realize the positioning of an AUV and complete the cooperative navigation. The MCACPF method is applied to the AUV cooperative navigation with outliers in measurement noise, higheraccuracy than existing PF, improved PF and robust filtering is obtained, and the computational complexity is lower than an existing improved particle filtering algorithm.

Description

technical field [0001] The invention belongs to the technical field of nonlinear robust filtering and cooperative navigation, and in particular relates to an AUV cooperative navigation method of Maximum Correntropy Adaptive Cubature Particle Filter (MCACPF). Background technique [0002] Using nonlinear filtering, such as Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF), Particle Filter (PF), etc. Vehicles, AUV) cooperative navigation for positioning, due to the water lock phenomenon of the Doppler Velocity Log (DVL), multiple acoustic propagation paths between the sound source and the receiver lead to the reflection of sound waves, causing the sound velocity to vary with Depth changes, reflections on the water surface and the seabed and other reasons will lead to outliers in the measurement noise, which seriously affects the accuracy of multi-AUV cooperative navigation and positioning. To solve this problem, it is proposed to use the nonlinear Filtering (Huber N...

AI Technical Summary

Problems solved by technology

[0004] In addition, the traditional PF and the improved PF have a large amount of calculation when used, so they are no longer suitable for occasions with high real-time requirements

To this end, people proposed a PF algorithm based on Kullback–Leibler distance (Kullback–Leibler Distance, KLD) sampling, but this method assumes that all particles come from the real posterior density function, which is difficult to achieve in practical applications

Images