Variational Bayesian and generalized maximum-likelihood based adaptive robust nonlinear filtering framework

•A variational Bayesian based cubature Kalman filter (VBCKF-R) is derived.•An adaptive robust nonlinear filtering framework is proposed based on the variational Bayesian (VB) method and generalized maximum likelihood estimation (GM estimation).•The VBCEECKF algorithm is developed by applying the CEE...

Full description

Saved in:
Bibliographic Details
Published in:Signal processing Vol. 215; p. 109271
Main Authors: Yang, Baojian, Wang, Huaiguang, Shi, Zhiyong
Format: Journal Article
Language:English
Published: Elsevier B.V 01.02.2024
Subjects:
Adaptive filter
Entropy
Nonlinear filter
Robust filter
Target tracking
Variational Bayesian
Variational Bayesian
Nonlinear filter
Target tracking
Entropy
Adaptive filter
Robust filter
ISSN:0165-1684
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•A variational Bayesian based cubature Kalman filter (VBCKF-R) is derived.•An adaptive robust nonlinear filtering framework is proposed based on the variational Bayesian (VB) method and generalized maximum likelihood estimation (GM estimation).•The VBCEECKF algorithm is developed by applying the CEE criteria to the proposed adaptive robust nonlinear filtering framework.•The proposed VBCEECKF algorithm has high precision in target tracking under non-ideal measurement situations. An adaptive robust nonlinear filtering framework is proposed based on the variational Bayesian (VB) method and generalized maximum likelihood estimation (GM estimation) to simultaneously handle the outliers and uncertain measurement noise covariance matrix (MNCM) for state estimation. This framework utilizes the cubature criterion to solve nonlinear integration problems. By embedding GM estimation in the VB method, the uncertain MNCM is modified while robust estimation is performed. The modified MNCM also provides more accurate model parameters for GM estimation, achieving the unity of adaptability and robustness. The VBCEECKF algorithm was obtained by applying the strong robust centered error entropy criterion to the adaptive robust nonlinear filtering framework. The simulation results of target tracking under different noise conditions verify the superiority of the adaptive robust filter compared to existing methods.
ISSN:0165-1684
DOI:10.1016/j.sigpro.2023.109271