Novel interacting multiple model filter for uncertain target tracking systems based on weighted Kullback–Leibler divergence

Bowen Hou, Jiongqi Wang, Zhangming He, Yongrui Qin, Haiyin Zhou, Dayi Wang, Dong Li

Research output: Contribution to journalArticlepeer-review

Abstract

Interacting multiple model (IMM) filter is a classical method to track targets in hybrid situations. However, it can exhibit divergence when the models are correlated or the system suffers from uncertainties. The generalized covariance intersection method based on the weighted Kullback–Leibler (K–L) divergence can solve the divergence problem of correlated estimates. A novel interacting multiple model (NIMM) filter is presented that combines two different algorithms, the adaptive fading Kalman filter and the maximum correntropy Kalman filter, based on the model interacting with the weighted K-L divergence to address the uncertainty problems of the system. The NIMM filter algorithm is designed and the stability and accuracy are analyzed. The simulation results demonstrate that the proposed filter can effectively improve the accuracy under different uncertainty conditions for classical examples and ballistic trajectory tracking scenarios.

Original languageEnglish
Pages (from-to)13041-13084
Number of pages44
JournalJournal of the Franklin Institute
Volume357
Issue number17
Early online date17 Sep 2020
DOIs
Publication statusPublished - 1 Nov 2020

Fingerprint

Dive into the research topics of 'Novel interacting multiple model filter for uncertain target tracking systems based on weighted Kullback–Leibler divergence'. Together they form a unique fingerprint.

Cite this