-

Nonlinear Filtering Based on Moment Approximation and Gaussian Approximation

组织者

焦小沛 , 康家熠

演讲者

胡三丰

时间

2025年05月22日 14:30 至 16:00

地点

A3-2-303

线上

Zoom 435 529 7909 (BIMSA)

摘要

This report investigates the estimation problem for discrete-time nonlinear stochastic dynamic systems. Within the moment approximation framework, a third degree moment-approximation based filtering algorithm is developed using a sequential homotopy method under a linear estimation structure. By propagating higher-order statistical information, the proposed filter improves estimation accuracy and overcomes the limitation of conventional linear estimation-based filters, which can only propagate the first two moments. Within the distribution approximation framework, two Gaussian filtering algorithms are proposed based on the minimization of Kullback-Leibler divergence. These algorithms transcend the constraints of linear estimation frameworks, offering high estimation accuracy while maintaining low computational complexity.

演讲者介绍

Sanfeng Hu received the B.S. degree in computer science and technology, the M.S. degree in applied mathematics, and the Ph.D. degree in mathematical statistics from Sichuan University, Chengdu, China, in 2015, 2018, and 2022, respectively. Since 2022, he has been with the School of Mathematics and Statistics, Yunnan University, Kunming, China, where he is currently a Lecturer. His research interests are in the fields of estimation, target tracking, information fusion, optimization algorithms, and machine learning.