时空学习驱动的混合重要性高斯滤波

张宏伟

doi:10.12263/DZXB.20230056

您当前的位置：

首页 >

文章列表页 >

时空学习驱动的混合重要性高斯滤波

更新时间：2023-10-20

- 时空学习驱动的混合重要性高斯滤波
- Spatiotemporal Learning Driven Mixture Importance Gaussian Filtering
- 电子学报 2023年页码：1-8
- 作者机构：
  
  中山大学航空航天学院，广州深圳 518107
- 作者简介：
  
  [ "张宏伟女，1982年生，河南南阳人. 中山大学航天航空学院，副研究员，博士.主要研究方向：智能信息处理，信息融合，目标跟踪.E-mail: zhanghw69@mail.sysu.edu.cn" ]
- 基金信息：
  
  广东省基础与应用基础研究基金(2019A1515111099);中山大学青年培育项目(20lgpy72);中国科学院空间精密测量重点实验室开放基金项目(SPMT2021002;SPMT2022001)
- DOI：10.12263/DZXB.20230056
  中图分类号： TN953;
- 收稿：2023-01-17，
  
  修回：2023-04-25，
  
  网络出版：2023-10-20，
- 稿件说明：
移动端阅览
张宏伟. 时空学习驱动的混合重要性高斯滤波[J/OL]. 电子学报, 2023,1-8.

ZHANG Hong-Wei. Spatiotemporal Learning Driven Mixture Importance Gaussian Filtering[J/OL]. ACTA ELECTRONICA SINICA, 2023, 1-8.
张宏伟. 时空学习驱动的混合重要性高斯滤波[J/OL]. 电子学报, 2023,1-8. DOI： 10.12263/DZXB.20230056.

ZHANG Hong-Wei. Spatiotemporal Learning Driven Mixture Importance Gaussian Filtering[J/OL]. ACTA ELECTRONICA SINICA, 2023, 1-8. DOI： 10.12263/DZXB.20230056.

摘要

为消减由非线性动态系统与状态约束不一致引起的高斯截断误差和线性化误差，提出时空学习驱动的混合重要性高斯滤波（Mixed Importance Gaussian Filtering，MIGF）算法.该算法采用软约束和稀疏正则技术表征目标状态真实参数和当前时刻测量之间的时空因果映射，利用瑞利熵混合历史状态和最新测量构建次优建议分布，根据因果不变结构融合重要性高斯-埃尔米特积分进行预测和更新，从而平衡样本数目和滤波精度并自适应地修正Sigma点权值，提升约束动态系统建模的准确性和参数估计的鲁棒性.实验结果表明，在综合性非线性标量状态估计中，相比无迹粒子滤波（Unscented Particle Filtering，UPF），MIGF的滤波状态协方差减小45.37%.在小型固定翼无人机纯方位跟踪实验中，相比混合截断无迹卡尔曼滤波（Mixture Truncated Unscented Kalman Filtering，MTUKF）、交互多模型无迹卡尔曼滤波（Interacting Multiple- model Unscented Kalman Filtering，IMMUKF）和多模型Rao-blackwell粒子滤波（Multi-model Rao-Blackwell particle filtering，MMRBPF），MIGF算法在时空一致约束下优化动态模型结构，获得更为准确、鲁棒的状态参数估计，整体运行时间相比IMMUKF增加一个数量级，而相比MMRBPF以模型集数目成倍减小.

Abstract

To reduce the Gaussian truncation error and linearization error caused by the inconsistency between the nonlinear dynamic and state constraints

this paper proposes a spatiotemporal learning-driven mixed importance Gaussian filtering (MIGF) algorithm. The algorithm uses soft constraint and sparse regularization technology to represent the spatiotemporal causal mapping between the true parameters of the target state and the current time measurement

uses Rayleigh entropy to mix the historical state and the latest measurement to establish the suboptimal recommendation distribution

and integrates the importance Gauss-Hermite integral to predict and update according to the causal invariant structure

adaptively correcting the Sigma points' weights and balancing the sample size and filtering accuracy to improve the accuracy of constrained dynamic system modeling and the robustness of parameter estimation. Experimental results show that

compared with the unscented particle filtering (UPF)

the filtered state covariance decreases by 45.37% for the comprehensive nonlinear scalar state estimation. Furthermore

in the bearings-only tracking experiment for a small fixed-wing UAV

compared with the mixture truncated unscented Kalman filtering (MTUKF)

interactive multi-model unscented Kalman filtering (IMMUKF) and multi-model Rao-blackwell particle filtering (MMRBPF)

the MIGF algorithm optimizes the dynamic model structure consistent spatiotemporal constraint to obtain more accurate and robust state parameter estimation. However

the overall running time is an order of magnitude higher than that of IMMUKF

while compared with MMRBPF

the reduction factor is the size of the model set.

关键词

Keywords

references

SARKKA S , SOLIN A , HARTIKAINEN J . Spatiotemporal learning via infinite-dimensional Bayesian filtering and smoothing: A look at Gaussian process regression through Kalman filtering [J]. IEEE Signal Processing Magazine , 2013 , 30 ( 4 ): 51 - 61 .

RONG LI X , JILKOV V P . Survey of maneuvering target tracking. Part I. Dynamic models [J]. IEEE Transactions on Aerospace and Electronic Systems , 2003 , 39 ( 4 ): 1333 - 1364 .

RONG LI X . Compatibility and modeling of constrained dynamic systems [C]// 2016 19th International Conference on Information Fusion (FUSION) . Piscataway : IEEE , 2016 : 240 - 247 .

Doucet A , Johansen A M . A tutorial on particle filtering and smoothing: Fifteen years later [J]. Handbook of nonlinear filtering , 2009 , 12 ( 656-704 ): 3 .

奚畅 , 蔡志明 , 袁骏 . 被动声纳粒子滤波检测前跟踪的检测门限设置方法 [J]. 电子学报 , 2022 , 50 ( 7 ): 1722 - 1734 .

XI C , CAI Z M , YUAN J . A method of determining detection threshold for particle filter track-before-detect in passive sonar [J]. Acta Electronica Sinica , 2022 , 50 ( 7 ): 1722 - 1734 . (in Chinese)

LIU C J , LI B B , CHEN W H . Particle filtering with soft state constraints for target tracking [J]. IEEE Transactions on Aerospace and Electronic Systems , 2019 , 55 ( 6 ): 3492 - 3504 .

DUAN B , ZHANG Q , GENG F , et al . Remaining useful life prediction of lithium-ion battery based on extended Kalman particle filter [J]. International Journal of Energy Research , 2020 , 44 ( 3 ): 1724 - 1734 .

Van Der Merwe R , Doucet A , De Freitas N , et al . The unscented particle filter [J]. Advances in neural information processing systems , 2000 , 13 .

ELVIRA V , MARTINO L , CLOSAS P . Importance Gaussian quadrature [J]. IEEE Transactions on Signal Processing , 2021 , 69 : 474 - 488 .

叶瑾 , 许枫 , 杨娟 , 等 . 一种基于多传感器的复合量测IMM-EKF数据融合算法 [J]. 电子学报 , 2020 , 48 ( 12 ): 2326 - 2330 .

YE J , XU F , YANG J , et al . A composite measurement IMM-EKF data fusion algorithm based on multi-sensor [J]. Acta Electronica Sinica , 2020 , 48 ( 12 ): 2326 - 2330 . (in Chinese)

YAO Y Q , XU X S , YANG D R , et al . An IMM-UKF aided SINS/USBL calibration solution for underwater vehicles [J]. IEEE Transactions on Vehicular Technology , 2020 , 69 ( 4 ): 3740 - 3747 .

HOSTETTLER R , SÄRKKÄ S . Rao-blackwellized Gaussian smoothing [J]. IEEE Transactions on Automatic Control , 2019 , 64 ( 1 ): 305 - 312 .

ZHANG H W . Multiple importance unscented Kalman filtering with soft spatiotemporal constraint for multi-passive-sensor target tracking [J]. International Journal of Robust and Nonlinear Control , 2023 , 33 ( 1 ): 264 - 281 .

RONG LI X , JILKOV V P . Survey of maneuvering target tracking. Part V. Multiple-model methods [J]. IEEE Transactions on Aerospace and Electronic Systems , 2005 , 41 ( 4 ): 1255 - 1321 .

SIMON D . Kalman filtering with state constraints: A survey of linear and nonlinear algorithms [J]. IET Control Theory & Applications , 2010 , 4 ( 8 ): 1303 - 1318 .

AFTAB W , MIHAYLOVA L . A Gaussian process regression approach for point target tracking [C]// 2019 22th International Conference on Information Fusion (FUSION) . Piscataway : IEEE , 2020 : 1 - 8 .

AFTAB W , MIHAYLOVA L . A learning Gaussian process approach for maneuvering target tracking and smoothing [J]. IEEE Transactions on Aerospace and Electronic Systems , 2020 , 57 ( 1 ): 278 - 292 .

SIMON D , CHIA T L . Kalman filtering with state equality constraints [J]. IEEE Transactions on Aerospace and Electronic Systems , 2002 , 38 ( 1 ): 128 - 136 .

BOYD S , VANDENBERGHE L . Convex Optimization [M]. Cambridge, UK : Cambridge University Press , 2004 .

LI B B , LIU C J , CHEN W H . An auxiliary particle filtering algorithm with inequality constraints [J]. IEEE Transactions on Automatic Control , 2016 , 62 ( 9 ): 4639 - 4646 .

GARCIA-FERNANDEZ Á F , MORELANDE M R , GRAJAL J . Truncated unscented Kalman filtering [J]. IEEE Transactions on Signal Processing , 2012 , 60 ( 7 ): 3372 - 3386 .

GARCÍA-FERNÁNDEZ Á F , MORELANDE M R , GRAJAL J . Mixture truncated unscented Kalman filtering [C]// 2012 15th International Conference on Information Fusion . Piscataway : IEEE , 2012 : 479 - 486 .

DE GEETER J , VAN BRUSSEL H , DE SCHUTTER J , et al . A smoothly constrained Kalman filter [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence , 1997 , 19 ( 10 ): 1171 - 1177 .

Särkkä S , Svensson L . Bayesian filtering and smoothing [M]. Cambridge university press , 2023 .

CRASSIDIS J L , JUNKINS J L . Optimal estimation of dynamic systems [M]. 2nd ed . Boca Raton, Fla .: Chapman and Hall/CRC , 2011 .

ARASARATNAM I , HAYKIN S , ELLIOTT R J . Discrete-time nonlinear filtering algorithms using gauss-Hermite quadrature [J]. Proceedings of the IEEE , 2007 , 95 ( 5 ): 953 - 977 .

BECKERS S . Equivalent causal models [J]. Proceedings of the AAAI Conference on Artificial Intelligence , 2021 , 35 ( 7 ): 6202 - 6209 .

HUBER M F . Recursive Gaussian process: On-line regression and learning [J]. Pattern Recognition Letters , 2014 , 45 : 85 - 91 .

Izanloo R , Fakoorian S A , Yazdi H S , et al . Kalman filtering based on the maximum correntropy criterion in the presence of non-Gaussian noise [C]// 2016 Annual Conference on Information Science and Systems (CISS) . IEEE , 2016 : 500 - 505 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

暂无数据