电子学报 ›› 2016, Vol. 44 ›› Issue (5): 1149-1155.DOI: 10.3969/j.issn.0372-2112.2016.05.019

• 学术论文 • 上一篇    下一篇

量测提升卡尔曼滤波

胡振涛, 胡玉梅, 刘先省   

  1. 河南大学图像处理与模式识别研究所, 河南开封 475004
  • 收稿日期:2014-10-20 修回日期:2015-05-18 出版日期:2016-05-25
    • 作者简介:
    • 胡振涛 男,1979年6月出生于河南永城市,现为河南大学计算机与信息工程学院副教授,硕士生导师.主要研究方向为复杂系统建模与估计,非线性滤波.E-mail:hzt@henu.edu.cn;胡玉梅 女,1990年10月出生于河南永城市,现为河南大学计算机与信息工程学院研究生.主要研究方向为复杂系统建模与估计,非线性滤波.E-mail:hym_henu@163.com;刘先省 男,1964年9月出生于河南潢川县,现为河南大学计算机与信息工程学院教授,博士生导师.主要研究方向为多源信息融合,传感器管理.E-mail:liuxianxing@henu.edu.cn
    • 基金资助:
    • 国家自然科学基金 (No.61300214); 中国博士后科学基金 (No.2014M551999); 河南省高校科技创新团队支持计划 (No.13IRTSTHN021); 河南省基础与前沿技术研究计划 (No.132300410148); 河南省博士后科学基金 (No.2013029); 河南省高校青年骨干教师资助计划 (No.2013GGJS-026); 河南大学优秀青年培育基金 (No.0000A40366)

Kalman Filter Based on Measurement Lifting Strategy

HU Zhen-tao, HU Yu-mei, LIU Xian-xing   

  1. Institute of Image Processing and Pattern Recognition, Henan University, Kaifeng, Henan 475004, China
  • Received:2014-10-20 Revised:2015-05-18 Online:2016-05-25 Published:2016-05-25
    • Supported by:
    • National Natural Science Foundation of China (No.61300214); China Postdoctoral Science Foundation (No.2014M551999); Science and Technology Innovation Team Support Plan of Colleges and Universities in Henan Province (No.13IRTSTHN021); Henan Research Program of Basic and Frontier Technology (No.132300410148); Postdoctoral Science Foundation of Henan Province (No.2013029); Funding Project for Young Backbone Teachers of Universities in Henan Province (No.2013GGJS-026); Excellent Youth Cultivation Fund of Henan University (No.0000A40366)

摘要:

滤波器设计是系统辨识和状态估计的重要基础.卡尔曼滤波通过状态预测和量测更新的实现框架,在最小方差准则下实现对目标状态的最优估计,但在单传感器量测环境中其滤波精度易受量测噪声随机性的影响.本文提出一种基于量测提升策略的卡尔曼滤波算法实现框架,新方法依据当前时刻量测和量测噪声先验统计信息构建虚拟量测,并通过对虚拟量测采样以及融合提升系统量测信息可靠性,进而改善状态估计精度.同时,针对算法在工程应用中实时性、准确性以及鲁棒性等需求,设计了分布式加权融合和集中式一致性融合的两种实现结构.理论分析和仿真实验结果验证了算法的可行性和有效性.

关键词: 卡尔曼滤波, 量测提升策略, 分布式加权融合, 集中式一致性融合

Abstract:

Filter design is the signification foundation for system identification and state estimation.Based on the realization construction of state prediction and measurement update, Kalman filter can obtain the optimal estimation of state estimated under the linear minimum variance criterion, but the filtering precision is vulnerable to the random characteristics in single sensor condition.A novel realization structure of Kalman filter based on measurement lifting strategy is proposed in the paper.At first, virtual measurement is constructed on the basis of latest measurement and the prior statistical information of measurement noise modeling.Then, virtual measurements are reasonably sampled and fusion to modify the measurement reliability, and the estimation precision is improved.In addition, aiming to the algorithm requirements including real-time, precise and robustness in engineering application, the distributed weight fusion structure and the centralized consistency fusion are designed respectively.Finally, the theoretical analysis and experimental results show the feasibility and efficiency of algorithm proposed.

Key words: Kalman filter, measurement lifting strategy, distributed weight fusion, centralized consistency fusion

中图分类号: