系统工程与电子技术 ›› 2023, Vol. 45 ›› Issue (3): 660-668.doi: 10.12305/j.issn.1001-506X.2023.03.06

• 电子技术 • 上一篇    下一篇

基于聚类求解TVAR模型的目标微多普勒分析

禄晓飞1, 靳硕静2, 洪灵3, 戴奉周2,*   

  1. 1. 中国人民解放军63620部队, 甘肃 兰州 732750
    2. 西安电子科技大学电子工程学院, 陕西 西安 710071
    3. 陕西师范大学计算机科学学院, 陕西 西安 710061
  • 收稿日期:2021-12-13 出版日期:2023-02-25 发布日期:2023-03-09
  • 通讯作者: 戴奉周
  • 作者简介:禄晓飞(1981—), 男, 高级工程师, 博士, 主要研究方向为雷达数据处理
    靳硕静(1996—), 女, 硕士研究生, 主要研究方向为复杂运动目标ISAR成像
    洪灵(1986—), 女, 副教授, 博士, 主要研究方向为机器学习、智能信号处理、高光谱图像处理
    戴奉周(1978—), 男, 副教授, 博士, 主要研究方向为智能感知与信号处理、电磁超材料、毫米波雷达
  • 基金资助:
    国家自然科学基金(61701290)

Target micro-Doppler analysis of TVAR model based on clustering

Xiaofei LU1, Shuojing JIN2, Ling HONG3, Fengzhou DAI2,*   

  1. 1. Unit 63620 of the PLA, Lanzhou 732750, China
    2. School of electronic engineering, Xidian University, Xi'an 710071, China
    3. School of computer science, Shanxi Normal University, Xi'an 710061, China
  • Received:2021-12-13 Online:2023-02-25 Published:2023-03-09
  • Contact: Fengzhou DAI

摘要:

现有时频分析方法对目标进行微多普勒分析时的时频分辨率不足。针对该问题,提出了基于聚类先验求解前后向时变自回归(time-varying autoregressive, TVAR)模型的时频分析算法,来进行空间锥体目标的微多普勒分析。使用基于扩展块稀疏贝叶斯学习(extended block sparse Bayesian learning, EBSBL)的改进算法对TVAR模型的时不变块稀疏系数采用了聚类结构的先验, 通过适当处理邻域的超参数来促进相邻稀疏系数之间的相关性, 并结合刚体目标的时不变块稀疏系数的块边界已知的先验信息来求解时不变系数。电磁仿真和实测数据实验结果表明, 所提算法在微多普勒分析时能够得到较传统方法更高的时频分辨率, 时频聚集性更高, 并且抗噪声性能较好。

关键词: 微多普勒分析, 前后向时变自回归模型, 块稀疏, 聚类结构先验

Abstract:

Aiming at the problem of insufficient time-frequency resolution of existing time-frequency analysis methods for micro-Doppler analysis of targets, a time-frequency analysis algorithm based on clustering prior to solve the forward and backward time-varying autoregressive (TVAR) models is proposed for micro-Doppler analysis of spatial cone targets. Using an improved algorithm based on extended block sparse Bayesian learning (EBSBL), the time-invariant block sparse coefficient of the TVAR model adopts a prior of the clustering structure, and the correlation between adjacent sparse cofficient is also promoted by appropriate handling of the hyperparameters of the neighborhood. The correlation between adjacent sparse coefficients is combined with the known prior information of the time-invariant block sparse coefficient's block boundaries of the rigid target to solve the time-invariant coefficients. The experimental results of electromagnetic simulation and measured data show that the proposed algorithm in this paper can obtain a higher time-frequency resolution, with a higher time-frequency aggregation, and a stronger anti-noise performance than traditional methods in micro-Doppler analysis.

Key words: micro-Doppler analysis, forward and backward time-varying autoregressive (TVAR) model, block sparsity, cluster-structured prior

中图分类号: