航空学报 > 2023, Vol. 44 Issue (6): 326823-326823   doi: 10.7527/S1000-6893.2022.26823

贝叶斯推理运动轨迹相干累积的动目标检测方法

杨文彬1,2, 王悦斌1,2, 李旦1,2(), 张建秋1,2   

  1. 1.电磁波信息科学教育部重点实验室,上海 200433
    2.复旦大学 信息科学与工程学院 电子工程系,上海 200433
  • 收稿日期:2021-12-16 修回日期:2021-12-30 接受日期:2022-03-21 出版日期:2023-03-25 发布日期:2022-03-30
  • 通讯作者: 李旦 E-mail:lidan@fudan.edu.cn
  • 基金资助:
    国家自然科学基金(11974082)

A moving target detector with coherent integration of Bayesian-inferred motion trajectories

Wenbin YANG1,2, Yuebin WANG1,2, Dan LI1,2(), Jianqiu ZHANG1,2   

  1. 1.Key Laboratory of EMW Information,Shanghai 200433,China
    2.Department of Electronic Engineering,School of Information Science and Technology,Fudan University,Shanghai 200433,China
  • Received:2021-12-16 Revised:2021-12-30 Accepted:2022-03-21 Online:2023-03-25 Published:2022-03-30
  • Contact: Dan LI E-mail:lidan@fudan.edu.cn
  • Supported by:
    National Natural Science Foundation of China(11974082)

摘要:

雷达理论和实践都表明相干累积是机动目标探测行之有效的方法,但其性能会因脉冲回波处理间隔内未知距离和/或多普勒的徒动而降低。为解决这个问题,提出了一种贝叶斯推理运动轨迹相干累积的动目标检测方法。首先,通过对脉冲压缩回波信号的距离频率进行翻转,就可从未翻转与翻转后的回波信号相乘的傅里叶逆变换中,提取出慢时间维描述待探测多目标运动轨迹的时频信号。视该时频信号中各目标运动轨迹为状态变量,获得的时频信号为观测,建立起多目标运动轨迹的状态空间模型。这样,据贝叶斯滤波推理出的目标运动轨迹,构建出快-慢时间维联合的二维匹配滤波器,就能补偿目标高速/机动带来的未知距离和/或多普勒徒动。理论分析和仿真实验均证明:所提算法适用具有复杂且未知运动状态的目标,且呈现出了优于文献报道方法的性能。

关键词: 脉冲多普勒雷达, 贝叶斯推理, 检测聚焦, 匹配滤波, 距离徒动, 多普勒徒动

Abstract:

It has been shown theoretically and practically that coherent integration is an effective method for moving target detection. Unfortunately, its performance is seriously damaged by the unknown range and/or Doppler frequency migrations in a coherent processing interval. To handle such a problem, a moving target detector with the coherent integration of Bayesian-inferred motion trajectories is proposed in this paper. Analyses show that by flipping the range-frequency of the pulse compressed echo signal, the time-frequency signal describing the motion trajectories of the detected targets in the slow-time domain is got by the inverse Fourier transform of the product of the original echo signal and the flipped one. When the time-frequency signal is observed and its phases describing the detected moving target trajectories are regarded as state variables, a state space model for inferring the motion trajectories of multiple targets is given. Based on the inferred target trajectories, the range-frequency-slow-time 2D matched filters are constructed to compensate the unknown range and/or Doppler frequency migrations in an echo signal. Numerical simulation results verify that the proposed method can be applied to high speed/maneuvering targets with unknown complex motion form, and has superior performance than the methods reported in the literature.

Key words: pulse-Doppler radar, Bayesian inference, detection focus, matched filter, range migration, Doppler frequency migration

中图分类号: