【作者】 罗军；

【导师】 张顺生；

【作者基本信息】 电子科技大学 ， 电子信息（专业学位）， 2024， 硕士

【摘要】 无源定位因其定位距离远、隐蔽性强以及安全性高等优点受到了广泛研究。传统多站时差定位体制和多站测向定位体制的定位精度与基线长度密切相关,若要定位远距离目标则需相应的长基线。当定位距离与基线长度的比值过大时,即在大距离基线比下多站时差定位和多站测向定位方法的定位精度会大幅下降甚至失效。直接定位方法较传统两步定位法具有低信噪比、多辐射源定位无需参数关联等优势,由此引入直接定位方法来解决大距离基线比下的辐射源定位问题。其主要工作总结如下:(1)分析了大距离基线比下传统多站时差定位方法和多站测向定位方法存在定位性能严重下降的问题,研究了一种面向圆阵接收的时域子空间直接定位方法,推导了该模型下定位精度的克拉美罗下界,并通过数值仿真验证了大距离基线比下直接定位方法的有效性。(2)针对大距离基线比下辐射源到各接收站的时间和角度信息差异过小导致传统直接定位方法定位精度不足的问题,提出了一种联合自适应LASSO和块稀疏贝叶斯的辐射源直接定位方法。建立了多站接收的稀疏定位模型,联合自适应LASSO和块稀疏贝叶斯学习重构稀疏信号,并利用网格密度聚类实现大距离基线比下辐射源定位。仿真结果表明:当距离基线比为100倍时,所提方法在低信噪比和少快拍条件下较传统直接定位方法和稀疏贝叶斯类方法具有更优的定位性能。(3)大距离基线比下辐射源到各接收站差异过小的时间和角度信息会增加离格条件下稀疏信号的重构难度,从而恶化离格辐射源的定位精度。针对该问题,提出了一种联合参数化字典和网格密度聚类的离格辐射源直接定位方法。该方法构建以自适应LASSO先验为基础的分层贝叶斯模型,通过泰勒展开动态更新字典,减少预设字典与真实字典之间的偏差,并利用网格密度聚类实现大距离基线比下离格辐射源定位。仿真结果表明:当距离基线比为100倍时,所提方法的定位性能在低信噪比和少快拍条件下优于稀疏贝叶斯类方法。更多还原

【Abstract】 Passive localization has been widely studied due to its advantages such as long localization distance,strong concealment,and high security.The accuracy of traditional multi-station time difference localization systems and multi-station direction finding localization systems is closely related to the baseline length.To locate distant targets,a corresponding long baseline is required.When the ratio of the localization distance to the baseline length is too large,the accuracy of multi-station time difference localization and multi-station direction finding localization methods will significantly decrease or even fail under large distance baseline ratios.Direct localization methods have advantages over traditional two-step localization methods,such as low signal-to-noise ratio and no need for parameter association for multi-radiation source localization.Therefore,direct localization methods were introduced to solve the radiation source localization problem under large distance baseline ratios.The main work is summarized as follows:(1)The traditional multi-station time difference positioning method and multistation direction finding positioning method under the large distance baseline ratio were analyzed,and a time-domain subspace direct localization method oriented to circular array reception was studied..The Cramér-Rao lower bound of positioning accuracy under this model was derived,and the effectiveness of the direct positioning method under the large distance baseline ratio was verified through numerical simulation.(2)To address the issue of traditional direct positioning methods insufficient accuracy due to the small differences in time and angle information from the radiation source to each receiving station under large distance baseline ratios,a direct positioning method for radiation sources that combines adaptive LASSO and block sparse Bayesian was proposed.A sparse positioning model for multi-station reception was established.The sparse signal was reconstructed by jointly learning the adaptive LASSO and block sparse Bayesian.The grid density clustering was used to achieve radiation source positioning under large distance baseline ratios.Simulation results indicated that when the distance baseline ratio is 100 times,the proposed method exhibits superior positioning performance under conditions of low signal-to-noise ratio and few snapshots compared to traditional direct positioning methods and sparse Bayesian class methods.(3)Under large distance baseline ratios,the small differences in time and angle information from the radiation source to each receiving station increase the difficulty of reconstructing sparse signals under off-grid conditions,thereby worsening the positioning accuracy of off-grid radiation sources.To address this issue,a direct positioning method for off-grid radiation sources that combines a parameterized dictionary and grid density clustering was proposed.This method constructs a hierarchical Bayesian model based on the adaptive LASSO prior,dynamically updates the dictionary through Taylor expansion,reduces the deviation between the preset dictionary and the real dictionary,and uses the grid density clustering to achieve off-grid radiation source positioning under large distance baseline ratios.Simulation results indicated that when the distance baseline ratio is 100 times,the positioning performance of the proposed method is superior to sparse Bayesian class methods under conditions of low signal-to-noise ratio and few snapshots.更多还原