外辐射源雷达信号处理与目标跟踪方法研究

发布时间：2018-06-17 09:30

本文选题：外辐射源雷达 + 目标检测　；参考：《西安电子科技大学》2016年博士论文

【摘要】：外辐射源雷达利用目标反射非协作照射源信号这一特点,通过处理静默接收到的电磁信号,来实现对特定区域的目标探测。依赖于其寂静接收、收发分置及所使用照射源信号频段的特点,外辐射源雷达具有常规雷达不可比拟的“四抗”优势,是一类发展前景非常广泛的雷达系统。如利用外辐射源雷达不会产生电磁污染这一优点,可将其部署在城市、郊区或人口密集等区域,以实现对运动目标的监测。传统外辐射源雷达通常采用“一发一收”模式对目标进行检测、定位和跟踪。然而,随着技术的不断发展和更新换代,传统“一发一收”的探测方式逐渐跟不上社会需求日新月异的步伐。当前,为了能有效应对不断提高的民用或军事需求,外辐射源雷达技术研究逐步走向了宽带多源的发展之路。与传统探测方式相比,宽带多源外辐射源雷达带来的一个最直接变化就是信号处理规模呈数十倍甚至上百倍的增长,这种增长为传统硬件处理能力带来了不小的压力。另一方面,外辐射源雷达所使用照射源信号的非合作性和起伏性决定了其检测到的目标航迹会出现不规律的断续问题,“精细化”的跟踪滤波方法是弥补这一体制缺陷的有效途径。本文在某部委项目的支持下,围绕外辐射源雷达信号处理和目标跟踪滤波方法两方面开展研究工作,取得的成果可归纳为：1.研究了常规外辐射源雷达信号处理存在的若干实际问题。针对实际工程应用,分析了常规处理方法存在的不足之处,并提出了相应的优化或改进方法。如提出了一种幅度修正、单元平均、补偿、远近分区、二次排序选大相结合的综合目标检测方法,有效地克服了因近距离强杂波剩余使距离-多普勒检测平台呈“斜坡”状而影响目标检测的问题。对于实际中参考通道也可能接收到目标回波的情况,推导分析了此种情况下真实目标能量被抵消和产生虚假目标的机理,并提出了一种利用杂波抑制权系数消除参考信号中所含目标回波的方法,该方法能够有效解决因参考信号含有目标回波信号而出现的衍生虚假目标及部分真实目标被削弱的问题。2.研究了多源全向外辐射源雷达信号实时处理的实现技术。针对多源全向外辐射源雷达信号实时处理运算量大的问题,提出了一种基于通用处理器GPU架构的并行实现方法,经过多层次纵向优化后,所提方法的处理效率明显改善,在应用中达到了实时处理的水平。与传统DSP处理架构相比,所提方法具有成本低廉、易于调试、硬件规模小的优势,而且可移植、可扩展,符合目前主流发展的“软件化雷达”的设计原则。最后,一系列仿真和实测数据的处理结果表明,所提方法的实现效率远高于传统方法,验证了所提实现方法的可行性和高效性。3.提出了一种可用于但不限于外辐射源雷达目标跟踪的加权最小二乘粒子滤波方法。针对非线性粒子滤波在对目标状态迭代估计中出现的难以避免的粒子退化现象,提出通过最小二乘估计将当前观测信息以加权的方式集成到粒子采样过程中,避免了传统方法中当前观测仅仅用来评判粒子好坏的不足,并以此来促进粒子由先验区域向目标真实状态的高似然区域的移动,进而达到提高粒子分布质量,缓解粒子退化程度和改善滤波精度的目的。另外,在推导粒子采样函数的过程中,提出了一种利用状态向量扩展观测向量的方法来解决潜在的矩阵秩亏问题,保证了所提滤波方法的普适性。最后,经过仿真和实测数据的处理验证了所提方法的有效性。4.提出了一种并行重采样方法,解决了传统系统重采样限制粒子滤波并行度的问题。所提方法的基本思想和系统重采样类似,它们的不同之处在于,所提方法使用每个粒子归一化权值与粒子数量乘积的取整部分作为该粒子在重采样中被保留的次数,同时为了保持粒子数量的恒定,理论验证了直接舍弃归一化权值最小的若干个粒子的合理性。该方法具有结构简单、便于并行的优点。
[Abstract]:The external radiation source radar uses the characteristic of the target reflecting the non cooperative source signal to realize the target detection of the specific region by processing the silent received electromagnetic signal. It depends on the characteristics of the silent reception, the transceiver division and the frequency of the source of the irradiated source. The external radiation source radar has the unparalleled "four anti" radar. Advantage is a kind of radar system with wide development prospects. For example, it can be deployed in a city, a suburb or a densely populated area to realize the monitoring of a moving target. However, with the continuous development and replacement of technology, the traditional "one hair and one collection" detection method gradually can not keep up with the rapid pace of social demand. At present, in order to effectively cope with the increasing demand for civil or military, the research of external radiation source radar technology has gone to the development of broadband and multi source. In comparison, one of the most direct changes in the wide-band and multi source radar is that the signal processing scale is tens to hundreds or even hundreds of times, which brings no small pressure to the traditional hardware processing capability. On the other hand, the non cooperative and undulation of the source signal used by the external radiation source radar determines its detection. The target track will appear irregular intermittence. The "fine" tracking filtering method is an effective way to make up the defect of the system. Under the support of a certain ministry project, the research work is carried out around two aspects of the signal processing and target tracking filtering method of the external radiation source radar. The achievements can be summed up as follows: 1. research This paper investigates some practical problems in conventional external radiant radar signal processing. Aiming at practical engineering applications, the shortcomings of conventional processing methods are analyzed, and the corresponding optimization or improvement methods are proposed. For example, a comprehensive target detection is proposed, which is a combination of amplitude correction, unit average, compensation, far and near partition, and two order sorting. This method effectively overcomes the problem that the target detection is affected by the "slope" of the distance Doppler detection platform because of the close range strong clutter remaining. The mechanism of the true target energy being offset and producing false targets is derived and analyzed in this case. A method of eliminating the echo of the target in the reference signal by using the clutter suppression coefficient, this method can effectively solve the problem that the derivative false target and some real target are weakened because of the reference signal containing the target echo signal..2. has studied the realization technology of the real-time processing of the multi source all outward radiant radar signal. A parallel implementation method based on the universal processor GPU architecture is proposed. After multi-layer longitudinal optimization, the processing efficiency of the proposed method is obviously improved and the real-time processing is reached in the application. Compared with the traditional DSP processing architecture, the proposed method has been proposed. It has the advantages of low cost, easy to debug and small hardware, and can be portable and extensible. It meets the design principle of "software radar". Finally, the results of a series of simulation and measured data show that the implementation efficiency of the proposed method is far higher than that of the traditional method, and the feasibility and high performance of the proposed implementation method is verified. A weighted least square particle filter, which can be used but not limited to the target tracking of the radar target of the external source, is proposed by.3.. In view of the hard to avoid particle degradation in the iterative estimation of the target state by the nonlinear particle filter, a least square estimation is proposed to integrate the current observation information into the particle in a weighted way. In the subsampling process, the current observation, which is used in the traditional method, is only used to judge the defects of the particle, and in order to promote the movement of the high likelihood region from the prior area to the true state of the target, to improve the quality of the particle distribution, alleviate the degree of particle degradation and improve the filtering accuracy. In the process of sampling function, a method of using state vector to extend the observation vector is proposed to solve the potential matrix rank deficiency problem, and the universality of the proposed filtering method is guaranteed. Finally, the effectiveness of the proposed method is verified by the processing of simulated and measured data.4., a parallel resampling method is proposed to solve the traditional system. The basic idea of the proposed method is that the basic idea of the proposed method is similar to the system resampling. The difference is that the proposed method uses the integral part of the product of each particle normalized weight and the number of particles as the number of times the particle is retained in the resampling, and in order to keep the number of particles constant, the method is used to keep the particle number constant. The rationality of some particles directly discarding the least normalized weight is verified. This method has the advantages of simple structure and convenient parallelism.
【学位授予单位】：西安电子科技大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN957.51

【相似文献】