【作者】 宋骊平；

【导师】 姬红兵；

【作者基本信息】 西安电子科技大学 ， 模式识别与智能系统， 2008， 博士

【摘要】 由于被动探测与传统的雷达探测相比具有隐蔽性好,不易受到攻击等许多优点,因此利用目标自身辐射的信号对目标进行被动探测和跟踪已成为现代防御系统的研究热点之一。构建被动多传感器防御系统,实现全被动的目标探测与跟踪,对于提高防御系统的能力,具有十分重要的意义。作为被动多传感器防御系统的一项关键技术,本文对被动多传感器目标跟踪方法开展了深入研究,提出了一些有效的新方法。全文共分七章,各章的主要内容如下:第一章,简要介绍了本文的研究背景、意义和被动多传感器系统的构成,概述了目标定位与跟踪问题的研究现状,列举了本文的主要研究成果和全文的内容安排。第二章,介绍了目标跟踪的基本理论,包括常用的目标运动模型与滤波算法。第三章,对多被动传感器情况下的目标跟踪问题进行了深入研究。从最基本的三维空间中的匀速运动目标入手,建立了多个被动传感器的目标运动模型与传感器模型;针对多被动传感器目标跟踪中存在的非线性问题,分别推导了基于推广卡尔曼滤波与基于无迹卡尔曼滤波的多被动传感器目标跟踪方法,提出了一种基于最小二乘的被动多传感器目标跟踪算法,实现了较高精度的非机动目标跟踪。第四章,针对机动目标跟踪中的机动检测问题,提出了基于三阶累积量的机动检测算法。由于高阶累积量能够抑制高斯噪声,因此在三阶累积量域更易于检测机动,同时通过采用逐点更新,可实时进行机动检测。在所提出的机动检测算法的基础上,进一步研究了基于这一算法的被动多传感器的机动目标跟踪。第五章,针对被动多传感器条件下的机动目标跟踪问题,提出了一种基于当前统计模型的最小二乘自适应跟踪算法,首先采用最小二乘对目标的状态进行粗估计,然后采用当前机动目标模型与自适应跟踪算法进行线性的卡尔曼滤波。另外将经典的推广卡尔曼滤波与当前统计模型相结合,推导了推广卡尔曼滤波自适应跟踪算法;将基于采样策略的平方根无迹卡尔曼滤波与当前统计模型相结合,推导了平方根无迹卡尔曼滤波自适应跟踪算法。由于机动目标当前统计模型为时变模型,能够更为真实地反映目标的机动变化,因此采用这一模型的三种算法能够较好地跟踪机动目标,尤其是所提出的最小二乘自适应跟踪算法在被动多传感器条件下,对机动目标的跟踪可以取得令人满意的效果。第六章,将无迹卡尔曼滤波与交互多模型算法相结合,应用于被动多传感器的机动目标跟踪中,提出了一种无迹卡尔曼滤波交互多模型算法,实现了在多个被动传感器只测角条件下对机动目标的跟踪。另外,将本文提出的多被动传感器的最小二乘融合应用于机动目标跟踪中,提出了一种被动多传感器最小二乘交互多模型算法,并通过仿真实验对几种算法进行了比较。第七章,总结了论文的主要工作,并对后续研究作了展望。更多还原

【Abstract】 Because of the merit of concealment and the capability to avoid attack for passive detection in comparison with conventional radar detection, passive detecting and tracking by utilizing the signal emitted from the target have been one of the hot topics of the modern defense system. Constructing the system of passive multi-sensor so as to achieve the passive target detecting and tracking is significant to enhance the capability of defense system. The passive multi-sensor target tracking, as a key technology of passive multi-sensor defense system, is studied and some effective new methods are presented in the thesis. The thesis is classified into seven chapters and their contents are outlined as follows.In Chapter 1, the background and importance of the research on the passive multi-sensor target tracking are explained; the structure of passive multi-sensor system is introduced. The status quo of the research on target locating and tracking techniques is reviewed. Finally, the main achievements and arrangements of the thesis are concluded.In Chapter 2, the basic theory of target tracking is described, including the common dynamical models and filters.In Chapter 3, target tracking with multiple passive sensors is discussed in deep. Begin with the target moving in constant velocity, dynamical model of the target and the measurement model of multiple passive sensors are set up. For the nonlinear relationship between the state of the target and the bearings measurements, the algorithms of passive multi-sensor target tracking based on extended kalman filter and unscented kalman filter have been deduced, respectively. And the algorithm of passive multi-sensor target tracking based on the least squares fusion is proposed. Non maneuvering target tracking with high precision is achieved in the chapter.In Chapter 4, the algorithm of maneuver detection based on higher-order cumulants is proposed. As the higher-order cumulants are blind to Gaussian noise (white or colored), the behavior of the maneuver becomes obvious and easy to be detected in the higher-order cumulant domains. Taking the sliding window processing, the maneuver detection algorithm is real-time without any time delay. On the basis of the maneuver detection algorithm, the maneuver target tracking for multiple passive sensors is studied further. In Chapter 5, for maneuvering target tracking with multiple passive sensors, a least squares adaptive algorithm based on the current statistical model is presented, in which the state of the target is approximately estimated by least squares algorithm at first, and then a current statistical model and an adaptive algorithm are employed. An extended kalman filter adaptive algorithm is presented in combination with the extended kalman filter and the current statistical model. A root square unscented kalman filter adaptive algorithm is presented in combination with the root square unscented kalman filter and the current statistical model. As a time variant model, the current statistical model is able to describe the target maneuver more reasonable, the above three algorithms has ability to track the maneuver target more effectively, especially does the least squares adaptive algorithm.In Chapter 6, the unscented kalman filter in combination with the interacting multiple model applied to the passive multi-sensor maneuver target tracking, a novel unscented kalman filter interacting multiple model algorithm is proposed. It can track maneuver target in multi-sensor bearings-only tracking. In addition, in application to the maneuver target tracking with the least squares fusion, the least squares interacting multiple model algorithm is presented. Finally, a comparison of the algorithmic performance is presented.In Chapter 7, a summarization to the whole thesis is given and the prospect of the issues concerned in the field is also made.更多还原