微弱信号的定位与跟踪技术研究

发布时间：2018-03-29 22:15

本文选题：无源机动弱目标　切入点：粒子滤波　出处：《电子科技大学》2015年硕士论文

【摘要】：微弱信号的定位与跟踪在军用和民用领域都有着广泛的应用。由于传感器截获的目标信号强度较小,难以将其从噪声中分离,采用传统目标检测与跟踪方法性能较差。检测前跟踪(Track Before Detect,TBD)算法基于原始观测数据,在检测之前建立跟踪模型,在进行一定时间的信号能量积累之后对目标进行检测判决,同时输出跟踪结果,可有效地解决微弱信号的检测与跟踪问题。粒子滤波(Particle Filter,PF)算法是实现TBD算法的一种有效手段。本文主要从非合作方的角度,利用PF技术,开展基于角度测量信息的无源传感器机动弱目标检测和跟踪算法的研究工作,主要研究成果如下:首先,介绍了贝叶斯估计和粒子滤波的理论基础,提出了基于角度测量信息的无源传感器TBD处理模型,并在该模型基础上研究了基于粒子滤波检测前跟踪(PF-TBD)的原理和统一理论框架,根据该原理实现了基于未归一化权值的优效PF-TBD算法。通过仿真实验验证了基于TBD处理实现无源传感器对微弱目标定位与跟踪的可行性,并分析了算法性能的影响因素。其次,针对机动弱目标的检测和跟踪问题,研究了多模型粒子滤波(MM-PF)算法,并将其应用到TBD算法中。根据一定准则随机选择各粒子的运动模型,通过重采样技术对符合目标运动特性的粒子进行自适应筛选,并将其融合得到目标的运动状态,可解决无源机动弱目标的定位与跟踪问题,并将UKF引入到算法中,提高了算法的性能。最后,针对MM-PF算法,研究了模型集的设计准则,针对多模型PF-TBD算法处理强机动性弱目标时具有跟踪性能较差的缺陷,提出了一种基于角速度估计的改进方法。通过对目标角速度进行实时滤波估计,并将其作为模型参数来设计各时刻的模型集,用相对较少的模型个数来自适应地精确匹配目标的实时运动状态,大大提高了多模型PF-TBD算法的实用性和性能。
[Abstract]:The positioning and tracking of weak signal in military and civilian fields have a wide range of applications. Since the target signal intensity of the sensor is small intercepted, it is difficult to be separated from the noise, the traditional methods of target detecting and tracking performance is poor. The track before detect (Track Before Detect TBD) algorithm based on the original observed data, establish the tracking model before the test, to detect the target in the decision after the accumulation of signal energy for a specified period of time, while the output tracking results, which can effectively solve the problem of detection and tracking of weak signals. The particle filter (Particle Filter PF) algorithm is an effective method to achieve the TBD algorithm. In this paper, mainly from the perspective of non cooperation, the use of PF technology to carry out passive sensor angle measurement based on the information of the research work for maneuvering weak target detection and tracking algorithms, the main research results are as follows: firstly, introduced the Pattra leaf Theoretical basis of Bayesian estimation and particle filter, we propose passive sensor TBD angle measurement based on the information processing model, and study the particle filter track before detection based on the basis of the model (PF-TBD) and the principle of unified theoretical framework, according to the principle of effective PF-TBD algorithm based on non normalization weights. Through the simulation experiment to verify the feasibility of TBD processing to achieve passive sensor positioning of weak targets and tracking based on, and analyzes the factors influencing the performance of the algorithm. Secondly, the problem of detection and tracking for maneuvering target, study the multiple model particle filter (MM-PF) algorithm and its application to the TBD algorithm. According to certain criteria random motion model the choice of particles, by re sampling technology to meet the target motion characteristics of particle adaptive screening, and the target state fusion, which can solve the passive The problem of positioning and tracking maneuvering target, and the UKF is introduced into the algorithm, improve the performance of the algorithm. Finally, according to the MM-PF algorithm, study the design criterion of the model set, the multiple model PF-TBD algorithm has strong maneuverability and weak target defect tracking performance, an improved method is proposed for angular velocity estimation based on the real-time filtering. To estimate the target angular velocity, and the model parameters to design each time the model set, the real-time state of motion to accurately match the target number of models with relatively small, greatly improves the practicability and performance of multi model PF-TBD algorithm.

【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN713

【参考文献】