点源目标被动式跟踪算法研究

【作者】 江锦；

【导师】 张洪才； 潘泉；

【作者基本信息】 西北工业大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 作为目标跟踪领域的一个重要分支，被动式跟踪对于提高武器系统在现代电子战环境下的生存能力和作战能力有着重要的研究价值和应用前景。本文从被动定向跟踪和被动定位跟踪两个方面进行研究，主要工作成果如下： 1．简要概述被动定向跟踪，推导了球坐标系下的匀速运动目标的定向跟踪模型；状态建模的不完备性引起了模型噪声方差阵带有未知径距参数，因而需要解决参数在线辨识问题。本文使用两种方法进行参数辨识：1)从滤波残差正交性原理出发，推导出参数自适应辨识滤波器—PIBF；2)根据多模型的思想，对未知参数在参数空间进行量化，进而通过交互式多模型算法(IMM)的模型切换实现参数的在线辨识。此外，本文从工程角度出发，导出了两通道解耦的滤波模型，仿真分析表明此算法在保证跟踪性能的前提下降低了计算量。 2．被动定位跟踪不可避免地会遇到非线性问题。本文将典型的Monte Carlo非线性滤波器—Unscented Kalman Filter(UKF)、Particle Filter(PF)和UnscentedParticle Filter(UPF)应用于被动定位跟踪，并在二维空间目标跟踪仿真中与经典的扩展卡尔曼滤波算法(EKF)进行了对比；本文将UKF算法与概率数据关联算法(PDA)相结合，给出了适用于杂波环境下多站被动式跟踪的UKFPDA算法。更多还原

【Abstract】 As an important branch of target tracking techniques, passive target tracking plays important roles in improving the viable ability and battle effectiveness of weapon systems in modern electronic warfares. This thesis focuses on the bearing-only target positioning and bearing-only target tracking problems .The main works are as follows:The problem of bearing-only target positioning is first outlined. The dynamic model in the spherical coordinate is presented for bearing-only target positioning. However, such model has unknown covariance of process noise, which is related to the range between the target and the passive sensor. Here two parameter identification methods are proposed: 1) Based on the principle of the orthogonality among filter residuals, a parameter identification method is proposed, which results in Parameter-Identification Based Filter (PIBF). 2) Based on multiple model idea, the parameter is quantified in its value range and then adaptively identified via model switches. In addition, the decoupling two-channel filter model is also presented, which is shown much low computation burden and satisfactory accuracy in computer simulations.The nonlinear problem is inevitable in bearing-only target tracking problem. Here the Monte Carlo based nonlinear filters are compared with the classical Extended Kalman Filter (EKF) in this field: the Unscented Kalman Filter (UKF), the Particle Filter (PF) and the Unscented Particle Filter (UPF). Furthermore, a multi-sensor passive tracking algorithm, through combining the UKF and Probability Data Association (PDA), is provided for bearing-only target tracking in dense clutters.更多还原