低信噪比条件下MIMO天波雷达多帧TBD方法研究

发布时间：2019-01-23 18:14

【摘要】：先检测后跟踪算法是传统的目标跟踪算法,利用目标检测统计量高于噪声检测统计量这一特性完成对目标的检测。然而,当积累信噪比低于13d B时,目标和噪声的统计量不可分辨,导致无法发现目标。检测前跟踪(TBD,Track-Before-Detect)利用现有雷达系统,能够对多帧原始数据进行积累,实现对目标的检测。本文围绕低信噪比条件下MIMO天波雷达多帧检测前跟踪方法,开展如下工作:首先,介绍了高频雷达的噪声模型、海杂波模型以及目标回波信号模型,并引入MIMO雷达数据模型,仿真低信噪比目标的回波谱数据。进一步的数据分析表明,检测统计量的构造和门限的求解是低信噪比条件下实现目标检测的关键。其次,根据目标数据模型,利用对未知参数的极大似然估计参数的方法推导了GLRT(The Generalized Likelihood Ratio Test)检测统计量,并且给出了门限的计算。利用信息论准则中的MDL(Minimum description length)准则,推导目标数目和方向估计算法。基于GLRT的推导,为了抑制检测帧中高斯噪声的干扰,提出多帧检测——MIMO-EL算法,该算法可充分挖掘并利用阵列采样数据的方向信息,利用最大似然估计得到信号幅度和角度的估计,进而得到目标检测统计量表达式。在低信噪比条件下,该方法比单帧GLRT检测效果更稳健,能够更好地抑制噪声从而提高目标检测能力。最后,利用目标运动特性推导检测前跟踪算法。根据匀速直线运动目标的轨迹呈直线的特性,给出基于Hough变换的TBD算法和基于粒子滤波的TBD算法。基于Hough变换的TBD算法的第一门限选取利用了结合GLRT检测器、MDL检测器以及MIMO-EL检测器的复合单帧检测算法。以单帧检测统计量和门限来进行第一门限检测。以第一门限检测结果作为Hough变换TBD算法的输入,比直接运用速度-距离谱数据作为第一门限检测数据相比较,具有更好的低信噪比条件下的检测能力。仿真结果表明,检测前跟踪算法在距离-多普勒-方向积累后的信噪比低于13d B时依旧能够实现对目标的有效检测。
[Abstract]:The first detection and then tracking algorithm is a traditional target tracking algorithm. The target detection statistics are higher than the noise detection statistics to complete the target detection. However, when the cumulative signal-to-noise ratio (SNR) is less than 13dB, the statistics of target and noise are indistinguishable and the target cannot be found. Pre-detection tracking (TBD,Track-Before-Detect) can accumulate multiple frames of raw data and realize target detection using existing radar systems. The main work of this paper is as follows: firstly, the noise model, sea clutter model and target echo signal model of high frequency radar are introduced, and the MIMO radar data model is introduced. The echo spectrum data of low SNR target are simulated. Further data analysis shows that the construction of detection statistics and the solution of threshold are the key to achieve target detection under low signal-to-noise ratio (SNR). Secondly, according to the target data model, the GLRT (The Generalized Likelihood Ratio Test) detection statistics are derived by using the method of maximum likelihood estimation of unknown parameters, and the threshold is calculated. By using the MDL (Minimum description length) criterion in the information theory criterion, an algorithm for estimating the number and direction of targets is derived. Based on the derivation of GLRT, in order to suppress the interference of Gao Si noise in the detection frame, a multi-frame detection (MIMO-EL) algorithm is proposed, which can fully mine and utilize the direction information of array sampling data. The maximum likelihood estimation is used to estimate the amplitude and angle of the signal, and then the expression of the target detection statistic is obtained. Under the condition of low SNR, this method is more robust than single frame GLRT detection, and can suppress noise better and improve the detection ability of target. Finally, the pre-detection tracking algorithm is derived by using the moving characteristics of the target. The TBD algorithm based on Hough transform and the TBD algorithm based on particle filter are presented according to the characteristic that the trajectory of the moving target is linear. The first threshold selection of TBD algorithm based on Hough transform uses a composite single-frame detection algorithm combining GLRT detector, MDL detector and MIMO-EL detector. First threshold detection is performed with single frame detection statistics and thresholds. Using the first threshold detection result as the input of the Hough transform TBD algorithm has better detection capability under low SNR than using the velocity-range spectrum data as the first threshold detection data. Simulation results show that the pre-detection tracking algorithm can still achieve effective target detection when the signal-to-noise ratio after range-Doppler direction accumulation is less than 13dB.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN958

【参考文献】