基于PHD滤波的相控阵雷达多目标跟踪算法  

Multi-target tracking based on PHD filter for phased array radar

作　　者：袁常顺 王俊[1] 雷鹏[1] 孙进平[1] 毕严先[1] 

机构地区：[1]北京航空航天大学电子与信息工程学院,北京100191

出　　处：《系统工程与电子技术》2016年第3期539-544,共6页Systems Engineering and Electronics

基　　金：国家自然科学基金(61171122;61201318;61471019;61501011;61501012);中央高校基本科研业务费专项资金(YWF-15-GJSYS-068)资助课题

摘　　要：对于相控阵雷达方向余弦量测,采用扩展卡尔曼概率假设密度(extended Kalman probability hypothesis density,EK-PHD)滤波进行多目标跟踪时,存在目标数估计偏高和目标状态估计准确度低的问题。针对上述问题,提出了一种新的多目标跟踪算法——无偏转换量测概率假设密度(unbiased converted measurements PHD,UBCM-PHD)滤波算法。该算法采用方向余弦量测下的量测转换方法,保留了更多的量测信息;同时对转换后的量测偏差进行补偿,使量测转换误差的均值、方差准确近似原始量测高斯分布的一、二阶矩。仿真实验表明,所提算法可提高目标数和目标状态估计准确性。The extended Kalman probability hypothesis density（EK-PHD）filter has a higher bias in the estimation of the number of targets and a lower estimation accuracy of their states by using the direction cosine coordinate measurements for the phased array radar.To solve this problem,a novel multi-target tracking algorithm called unbiased converted measurements probability hypothesis density（UBCM-PHD）filter algorithm is proposed.The proposed algorithm utilizes the unbiased converted method to remain more information about the direction cosine coordinate measurements.Meanwhile,it compensates the bias caused by the converting direction cosine coordinate to Cartesian coordinate measurements,and the means and variances of the converted errors could accurately approximate the first-order and second-order moments of Gaussian distribution for original measurements.The simulation results indicate that the proposed algorithm improves the estimation accuracy of both the number of targets and their states.

关 键 词：相控阵雷达 多目标跟踪 无偏转换量测 随机有限集 概率假设密度滤波 

分 类 号：TN953[电子电信—信号与信息处理]