一种精确估计恒虚警检测器阈值因子的方法  

A Method to Accurately Estimate threshold Factors for Radar Constant False Alarm Ratio Detectors

作　　者：刘盼芝[1] 韩崇昭[1] 介婧[1] 

机构地区：[1]西安交通大学电子与信息工程学院,西安710049

出　　处：《西安交通大学学报》2009年第2期67-71,共5页Journal of Xi'an Jiaotong University

基　　金：自然科学基金资助项目(60574033);国家重点基础研究发展规划资助项目(2007CB311006)

摘　　要：针对恒虚警检测器在利用求解阈值因子方法时需要大量样本且搜索耗时较长的问题,提出了一种基于仿生微粒群的精确估计雷达恒虚警检测器阈值因子的方法.将指定虚警概率下的求解阈值因子问题转化为一个最小化问题,再利用微粒群方法对其进行求解,而微粒群优化采用基于种群的搜索方式.惯性因子可以随着迭代数的变化自适应调整,从而线性地减少了惯性因子的数值,致使算法具有平衡全局搜索、局部搜索的能力和较高的搜索效率.仿真结果表明,所提方法可在指定的精度下快速实现单雷达或者雷达组网等多种恒虚警检测器阈值因子的精确估计,节省了近50%的时间,提高了解的检测精度,并具有良好的鲁棒性和快速收敛等特点.A new method is proposed to improve the problem that many samples and long search time are required in the traditional methods to determine the threshold factor of constant false alarm ratio （CFAR） detector. The method estimates threshold factors for radar CFAR detectors based on particle swarm optimization （PSO） algorithm. The problem to determine the threshold factor under given false alarm probability is first converted into a minimization problem, then the minimization problem is solved using PSO method. The inertia weight is adaptively adjusted with the iteration during minimization. Therefore the method has better performance of global search and local search, and higher efficiency of scarch. Simulation results show that the method can accuralely estimate the threshold factor for single radar or radar netting system. Compared with a genetic algorithm. 50 percent of running lime is saved, and the accuracy of estimation is improved. The numerical results indicate that the proposed scheme can quickly and accurately find the estimation of threshold factors for most CFAR detectors.

关 键 词：恒虚警检测 虚警概率 阈值因子 微粒群优化 遗传算法 

分 类 号：TN952[电子电信—信号与信息处理]