基于蜂群算法优化向量机的微弱信号检测方法  

Weak Signal Detection Method Based on Support Vector Machine Optimized by Bee Colony Algorithm

作　　者：陈伟[1,2] 吕文华[3] 行鸿彦[1,2] 

机构地区：[1]南京信息工程大学江苏省气象灾害预报预警与评估协同创新中心,江苏南京210044 [2]南京信息工程大学江苏省气象探测与信息处理重点实验室,江苏南京210044 [3]中国气象局气象探测中心,北京100081

出　　处：《探测与控制学报》2018年第1期5-10,共6页Journal of Detection & Control

基　　金：国家自然科学基金项目资助(61671248);江苏省高校自然科学研究重大项目资助(15KJA460008);江苏省"六大人才高峰"计划;江苏省"信息与通信工程"优势学科项目资助

摘　　要：针对传统检测方法对混沌背景下微弱信号检测能力的不足,提出了基于蜂群算法优化支持向量机的微弱信号检测方法。该方法通过混沌信号的时间延迟和嵌入维实现相空间重构,利用蜂群算法对支持向量机的惩罚系数和核函数参数进行优化,结合支持向量机建立混沌序列的单步预测模型,从预测误差中检测淹没在混沌背景中的微弱信号(包括瞬态信号和周期信号)。以Lorenz系统和雷达实测得到的海杂波数据作为混沌背景噪声进行验证研究,仿真实验结果表明,所提方法能有效地抑制噪声对混沌背景信号的影响并检测出混沌背景噪声中的微弱信号。与传统方法相比,预测精度和检测门限方面的性能都有显著的提高。In view of the weakness of traditional detection methods for detecting weak ground ’ a weak signal detection method based on artificial bee colony algorithm optimizchines was proposed. The method realized phase space reconstruction by time delay and embedding dimension ofchaotic signals. The penalty coefficient and the kernel function parameter of support vector machine were opti-mized by artificial bee colony algorithm. Combined with support vector machine ’ singlestep the chaotic sequence was set up ’ so the weak signal in chaos could be detected from the prediction error（inclu-ding the transient signal and periodic signal）. Lorenz attractor and the data from the radar sea clutter databasewere used in the simulation, The simulation resuit showed that the method could effectivelget from chaotic signal and suppressed the effect of noise on chaotic background signal. Compareditional method , the performance of the prediction accuracy and detection threshold were improved significantly.

关 键 词：蜂群算法 支持向量机 海杂波 微弱信号检测 

分 类 号：TN911.7[电子电信—通信与信息系统]