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

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.