摘要
针对传统方法不能对偏振不敏感光纤振动传感网络入侵特征进行准确提取的问题,提出一种基于支持向量机与主成分分析的偏振不敏感光纤振动传感网络入侵特征提取方法。设计了一种低通滤波器,对偏振不敏感光纤振动传感网络入侵信号进行滤波,利用信号的短时能量和过零率对入侵信号进行自适应分离,引入信息损耗的概念对入侵信号数据进行离散化处理,利用支持向量机获取最优分类界面,利用主成分分析法,对入侵特征提取问题进行求解,将入侵特征重要程度利用优先级进行表示,按照特征的重要性和破坏性将其划分等级后进行特征提取,完成入侵特征提取。实验结果表明:与传统方法相比,该方法可以有效抑制和过滤噪声和其他干扰,有效的对偏振不敏感光纤振动传感网络入侵特征进行提取。
The tradition method cannot accurately extract the intrusion feature of polarization insensitive optical fi- ber vibration sensing network. Therefore, this paper proposed a extraction based on support vector machine （SVM） and principal component analysis. A low pass filter was designed to process the polarization insensitive optical fiber vibra- tion sensing network intrusion signal, which separated the intrusion signal adaptively by its short energy and zero cross- ing rate. Discretizing the intrusion signal data by the concept of information consumption, acquiring optimized classifi- cation interface by supporting vector machine, solving the intrusion feature extraction problem by the principal compo- nent analysis, to express the importance of the intrusion feature by priority, and to classify it by its importance and de- structiveness. Finally, do the extraction. The experimental results show that this method can effectively retrain and fil- ter noise and other interference, and can effectively extract the polarization insensitive optical fiber vibration sensing network intrusion feature.
作者
李刚
LI Gang(Guangxi Police College, Nanning 530023, Chin)
出处
《激光杂志》
北大核心
2017年第4期155-158,共4页
Laser Journal
基金
国家自然科学基金项目(71101137)
关键词
光纤振动传感网络
入侵特征
主成分分析
优先级
optical fiber vibration sensing network
intrusion feature
principal component analysis
priority
