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。 Aiming at the problem that the traditional method can not accurately extract the intrusion characteristics of the polarization-insensitive fiber optic vibration sensor network, a method based on support vector machine and principal component analysis is proposed to extract the intrusion characteristics of the polarization-insensitive fiber optic vibration sensor network. A low-pass filter is designed to filter the intrusion signal of the polarization-insensitive fiber optic vibration sensor network. The short-time energy and zero-crossing rate of the signal are used to adaptively separate the intrusion signal. The concept of information loss is introduced to analyze the intrusion signal data Then we use the support vector machine to get the optimal classification interface, use the principal component analysis to solve the intrusion feature extraction problem, and use the priority of the intrusion feature to represent it. According to the importance and destructiveness of the feature, Divide the level after the feature extraction, to complete the invasion feature extraction. The experimental results show that compared with the traditional method, the proposed method can effectively suppress and filter noise and other interferences and effectively extract the intrusion characteristics of the polarization-insensitive fiber optic vibration sensing network.