【摘要】 光纤安防监测系统信号的特征提取与识别方法是当前的研究热点。光纤振动信号的随机性、非平稳性,以及各种信号的相似性,导致信号的识别容易产生误报现象。识别入侵事件类型的关键是信号的特征提取和高效的识别方法。对光纤振动信号的各种特征提取方法和识别方法进行分析和比较,把特征提取方法分为基于小波分解的特征提取法、基于其他分解模型的特征提取方法和基于波形统计参数的特征提取法;把对光纤振动信号的识别方法分为经验阈值识别方法、支持向量机识别方法和神经网络识别方法,最后对特征提取方法和识别方法进行总结和展望。更多还原

【Abstract】 The feature extraction and recognition method of optical fiber security monitoring system is the current research hotspot. The randomness, nonstationarity and similarity of various incident signals of optical fiber vibration signals cause false identification. The key to identify the types of intrusion event is signal feature extraction and efficient recognition. The characteristics extraction methods and recognition methods of optical fiber vibration signals are analyzed and compared. Feature extraction methods are divided into feature extraction based on wavelet decomposition, feature extraction based on other decomposition models and feature extraction based on waveform statistical parameters. The identification methods of optical fiber vibration signals are divided into empirical threshold recognition, support vector machine recognition and neural network recognition. Finally, the feature extraction methods and recognition methods are summarized and prospected.更多还原