光纤周界安防系统的振动信号识别研究

发布时间：2019-03-03 17:44

【摘要】：光纤周界安防系统利用光纤作为传感器实现分布式周界安防监测，，是光纤技术在非通信领域的一项重要发展。该系统主要传感部件是振动光纤，这种设计独特的光纤对运动、压力和振动非常敏感，可沿围栏、围墙铺设，探测攀爬、敲击等入侵行为，也可以在土壤、草坪下铺设，探测踩踏等入侵行为。光纤传感器检测到的信号纷乱复杂，如何对入侵发生时引起的振动信号进行有效处理，以便对入侵行为进行识别、分类，是光纤周界安防系统的关键技术之一，处理效果将直接影响系统对入侵行为的监测性能。本文对光纤周界安防系统的振动信号特征提取与识别技术展开研究，主要内容如下：首先，分析了光纤振动传感系统的基本结构与工作原理，设计了运用小波阈值法对光纤振动信号进行消噪的方法，有效地去除了背景噪声，减弱了冗余噪声信息对后续特征提取与识别的影响。其次，研究了光纤振动信号时频域特征提取方法，包括时域与频域特征提取法、基于小波包分解的能量特征与熵特征提取法，以及Mel倒谱系数特征提取法。第三，重点研究了模糊函数用于光纤振动信号特征提取与表示，提出了将模糊函数切片作为光纤振动信号特征的方法，并使用ReliefF特征选择方法对所选切片进行了优化，以获得更稀疏的特征子集。最后，采用支持向量机进行光纤振动信号识别，分别利用本文四种特征提取方法所获得的特征向量作为输入，进行了光纤振动信号的分类识别实验，验证了模糊函数切片与支持向量机相结合的光纤振动信号识别方法的有效性和可靠性。
[Abstract]:Optical fiber perimeter security system uses optical fiber as sensor to realize distributed perimeter security monitoring. It is an important development of optical fiber technology in non-communication field. The main sensing component of the system is the vibrating fiber, which is uniquely designed to be sensitive to motion, pressure and vibration, and can be laid along fences, walls, detection, climbing, knocking, and other intrusions, as well as under soil and lawns. Detect intrusions such as stampede. The signal detected by optical fiber sensor is complicated. How to process the vibration signal caused by the invasion effectively in order to identify and classify the intrusion behavior is one of the key technologies of the optical fiber perimeter security system. The processing effect will directly affect the monitoring performance of the system to the intrusion behavior. In this paper, the vibration signal feature extraction and identification technology of optical fiber perimeter security system is studied. The main contents are as follows: firstly, the basic structure and working principle of fiber optic vibration sensing system are analyzed. A method of de-noising optical fiber vibration signal using wavelet threshold method is designed, which effectively removes the background noise and reduces the influence of redundant noise information on the subsequent feature extraction and recognition. Secondly, the time-frequency domain feature extraction method of optical fiber vibration signal is studied, including time domain and frequency domain feature extraction method, energy feature and entropy feature extraction method based on wavelet packet decomposition, and Mel cepstrum coefficient feature extraction method. Thirdly, the fuzzy function is used to extract and represent the feature of optical fiber vibration signal, and the method of taking fuzzy function slice as the feature of optical fiber vibration signal is put forward, and the selected slice is optimized by using ReliefF feature selection method. To obtain a more sparse subset of features. Finally, using support vector machine to identify optical fiber vibration signal, using the feature vector obtained by four feature extraction methods as input, the classification and recognition experiment of optical fiber vibration signal is carried out. The validity and reliability of the optical fiber vibration signal recognition method based on fuzzy function slice and support vector machine (SVM) are verified.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.7

【参考文献】