仿听觉感知机制的分布式光纤振动传感扰动信号识别方法研究

Illegal intrusion and destructive behavior poses a serious threat to the basic energy facilities and wide range of regional security. Using the distributed optical fiber sensing technology is an effective means to solve the problem of the monitoring. Due to the complexity of the environment and the randomness of the intrusion, the measured signal about the intrusion will be overwhelmed by noise. Because of the lack of the prior knowledge and the hardness of the feature characterization, the false alarm rate and recognition rate of the existing.methods were still not ideal.Therefore, inspired by the mechanism of biological auditory perception, this study aimed at the interferometric distributed optical fiber sensing intrusion detection problem, put forward a new intrusion detection method in complex environment and establish a bionic intursion detection model and design the detection algorithm based on the model.This study is focused on solving those questions showed as follow. 1)Simulate the sensitivity and selectivity mechanism of the auditory system, constructs the detection model of disturbance signal; 2) The intrusion feature characterization based on information processing mechanism of bionic auditory neuron ; 3) The recognition method based on the cognitive frame of detected object. This study will enrich and develop the theories of intrusion detection methods for distributed optical fiber sensing under complex environment and provide a theoretical basis and experimental guidance for the future application.

非法入侵和破坏性行为对国家基础能源设施、大范围区域安全造成严重威胁。采用分布式光纤传感技术对这些扰动行为进行监测是解决该问题的有效手段。因大范围监测应用通常处于野外复杂环境下，使得光纤传感低频扰动信号极易被噪声淹没，特征弱且难以表征，致使现有方法识别率低、虚警率高。为此，本研究借鉴人类听觉认知机理，拟建立一套复杂环境下分布式光纤传感扰动信号识别方法：构建仿听觉认知的扰动信号辨识模型，设计识别算法，并进行性能评估。重点解决（1）仿听觉系统敏感性和选择性机制，构建扰动信号的检测模型；（2）仿听觉高阶神经元信息处理机制的信号特征提取；（3）基于认知框架的扰动信号识别方法。本研究将丰富和发展复杂环境下光纤传感扰动信号识别理论和方法，并为该方法的应用提供理论依据和实验指导。

非法入侵和破坏性行为对国家基础能源设施、大范围区域安全造成严重威胁。采用分布式光纤传感技术对这些扰动行为进行监测是解决该问题的有效手段。因大范围监测应用通常处于野外复杂环境下，使得光纤传感低频扰动信号极易被噪声淹没，特征弱且难以表征，致使现有方法识别率低、虚警率高。为此，本研究借鉴人类听觉认知机理，建立一套复杂环境下分布式光纤传感扰动信号识别方法：构建仿听觉认知的扰动信号辨识模型，设计识别算法，并进行性能评估。重点解决（1）仿听觉系统敏感性和选择性机制，构建扰动信号的检测模型；（2）仿听觉高阶神经元信息处理机制的信号特征提取；（3）基于认知框架的扰动信号识别方法。项目进展三年期间，项目组成员按照项目进展计划逐步实现了相应的研究目标。搭建了一套光纤振动信号采集系统，研究光纤扰动检测系统实现了对振动发生地点的定位算法；并对作用于传感光纤上任意大小的振动信号进行解调，为后续扰动信号识别方法的研究做好充分准备工作；针对传统的单一特征识别率不高的问题，借鉴语音信号的处理方法，结合听觉注意力机制，将光纤扰动信号的梅尔频率倒谱系数特征和基于注意力机制的显著性特征相结合，研究了一种基于多维度结合的光纤振动信号特征提取的方法；基于听觉感知机理提取信号特征，研究采用卷积神经网络通过多特征融合进行光纤振动信号的分类方法，分析卷积神经网络性能，构建最优信号识别机制。本研究将丰富和发展复杂环境下光纤传感扰动信号识别理论和方法，并为该方法的应用提供理论依据和实验指导。