φ-OTDR光纤入侵检测识别算法研究与实现

发布时间：2018-05-10 08:59

本文选题：光纤预警系统 + 恒虚警　；参考：《北方工业大学》2017年硕士论文

【摘要】：基于相位敏感光时域反射体制(Phase-sensitive Optical Time Domain Reflection,Φ-OTDR)的光纤预警系统具有良好的隐蔽性、精准的定位能力、可靠的实时性和极强的抗电磁干扰能力等优点,在油气管道安全监测领域有着良好的应用前景。近年来,该领域的研究多集中在光纤传感体制设计和振源检测识别算法设计上,而针对算法实现方面的研究较少。光纤预警系统的检测识别算法需要在嵌入式的信号处理板卡中实现,算法实现的优劣程度影响着系统的整体性能。由于算法的复杂度较高且信号处理板卡的资源有限,因此,需要设计一套较为合理的数字信号处理器(Digital Signal Processor,DSP)软件架构,以保证算法实现简洁、高效且易于升级。本文深入研究了光纤预警系统检测识别算法,将算法分为两部分:振源的检测定位部分和振动类型识别部分,并依托光纤预警系统平台,设计了一套检测识别算法在DSP中实现的解决方案。针对振源的检测定位部分,在算法设计方面,本文对传统振源检测算法进行了改进,设计了一种两级恒虚警检测器,该检测器结合了单元平均恒虚警(Cell Averaging Constant False Alarm Rate,CA-CFAR)检测器和最大/最小选择恒虚警(Greatest-Of/Smallest-Of Constant False Alarm Rate,GO/SO-CFAR)检测器的优点,在保证检测性能的同时尽可能提高算法的执行效率;在算法实现方面,本文设计了基于流水线技术的DSP软件架构,并采用滑动窗口结构和有序链表结构对两级恒虚警检测器的处理流程进行优化,进一步提高了算法的执行效率。针对振动类型识别部分,本文提出了通过信号的小波功率熵特征和基音周期特征对振动信号进行识别的方法,并设计了基于环形队列技术的DSP软件架构,将每一待识别的振动列当作一项任务压入环形队列中,遵循先进先出(First In First Out,FIFO)的原则依次对环形队列中的任务进行处理,使得算法的执行流程更加简洁,软件的开发更具模块化。经现场测试证明,本文提出的DSP编程解决方案合理、高效地实现了检测识别算法,达到了预期的检测识别性能。模块化的软件架构设计为后续算法版本的升级和维护打下了良好的基础,有效推进了光纤预警系统的实用化进程。
[Abstract]:The optical fiber early warning system based on Phase-sensitive Optical Time Domain Reflection, 桅 -OTDR has the advantages of good concealment, accurate positioning ability, reliable real-time performance and strong anti-electromagnetic interference ability. It has a good application prospect in the field of oil and gas pipeline safety monitoring. In recent years, the research in this field has focused on the design of optical fiber sensing system and vibration source detection and identification algorithm, but less on the realization of the algorithm. The detection and recognition algorithm of optical fiber early warning system needs to be implemented in the embedded signal processing board. The performance of the system is affected by the degree of the realization of the algorithm. Due to the high complexity of the algorithm and the limited resources of the signal processing board, it is necessary to design a reasonable digital signal processor (DSP) software architecture to ensure that the algorithm is simple, efficient and easy to upgrade. In this paper, the detection and identification algorithm of optical fiber early warning system is deeply studied. The algorithm is divided into two parts: the detection and location of the vibration source and the identification of the vibration type, and rely on the optical fiber early warning system platform. Design a set of detection and recognition algorithm in DSP implementation of the solution. Aiming at the detection and location of vibration source, in the aspect of algorithm design, this paper improves the traditional detection algorithm of vibration source, and designs a two-stage constant false alarm detector (CFAR). The detector combines the advantages of the cell average CFAR Averaging Constant False Alarm CA-CFAR detector and the maximum / minimum selective CFAR detector, which ensures the detection performance and improves the efficiency of the algorithm. In this paper, the DSP software architecture based on pipeline technology is designed, and the processing flow of the two-stage CFAR detector is optimized by using the sliding window structure and the ordered linked list structure, which further improves the execution efficiency of the algorithm. In the part of vibration type recognition, this paper proposes a method to identify vibration signal by wavelet power entropy feature and pitch period feature, and designs a DSP software architecture based on ring queue technology. Each vibration sequence to be identified is pressed into the ring queue as a task, and the tasks in the ring queue are processed in turn according to the principle of first-in-first-out (FIFO), which makes the execution flow of the algorithm more concise. Software development is more modular. It is proved by field test that the proposed DSP programming solution is reasonable, and the detection and recognition algorithm is realized efficiently, and the expected detection and recognition performance is achieved. The modular software architecture design lays a good foundation for the update and maintenance of the subsequent algorithm version, and effectively promotes the practical process of the fiber optic early warning system.
【学位授予单位】：北方工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TE973.6

【参考文献】