Research And Implementation On Detection And Recognition Algorithm For ?-OTDR Optic Fiber Pre-warning System

The optical fiber pre-warning system based on phase-sensitive optical time-domain reflection system has good application prospects in the field of oil and gas pipeline safety monitoring.It is featured with good concealment,accurate positioning ability,reliable real-time and strong anti-electromagnetic interference ability.In recent years,researches in this field has mostly focused on optical fiber sensing system design and algorithm of detecting and recognizing vibration source rather than algorithm implementation.The detection algorithm of optic fiber pre-warning system is operated on embedded signal processing board,so the overall performance of the system is strongly affected by the performance of the software.The resources of signal processing board are limited,and many algorithms with high complexity are difficult to operate on the board.Therefore,a more fitted DSP software architecture need to be designed,which could make the implementation of the algorithm more concise,efficient and modifiable.In this paper,the detection and recognition algorithm of optic fiber pre-warning system is studied deeply,and is divided into two parts:vibration detection part and vibration type recognition part.Based on the fiber optic pre-warning system platform,this paper puts forward a solution of the detection and recognition algorithm.In terms of the vibration source detection algorithm,a two-level CFAR detection algorithm is proposed,featured with the advantages of CA-CFAR detector and GO/SO-CFAR detector.And a software architecture based on the pipeline technology is designed to improve the efficiency of the software and ensure the maintainability and stability of the software,using the sliding window structure and the ordered list structure.In terms of the vibration type recognition,characteristics of wavelet power entropy and pitch period are extracted to identify the type of vibration signal.And a software architecture based on the circular queue technology is designed for the vibration type recognition part.Each of the vibration columns is pressed into a circular queue as a task.The tasks in the circular queue are processed according to the principle of first-in first-out,making the implementation of the system software more concise and modular.The field test proves that the proposed DSP programming solution is reasonable and efficient.It achieves the expected performance on detection and recognition.Modular design as laid a good foundation for the future algorithm version,effectively promoting the practicability of the optic fiber pre-warning system.