Study Of Pattern Disturbance Recognition For Φ-OTDR Fiber-optic Distributed Sensor System

Fiber-optic distributed disturbance sensor (FDDS) is wildly used in areas such as intrusion, pipelines monitoring, communication link monitoring due to its distinct advantages such as simple structure, no external supply power, wide detection range as well as no electromagnetic interference. Phase-sensitive optical time domain reflectrometer (Φ-OTDR) has become a new hot spot in the field of distributed optical fiber sensing because it has simple structure, an ability of detecting multipoint disturbances simultaneously, high resolution and accuracy of locating the disturbance event position, et al. The recognization of the disturbance patterns is crucial for the Φ-OTDR fiber-optic distributed sensor system. In this thesis, we focus on developing the pattern recognition algorithm for Φ-OTDR fiber-optic distributed sensor system. A novel multi-parameters method is proposed and demonstrated experimentally to recognize different disturbance signals.The main achievements are described as the following.(1) The theory of disturbance is investigated for Φ-OTDR fiber-optic distributed sensor system. The sensing principle and sensing signal’s characteristics are analyzed for Φ-OTDR fiber-optic distributed sensor system. Based on the theory of optical waveguide and elastic mechanics, mathematical model of the optical path of the disturbance is established for Φ-OTDR fiber-optic distributed sensor system. It offers theoretical basis of the signal processing and pattern recognition of the output signal in the following work.(2) Multi-parameters based pattern disturbance recognition is proposed for Φ-OTDR fiber-optic distributed sensor system. The multi-parameters fuzzy evaluation matrix, fuzzy attribute evaluation model and decision rule are discussed by fuzzy mathematics and signal’s characteristics are analyzed in Φ-OTDR fiber-optic distributed sensor system. Four characteristic parameters are selected for multi-parameters method based pattern recognition. Multi-parameters method based pattern recognition has great significance for reduce the complexity of the algorithm and improve the performance of the system.(3) Multi-parameters based pattern disturbance recognition is demonstrated experimentally with the setup of Φ-OTDR fiber-optic distributed sensor system. The experimental results prove that four disturbance modes including stress damage (breaking), climbing, watering (raining) and crushing (harmless artificial disturbance), can be effectively recognized by the proposed method. The accuracy of disturbance pattern recognization reaches to 91.2%,88.3%,90% and 86.7%, respectively. This method breaks though the limitation of disturbance modes recognized by current single-parameter method and makes the reliability of disturbance pattern recognization improved.