Classification Algorithm And Implementation Of Disturbance Signal Based On ?-OTDR Fiber-optic Distributed Sensing System

With the development of science and technology,optical-fiber sensing technology and machine learning algorithms are constantly improving.The distributed optical-fiber sensing system based on the Phase-sensitive Optical Time Domain Reflectometer(?-OTDR)has many advantages,such as simple design structure,long monitoring distance,and multi-point positioning.So it has been widely used in long-distance airport security,oil and gas pipelines detection,and culvert tunnel inspection.The classification of disturbance events based on ?-OTDR distributed optical fiber sensing system has been becoming more and more popular.This dissertation identifies five disturbance events based on AdaBoost(Adaptive Boosting)algorithm.The main work is described as follows:(1)The theories of ?-OTDR distributed optical-fiber perturbation sensing technology are studied and analysed,and a simple theoretical analysis of the output signal is carried out.And the disturbance signals under five conditions(watering,climbing,tapping,crushing,no disturbance)are extracted.The collected sample signals are subjected to preprocessing operations such as data normalization,data grouping,and data difference.These operations can eliminate the interaction between the original data and facilitate the subsequent feature extraction operations.Then the 30 features of the disturbance signal in the time domain are extracted,which lays a foundation for the classification of the disturbance signals.(2)The signal classification method for ?-OTDR distributed optical fiber disturbance sensing system based on vote is proposed,implemented and tested.The results show that the method can effectively identify five kinds of disturbance events,and the recall rates of five disturbance events reach 91.93%,98.20%,91.57%,90.83%,and 99.23%,respectively;the average recognition recall rate is 94.35%.The recognition accuracy rates of the five models reach 94.85%,87.54%,97.76%,95.83%,99.21%,and the average recognition accuracy rate is 95.04%.(3)The signal classification method for ?-OTDR distributed optical fiber disturbance sensing system based on binary code is proposed,implemented and tested.The results show that the method can effectively distinguish the five disturbance events,but the recognition rate has a certain decline compared with the method based on the voting method.The recall rates of the five disturbance events reach 89.43%,97.70%,90.90%,88.30%.and 99.13%,respectively;the average recall rate is 93.09%.The recognition accuracy rates of the five models reach 93.28%,87.35%,96.23%,92.75%,99.64%,and the average accuracy rate is 93.85%.In addition,the number of classifier is reduced,so the recognition time has a certain decline with respect to the voting method.(4)The signal classification method for ?-OTDR distributed optical fiber disturbance sensing system based on binary code and binary tree is proposed,implemented and tested.The results show that the method can effectively identify five disturbance events,and the recognition rate is improved compared with the previous two identification methods.The recall rates of the five disturbance events reach 92.27%,97.83%,94.47%,90.33%,and 98.73%,respectively;the average recall rate is 94.73%.The recognition accuracy rates of the five models reach 92.95%,90.90%,97.05%,95.67%,99.76%,and the average accuracy rate is 95.27%.The recognition recall rate and accuracy rate of the method are better than the previous identification methods,but the recognition time is increased,and the average recognition time of 100 tests is 0.2325s.