Research On Signal Processing Of Intrusion Monitoring Based On ?-OTDR

Distributed optical fiber sensor has the advantages of anti-electromagnetic interference, corrosion resistance, good electrically insulating, high sensitivity, and monitoring a wide range of area. Besides, the sensing fiber can be buried underground, which bring concealment and security to the system. Therefore, distributed optical fiber sensor attract plenty of research, and recent has been used to monitor borders, military bases, power plants, oil pipelines, nuclear facilities, prison and other important places. Based on phase-sensitive Optical Time Domain Reflectrometer(?-OTDR), This paper is about a research on design a method to improve the accuracy rate of monitoring intrusion incidens. ?-OTDR with this method, which only need a single fiber, can monitor and locate intrusion incidents in different places at the same time, and the monitoring length can be as large as 50 km. This paper focus on improving the accuracy rate of monitoring intrusion incidents, carries out the following researchs.1) Analysing the application of distributed optical fiber sensing technology in security monitoring, mainly focus on the principle of monitoring intrusion incidents based on ?-OTDR.2) Focus on the difficulty to monitoring intrusion incidents, based on method of subtracting present data from earlier one, this paper research on method of combing elements, which are the past records of single point, the data of neighboring points, and the ratio of peak value. Experiments confirm that the method combing the past records of single point and the data of neighboring points achieve the best performance, the identification accuracy can be as high as 100%, and the results also confirm that this method has a good adaptation in different frequencies.3) In the method of subtracting present data from earlier one or the past records of single point, the main work is about design of three types of threshold, they are the unity threshold, the data fitting threshold, the self-adaptive mean value threshold, and experiments confirm that the self-adaptive mean value threshold achieve the best performance.4) Research on the simulation of frequency response of method combining the past records of single point and the data of neighboring points. In order to classify frequency information of different intrusion incidents, this paper design a method, which then has been confirmed by experiments.5) This paper analyzes the working principle of wavelet de-noising, and the adaptation in ?-OTDR. Based on the method to classify frequency information, this paper also analyze on pattern recognition, mainly on BP neural networks improved by genetic algorithms, for different modes of intrusion incidents classifying.