Research On The Disturbance Signal Recognition Of Distributed Optical Fiber Warning System

Distributed optical fiber warning system has many advantages,such as high sensitivity,real-time and adaptability to large-scale monitoring.It has important application prospects in civil aviation airport perimeter security.The working environment of the system is complex and the types of intrusion are changeable.This requires that it can monitor multiple intrusion disturbances and provide real-time accurate alarm while anti-interference.In this paper,an effective feature extraction algorithm is studied for the disturbance signal of the optical fiber warning system by using the method of signal recognition.A pattern recognition classifier is designed to quickly recognize the disturbance signal to improve the correct recognition rate of the disturbance signal in the system.The validity of each method is demonstrated by experimental data.This dissertation mainly does the following research work:1.The mathematical model of optical fiber warning system based on double Mach-Zehnder interferometer is deduced theoretically.For disturbance signal endpoint detection in signal acquisition,an endpoint detection method based on saturated embedding dimension fractal frame signal and information dimension is studied,which is verified by experimental signals and compared with other methods.The results show that the detection time is 0.17 s and the average error is 2%.The optical fiber warning system is built to collect different types of disturbance signals and analyze their characteristics.2.Aiming at the problem of sensitive information of disturbance signal and low recognition rate caused by single feature in the process of extracting,a disturbance signal recognition method based on local mean decomposition and serial feature fusion is studied.The method is validated by the single disturbance signal collected in the field and the disturbance signal under the wind and rain weather,separately.The results show that the average recognition rate of this method is 96% and 96.67%,and the recognition time is 0.87 s and 0.91 s respectively.The method satisfies the real-time performance and improves the recognition rate of the system.3.Aiming at the deficiencies of self-adaptability in feature extraction and generalization ability of fixed classifier in pattern recognition,a disturbance signal recognition method based on multi-fractal spectrum and improved probabilistic neural network is studied.Validating by actual signal,the results show that the average recognition rate of this method is 96.25%,and the recognition time is 1.63 s.So it can effectively identify different disturbance signals and improve the generalization ability of the classifier and the correct recognition rate of the system.