Research On Alarm Method Of Optical Cable Stealing Based On Distributed Optical Fiber Vibration Sensing In Railway Communication

In recent years,driven by interests,cases of artificially excavating railway communication optical cables have emerged in endlessly,which has caused great harm to the safety of railway operation.Therefore,it is very important to prevent theft and excavation of optical cable lines.Compared with traditional cable protection technology,distributed optical fiber sensing technology has many advantages,such as high sensitivity,long monitoring distance and strong environmental adaptability.It is very suitable for the research of anti-theft and excavation methods of optical cable.Therefore,this paper proposes an alarm method of optical cable stealing based on distributed optical fiber vibration sensing.In the research of the alarm method of optical cable stealing,the key problem is to identify the fiber sensing signal generated by the pirate,and then generate an alarm signal.However,the existing classification and recognition methods of optical fiber sensing signals have some drawbacks:on the one hand,the existing signal classification and recognition methods are based on single feature extraction,the application environment of the problems studied in this paper is complex,and the single feature extraction methods commonly used in current research can not achieve good recognition effect;on the other hand,in the commonly used SVM pattern classification algorithm,when the parameter selection is inappropriate,it has a great influence on the classification performance.In view of the above problems,this paper mainly does the following research:(1)A feature extraction method based on time domain and wavelet domain composite eigenvectors is proposed to enhance the adaptability of the algorithm to complex environments.Based on the time domain and wavelet domain characteristics of optical fiber sensing signals,the sample data with vibration and non-vibration bands are analyzed and improved,and the validity of feature extraction in time domain and wavelet domain is proved.(2)A two-stage vibration event detection algorithm based on time domain-wavelet domain is proposed,which reduces a large number of complex wavelet transform calculations and improves the timeliness of the algorithm.(3)The DT-SVM multi-classification model based on adaptive genetic algorithm is studied to improve the generalization ability of DT-SVM classification model.The optimized DT-SVM multi-classification model is used to classify the three types of vibration events:train passing,pedestrian passing and artificial stealing.(4)Collect real data of the test site and conduct a large number of experiments to verify the effectiveness of the algorithm.The experimental results show that the feature extraction method based on the composite feature vector of time domain-wavelet domain is better than the feature extraction method based on single feature.