Research On Location And Recognition Technology For Optical Fiber Pre-Warning System

As distributed optical fiber pre-warning system has the advantages of small size,light weight,low cost,long service life,corrosion resistance,good electrical insulation and strong concealment,it can be applied to perimeter security,pipeline safety monitoring and other fields.The distributed optical fiber pre-warning system based on phase-sensitive optical time domain reflectometer has become a research hotspot in the field of distributed optical fiber vibration sensing system because of its advantages such as simple structure,stable system,high sensitivity,and the ability to simultaneously locate multi-point vibration.Based on the detection principle of the system,this paper conducts an experimental study on the signal processing methods of vibration intrusion location,feature extraction and type recognition.First of all,because the vibration signal is easily submerged in the background noise and the false alarm rate of the system intrusion detection is high,a positioning detection method is proposed in this paper.The improved separating averaging and differential algorithm is used to improve the SNR and extract the useful vibration signal.The SNR is 20.4366 dB.The cell average constant false alarm rate detection algorithm is used to set the adaptable threshold.The continuous vibration section is found and the middle position of the section is judged to be the vibration position.The measured data is used to verify the effectiveness of the positioning detection method.The results show that the location detection method can effectively detect the vibration intrusion and reduce the calculation amount of the subsequent identification module.Then,in order to fully describe the vibration signal information,extracting features and selecting features method is proposed in this paper.The least square method and the wavelet threshold de-noising method are used to eliminate trend error and improve the SNR.Short time analysis method and wavelet analysis method are introduced to extract signal features.The feature selection is performed based on the criterion value of dispersion matrix.The best feature vectors of the signal are obtained to prepare for the identification module.Finally,because of the high error rate of the threshold recognition method,probabilistic neural network and support vector machine are used to identify intrusion types.The classifier is trained by the feature vector set extracted from the measured data.Particle swarm optimization is used to optimize classifier parameters.The cross-validation method is used to obtain the average recognition accuracy and the optimal dimension of the feature vector.Comparing the results of the two classifiers,the classifier with the best performance is used to classify and identify the intrusion types along the pipeline.The results show that the support vector machine identification method requires fewer signal features and has short time for optimizing parameters.The correct recognition rate can reach 95.1724%.For the distributed optical fiber pre-warning system based on phase-sensitive optical time domain reflectometer,the improved fiber vibration signal processing method proposed in this paper can identify the type of intrusion and meet the real-time requirements of the system.