The Research On Optical Fiber Intrusion Signal Identification Algorithm Based On Fourier Decomposition

At present,the widely used optical fiber early warning system in the field of pipeline safety early warning has the advantages of strong anti-interference ability and high sensitivity,and can monitor,identify and locate external intrusion events during long-distance pipeline transportation.The system uses optical fiber sensing technology to collect vibration signals generated by intrusion events,and then combines signal processing technology and pattern recognition technology to identify intrusion signals,thereby preventing oil and gas pipeline accidents caused by intrusion events.In this paper,based on the optical fiber early warning system based on the phase sensitive optical time domain reflectometry technology,the feature extraction and identification algorithm of the optical fiber intrusion signal in the system is carried out.First of all,this paper analyzes the characteristics of time-domain waveform and spectral distribution of fiber intrusion signals,and studies the conventional feature extraction methods such as pitch period,over-average rate and first-order difference sum.Aiming at the problem that it is difficult to realize the identification of multi-type optical fiber intrusion signals by using a single conventional feature extraction method.In this paper,Fourier Decomposition Method(FDM)is used to process signals with the advantages of orthogonality,completeness and locality,and a feature extraction algorithm for fiber intrusion signals is proposed.The algorithm first automatically searches for signal components from low frequency to high frequency,selects the components with strong correlation by calculating the cross-correlation coefficient between each component and the original signal,reconstructs the signal using the filtered components,and calculates the approximate entropy,energy and over-average rate and construct feature vector,and finally use SVM algorithm for model training and identification.The experimental results show that the algorithm is significantly better than the EEMD algorithm in terms of signal reconstruction,and the algorithm can effectively identify optical fiber intrusion signals such as trotting,pick-planing and passing vehicles.However,when the signal-to-noise ratio is low,using the cross-correlation coefficient to filter the components is susceptible to the influence of noise components,so that the reconstructed signal contains part of the noise energy,resulting in a low recognition rate of the algorithm.In order to better filter out noise components for signal reconstruction,this paper proposes an improved FDM algorithm based on permutation entropy.By calculating the permutation entropy of components to determine the degree of noise contained,combined with the characteristics of fiber intrusion signal energy concentrated in low frequency bands,filter out noise components.Based on the improved FDM algorithm,the approximate entropy,energy and over-average rate characteristics of the optical fiber intrusion signal are extracted.In the process of model training,through the comparative analysis of the gradient boosting tree algorithm and the support vector machine algorithm,it is verified that the improved algorithm can improve the recognition rate of optical fiber intrusion signals.The experimental results show that the feature extraction and recognition algorithm based on improved FDM proposed in this paper can effectively identify four types of optical fiber intrusion signals: trotting,tapping,pickaxe and passing cars.