Research On Location And Recognition Of Traffic Incidents Based On ?-OTDR System

With the continuous reforming of China's infrastructure construction,the transportation network is becoming more and more developed.In the field of transportation,there is still no reliable,economical,and practical long-distance continuous measurement and monitoring for the detection of traffic events such as the state of the transportation network.In recent years,optical fiber vibration sensing technology,with its technical characteristics of distributed,long-distance,all-weather monitoring,has been initially applied in large-scale linear infrastructure such as oil pipeline leakage monitoring,orbital operation monitoring,and long-distance perimeter security.However,it still faces many technical difficulties.This thesis mainly takes the optical fiber vibration sensing system as the research object,and conducts in-depth research on the mechanism,induced physical quantity,signal processing,pattern recognition and other issues involved in the phase-sensitive distributed optical fiber vibration sensing system.At the same time,this thesis attempts to introduce the vibration sensing system into the traffic system to solve the problem of continuous measurement and monitoring of long-distance traffic.This thesis proposed an intelligent recognition model of optical fiber vibration signals based on Affinity Propagation Clustering(AP Clustering).In the model,a new type of vibration signal location method was originally proposed,which can filter out a large number of irrelevant non-event vibration signals and increase the recognition efficiency.At the same time,the method of Variational Mode Decomposition(VMD)is proposed on the original fiber vibration signal filtering and denoising,and it is verified in the experiment that the method is better than Both Empirical Mode Decomposition(EMD)and Ensemble Empirical Mode Decomposition(EEMD)to obtain the best signal-to-noise ratio,when combined with the new vibration signal positioning algorithm.Meanwhile,through the experimental analysis of multiple groups of different vibration events,this thesis found that the combination of VMD,vibration signal positioning and AP can achieve a high recognition rate of more than 90% of events.The characteristics of supervised learning make this intelligent recognition model have good application prospects.Based on the in-depth study of the ?-OTDR(phase-sensitive optical time-domain reflectometer)optical path and signal intelligent recognition,and based on the traditional single-channel coherent detection,due to the possibility of signal fading and noise in the single-channel system,this thesis proposes a new type of dual-channel ?-OTDR(DualChannel-?-OTDR,DC-?-OTDR)detection system.Firstly,through experiments based on the affine propagation clustering intelligent recognition model,it is verified that the DC-?-OTDR detection system can reduce the failure of vibration event detection and improve the signal-to-noise ratio of the detection system at the same time.Compared the single-channel?-OTDR(Single-Channel-?-OTDR,SC-?-OTDR)detection system,this DC-?-OTDR detection system improves event recognition accuracy by approximately 14%.Secondly,in order to solve the problem which appeared in the intelligent recognition model based on affine propagation clustering,such as complex algorithm,large overhead machine memory and difficult to achieve real-time detection,this thesis innovatively proposes a fast vibration location algorithm suitable for DC-?-OTDR system.This method also overcomes the shortcomings of traditional conventional differential method(CDM)relying on step parameter selection.According to the combined differential signal,the vibration position can be located in real time,and compared with CDM the method obtains a signal-to-noise ratio higher than 3-5d B,showing a good practical effect.In addition,due to the spatial independence of the two optical fibers in the optical cable of the DC-?-OTDR detection system,the influence of random disturbance factors on the intensity and phase of the transmitted light waves in the dual channel is not synchronized.At the same time,the 5G communication system needs higher requirements in the future.Long-distance and large-capacity optical fiber communication technology represented by wavelength division multiplexing and coherent optical communication technology will be used on a large scale.Therefore,this thesis exploratively built a dual-channel ?-OTDR sensing system based on two-mode fiber,and further studied the dual-channel ?-OTDR system based on the LP01 mode and LP11 mode in the two-mode fiber,and explored the LP01 mode and LP11 mode respectively.The vibration detection performance of the two modes are also studied.The photon lantern is used as the mode division multiplexer/demultiplexer,and the LP01 mode and LP11 mode are injected into the twomode fiber,and the corresponding backward Rayleigh scattering signals LP01-01 and LP11a-11 a are fed back to the signal processing system for the vibration signal Perform demodulation.The results show that the LP01-01 produced by the LP01 mode has higher signal-to-noise ratio and time-frequency recovery characteristics.It shows that the LP01 and LP11 mode of the two-mode fiber are able to extract the vibrating time-frequency components,giving a good foundation for feature extraction and intelligent recognition of traffic incidents.