Research On Vibration Signal Recognition Of Subway Tunnel Based On Interferometric Grating Array Sensing

In recent years,many unauthorized construction activities,such as adjacent construction and unauthorized drilling,often damaged the subway tunnel structure,thus affecting the safe operation of the subway.The safety inspection of subway tunnels has become an important engineering problem in the process of urban construction.Traditional tunnel safety detection technology can not realize distributed dynamic detection,and can only alarm when the tunnel structure has changed,cannot really avoid the occurrence of related accidents.Therefore,it is urgent to develop a prewarning detection method for subway tunnel boundary intrusion.Ultra-weak fiber Bragg grating is widely used in distributed vibration detection because of its low reflectance,wide reflection spectrum and easy to construct largecapacity multiplexing network.Moreover,it has higher sensitivity when combined with interferometric sensing technology,which is suitable for vibration detection of subway tunnels.However,in practical application,only detecting and locating vibration events is not enough to realize accurate warning,so the accurate signal recognition is necessary to reduce false positives.Based on the interference grating array vibration sensing technology and system,this paper proposes a vibration signal recognition method for subway tunnel.It realizes the identification and classification of vibration events near subway tunnels through three stages: signal preprocessing,feature extraction and recognition classification,so as to ensure the accuracy of pre-warning of boundary invasion.The main contents of the paper are as follows:(1)Research on the vibration sensing technology of interferometric ultra-weak grating array.Based on coupled-mode theory,the principle of grating sensing is derived,and the distributed sensing technology based on ultra-weak grating array is studied.The phase modulation principle of the sensing system is studied and the mathematical model of strain sensing is derived.The phase demodulation principle of the sensing system is studied and the vibration source is located according to the optical time domain reflection technique.The experimental platform is built using ultra-weak grating array to test the frequency response,stability and distributed detection characteristics of the system.(2)Research on signal preprocessing of interferometric grating array vibration sensing.Firstly,in view of the influence of various environmental and system noises on the collected vibration signals,the spectral subtraction method is proposed to reduce the noise of the collected signals,so as to effectively improve the signal-to-noise ratio of the signals.Then,to solve the problem that the recognition of a large number of nonevent signals will affect the recognition efficiency of the system,the interception technology is selected to only identify the event signals,but the accuracy of the interception will affect the recognition accuracy.Therefore,a vibration signal intercept method is proposed,which realizes the endpoint detection based on short-time power spectral density in the process of noise reduction.The root mean square(RMS)of the short-time power spectrum density of the signal in the effective frequency band after spectrum subtraction is used to describe the vibration intensity change of the signal,so as to enhance the accuracy of endpoint detection and effectively intercept the event vibration signal.(3)Research on vibration sensing signal recognition method of interferometric grating array.The feature extraction methods based on time-frequency domain,wavelet domain and empirical mode domain are studied.Aiming at the problem that the recognition precision of the above method is not high because of the single feature,an idea of improving the recognition accuracy by acquiring the characteristics of signals at more scales is put forward.The local characteristic scale decomposition method is studied and analyzed to obtain the intrinsic scale component with richer local information.At the same time,the multi-scale permutation entropy with appropriate parameters is used to maximize the difference of feature vectors among different events,so as to effectively improve the performance of the recognition method.The recognition and classification of subway tunnel vibration signals are realized by combining the support vector machine(SVM)algorithm with the multi-scale feature extraction of signals.(4)Feasibility analysis of subway tunnel vibration signal monitoring system and identification method based on interferometric grating array sensing.The experiment is conducted in the operation section of Wuhan Metro Line 7 from Hubei University of Technology Station to Yezhi Lake Station.According to the field conditions,the experimental scheme of simulating the boundary invasion is designed.The feasibility of the sensing system in the application of vibration monitoring of subway tunnels and the effect of the identification method are verified using the actual acquisition signals.The results show that the sensing system adopted in this paper can realize long-term and accurate distributed monitoring of subway tunnel vibration signals,and the accuracy of the identification method proposed in this paper is up to 96.57%,which can meet the actual needs.