Design And Optimization Of Optical Fiber Fence Intrusion Monitoring System

The safety monitoring of third-party intrusion events such as climbing,overpassing,and digging has been achieved by the perimeter security monitoring system.The optical fiber fence intrusion monitoring system based on optical fiber sensing technology has the merits of low cost,corrosion resistance,and long measurement distance,and it has been widely used in aerospace,rail transit,power transmission and other industrial fields,becoming a research hotspot in new intelligent security technology.As a typical representative optical fiber sensing technology,the optical fiber vibration sensing system based on the Sagnac interference effect has the advantages of high signal restoration accuracy,large frequency response range,and high sensitivity.However,the system has been affected by the factors such as environmental interference and system noise since the continuous increase of optical fiber sensing distance.There has been a large number of nonintrusive signals in the signal detected by the system,which leads to a high false alarm rate in the system and restricts the further development of Sagnac interference optical fiber vibration sensing system.A Sagnac interference optical fiber vibration sensing scheme based on pattern recognition theory is proposed in this paper.Firstly,the signal denoising is performed based on the mel cepstrum coefficient method,then the ensemble empirical mode decomposition algorithm is used to extract the feature vector,and finally the support vector machine is used to classify the feature vector,and then the non-intrusive signal can be quickly separated from the system detection signal.The above scheme can accurately identify vibration events,reduce the false alarm rate of the system,and facilitate the system to make accurate and real-time security warnings for intrusion events.The work of this article mainly includes the following parts:(1)Firstly,the mechanism of optical fiber vibration sensing based on the Sagnac interference effect is studied in this article,and meanwhile the multi-channel ring Sagnac interference optical fiber fence intrusion monitoring system is designed and the system device parameters are selected and analyzed.(2)Secondly,three endpoint detection methods: short-term average energy,short-term zero-crossing rate and mel cepstrum coefficient are used to separate and denoise the environmental noise signal and the effective vibration signal.Then the vibration signal's energy value is selected as the eigenvalue,and wavelet decomposition,empirical mode decomposition and overall average empirical mode decomposition are used as three feature vector extraction methods to achieve feature extraction of typical events such as tapping and walking.(3)Thirdly,the support vector machine multi-classification model is built,and the Gaussian kernel function is selected to achieve the multi-dimensional feature spaces mapping of feature vectors.Then the optimal solution of the penalty factor is obtained through the training of the soft interval of the support vector machine,and the K-fold cross-validation is introduced to prevent samples from overfitting,and finally the performance of the classification model is evaluated through the performance measurement.(4)Finally,a fiber-optic fence intrusion monitoring system software is designed based on the MFC compiling environment,the system functions are achieved such as vibration perception,interface display and event classification,and the construction and debugging of the engineering prototype are completed.Experimental results and system tests show that the Sagnac interference optical fiber vibration sensing system based on the pattern recognition method can effectively monitor vibration signals in the fiber-optic fence area,and has achieved the classification accuracy rate to 89.83% for vibration events.Therefore,the false alarm rate of the system has been reduced,and a new solution is provided for the security early warning of the optical fiber fence intrusion monitoring system in the actual environment.