The Mach-Zehnder And ?-OTDR Fiber Perimeter System Combined Signal Acquisition And Processing

Optical fiber sensor has the advantages of high sensitivity,simple structure,strong anti-electromagnetic interference ability and adaptability to large-scale monitoring,and it has an important application prospect in the security of civil airport perimeter.However,in the process of data acquisition and signal recognition,there are some problems,such as complex working environment,changeable intrusion behavior,signal acquisition can not be synchronized and low accuracy of signal classification and recognition.As for the above problems,this paper adopt the joint system which based on the Mach-Zehnder and ?-OTDR to realize the work about intrusion signal identification and localization.Emploied the labview software to solve the problem of signal cycle preservation and synchronous acquisition.Utilized the method of Singular Value Decomposition(SVD)to denoise the signal and improve the signal-to-noise ratio.Using the improved Particle Swarm Optimization-Support Vector Machine(MPSO-SVM)to classify and recognize the signal.Data were collected in an area of tianjin airport,do data processing and verify the effectiveness of this method.The thesis mainly develops from the following aspects1.Labview software was adopted to compile a signal joint acquisition program.According to the Mach-Zehnder and ?-OTDR joint system schematic diagram to decorate experimental apparatus,adopted network-connected and buried optical fiber sensors to detect intrusion signals.Employed labview to realize joint system continuous signal acquisition,synchronous acquisition,loop experiment requirement.And read the signal data to verify whether the data acquisition program is correct.2.As for the acquisition data of optical fiber sensing perimeter warning system has the problem of partial noise interference and low signal-to-noise ratio,Adopted SVD method to realize signal denoising.A method was proposed to determine the rank order of signal reconstruction and based on the singular value sequence of second-order difference spectrum one side minimum principle.The denoising effect is verified by the measured signal andcompared with the traditional denoising method.The results show that the denoising efficiency of the proposed method is better than that of the traditional method in both signal-to-noise ratio and mean-square error.3.As for the problem of feature extraction and classification recognition of optical fiber sensing vibration signals,Serial Feature Fusion(SFF)method is firstly adopted to extract the signal features.Secondly,proposed a signal recognition method which based on improved particle swarm optimization support vector machine.By improving the inertia weight and learning factor of particle swarm optimization algorithm,the kernel function g and penalty factor C of support vector machine are optimized,and use MPSO-SVM to realize the vibration signal classification and recognition,so as to improve the recognition accuracy and algorithm efficiency.The results show that the average recognition rate of this method is97.5%,which is 5% higher than the traditional particle swarm optimization support vector machine,and it has practical application value.