Research On Intrusion Signal Recognition Of Perimeter Alarm System Based On Capacitive Disturbance

Due to the vigorous development of airport transportation production in the country and the increase of the perimeter of the flight area,the prevention management of the airport flight area is becoming more and more important.The capacitive disturbance alarm system is an important technology in the field of airport defense security.How to effectively identify intrusion signals is an important part of the capacitive disturbance perimeter alarm system.This thesis is supported by Sichuan Province Science and Technology Plan "The Major Technology and Application of Large-scale High-availability Perimeter Alarm System in Civil Aviation Airport"(No.2018GZ0069).In order to solve the current false positives problems of capacitive disturbance alarm system and identify the typical classifications of capacitive disturbance alarm signals,this thesis is taking the capacitive disturbance alarm signal as the research object and carrying out the research of pattern recognition algorithm which is suitable for capacitive disturbance of alarm system.The work of this thesis mainly covers the following aspects:Firstly,according to the working principle of the capacitive disturbance alarm system,the reason for the false alarm generated by the current capacitive disturbance perimeter alarm system is analyzed from the acquisition layer.In this thesis,the wavelet packet threshold method which can remove the multiplicative noise is used to preprocess the capacitive disturbance signal,and the wavelet packet base and the decomposition layer suitable for the capacitive disturbance signal are selected through experimental simulation.Secondly,the thesis studies the method of feature extraction of capacitive disturbance signals.The wavelet base and decomposition layer obtained by the analysis are combined with time-frequency feature analysis to extract the wavelet packet energy of capacitive disturbance signal,and their kurtosis value is calculated as a feature of capacitive disturbance alarm signal.Then introducing ITD(Intrinsic Time-Scale Decomposition)method,the thesis innovatively proposes a PRC screening system based on PRC orthogonality and using the KL divergence as a judging value for the distortion phenomenon of the ITD method and the problem of false PRC(Proper Rotation Component).When decomposing the capacitive disturbance signal,the dominant component can be adaptively selected and the false component can be screened out.On this basis,the PRC information entropy and the PRC sample entropy are calculated as the other two characteristics of the capacitive disturbance signals.Finally,the thesis studies the classifiers of capacitive disturbance signals.BP neural network model and PNN neural network model were established respectively for the three features extracted above.Their models are trained by using the measured capacitive disturbance alarm signal data in the database.The recognition test results of the two classifiers are compared and analyzed.The optimal feature extraction method and classifier method are used to construct the intrusion signal recognition algorithm of perimeter alarm system based on capacitive disturbance.In this thesis,the experimental environment is built to test the intrusion signal recognition algorithm proposed in this thesis.The experimental results show that the intrusion signal recognition algorithm in the thesis has high accuracy,which verifies the feasibility of the research content and has practical application significance for the intrusion signal recognition of the capacitive disturbance perimeter alarm system.