Research On Fiber-optic Vibration Source Recognition Algorithm Based On Time-frequency Characteristics

Recently,the perimeter protection under the optical fiber early warning system(OFPS)has been developed rapidly,and the research on the identification of the invading vibration source plays an important role in the system.By recognizing the type of intrusion signal,we can judge whether it is an intrusion event or not and do some preventive measures in time.This paper mainly aims at the intrusion in OFPS from the time domain and the frequency domain,designs the signal feature extraction and recognition classification,and completes the localization and the type recognition to the system intrusion signal.The first step is to detect the vibration source signal collected by the system.Background homogeneity adaptive(BHA-)CFAR is provided in this paper using the homogeneity detection,the spatial dimension detection is carried out.In order to ensure the detection of high performance,the time lost by the method should be reduced as much as possible,and the noise removal should be completed.Secondly,the detected data is preprocessed.In this paper,we mainly use Daubechies Wavelet analysis and hard threshold to reduce the noise of the detection signal.Thirdly,the preprocessing data is extracted by feature extraction.This paper is mainly from the time domain and frequency domain to extract the characteristics of the signal analysis to reflect the time-frequency characteristics of the signal.The time domain is mainly based on the temporal and spatial characteristics of the signal in the time domain,which is transformed into the gray image representation of morphology to extract the characteristics of the vibration signal.The frequency domain mainly from the fast Fourier Transform(FFT)different signal spectrum distribution has obvious difference,calculates each signal different frequency band energy ratio,and takes it as the signal classification recognition characteristic,establishes the multidimensional characteristic vector.Finally,the feature extracted data are identified.Starting from the time domain and frequency domain,the gray scale image is recognized and classified by binary corrosion operation in the time domain and frequency domain.In the frequency domain feature,the multi-dimensional eigenvector of the calculated energy ratio is sent to the classifier to recognize the vibration source signal.In the selection of classifier,the Linear Discriminant Analysis(LDA)classifier is used to identify the signal.LDA can keep the original data information to the maximum extent and distinguish the vibration signal effectively.According to the characteristics of different intrusive vibration source events,the characteristics of walking,manual operation and mechanical signals are studied.When signal analysis is carried out in time domain,when the parameters of different vibration sources are the same,the frequency domain analysis can distinguish different vibration sources by analyzing the frequency structure,and the two can complement each other.Through algorithm analysis and experimental verification,the results show that the intrusion signal feature extraction and classification recognition algorithm from the time domain and frequency domain,on the basis of verifying the feasibility of the algorithm,the vibration source recognition algorithm is further complementary.On the one hand,it can effectively reduce the recognition time;on the other hand,it can improve the recognition rate of intrusion signal through double monitoring.