CN115374821A - Intrusion event identification method for fiber grating perimeter security system - Google Patents
Intrusion event identification method for fiber grating perimeter security system Download PDFInfo
- Publication number
- CN115374821A CN115374821A CN202211015995.XA CN202211015995A CN115374821A CN 115374821 A CN115374821 A CN 115374821A CN 202211015995 A CN202211015995 A CN 202211015995A CN 115374821 A CN115374821 A CN 115374821A
- Authority
- CN
- China
- Prior art keywords
- intrusion
- signal
- fiber grating
- event
- signals
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Burglar Alarm Systems (AREA)
Abstract
The invention belongs to the technical field of fiber perimeter security, and particularly relates to an intrusion event identification method for a fiber grating perimeter security system, which comprises the following steps: fixedly winding the fiber grating array on the perimeter fence to form a sensing unit so as to monitor an external invasion event; calculating the zero crossing rate of the monitored intrusion signals and carrying out variation modal decomposition on the calculated zero crossing rate; calculating correlation coefficients of IMF components obtained by decomposition and the intrusion signals, selecting the IMF component with the highest correlation coefficient and extracting multi-scale fuzzy entropy of the IMF component; and inputting the signal zero crossing rate and the multi-scale fuzzy entropy serving as mixed feature vectors into a support vector machine fitted by a Sigmoid function to identify the intrusion events, and outputting the occurrence probability of each type of intrusion events while outputting the types of the intrusion events. The invention can effectively identify the man-made invasion event and the non-man-made invasion event, thereby reducing the false alarm rate of the man-made invasion event and the non-man-made invasion event.
Description
Technical Field
The invention relates to the technical field of optical fiber perimeter security, in particular to an intrusion event identification method for an optical fiber grating perimeter security system
Background
Traditional perimeter security protection system for example fence system, reveal cable type perimeter system, these systems exist easily receive electromagnetic interference, and positioning accuracy is poor, easily receives the environmental impact, and life is short, with high costs scheduling problem. The optical fiber sensing technology in the field of perimeter security mainly comprises an Optical Time Domain Reflectometer (OTDR) technology, an optical fiber interference technology and an optical Fiber Bragg Grating (FBG) sensing technology. Compared with other technologies, the fiber grating-based sensing technology is little affected by environmental interference, has higher monitoring precision and low false alarm rate, can sense the intrusion vibration signal more accurately, can set the sensing interval reasonably as required, becomes an economic and efficient identification system, and provides high cost performance for perimeter security protection in various industry fields such as petrochemical industry, energy and electric power, airports and the like.
How to accurately describe the characteristics of the intrusion vibration signal is the key for improving the accuracy rate of mode identification, and the chinese patent CN103617684B discloses an interference type optical fiber perimeter vibration intrusion identification algorithm, and performs characteristic value calculation on the result through Empirical Mode Decomposition (EMD), thereby realizing intrusion alarm under the condition of interference. However, the empirical mode decomposition has a serious mode aliasing problem, and the method is based on a single sensing optical fiber sensing signal and has poor flexibility. Chinese patent CN106127135B discloses a vibration signal feature extraction and classification recognition algorithm for rail area intrusion, which performs integrated empirical mode decomposition (EEMD) on the collected vibration signals and calculates EEMD energy entropy to realize the discrimination of the intrusion signals. However, some noise signal remains in the EEMD during the decomposition process, and only two types of intrusion signals are identified by the method.
Disclosure of Invention
The invention mainly aims to overcome the defects of the prior art and provide an intrusion event identification method for a fiber grating perimeter security system, so that the mode identification of human intrusion, non-human intrusion and non-intrusion interference events is realized, and the low false alarm rate can be realized.
In order to achieve the purpose, the invention provides the following technical scheme:
an intrusion event identification method for a fiber grating perimeter security system includes
and 4, testing the trained recognition model by using the test set.
Preferably, in step 1, performing variation modal decomposition on each extracted intrusion signal to obtain IMF components with different time-frequency characteristics and calculating a correlation coefficient between each IMF component and the original signal, where the correlation coefficient is a correlation coefficient between each IMF component and the original signal
Wherein, y si Is the data point in the s-th IMF component,
is the mean of the magnitudes of the s-th IMF component, x i (t) is the amplitude of the ith data point,
is the average of the amplitudes of the original signal. And selecting the IMF component with the highest correlation coefficient to obtain the IMF component with the highest correlation of each intrusion signal.
Preferably, in step 2, a feature extraction method of multi-scale fuzzy entropy (MFE) is used for the highest correlated IMF component in each intrusion signal, and the feature extraction method and the zero crossing rate of the intrusion signal are constructed into a mixed feature vector.
Preferably, in decomposing the signal, the number of decomposition layers K =7; the MFE parameters were chosen as follows: the embedding dimension m =2, the scale factor s =4, the similarity tolerance r =0.15std, std being the standard deviation of the intrusion signal.
Preferably, the known types of intrusion signals include man-made intrusion signals, non-man-made intrusion signals and non-intrusion interference signals; wherein, the artificial invasion signal comprises shaking and shearing, and the non-artificial invasion signal comprises wind blowing and rain. The non-invasive interference is the acquired background noise signal.
Preferably, the probability output of the intrusion event is realized by mapping the output value of the SVM between [0,1] by adopting a Sigmoid function. For any x-th class event, the probability value of the occurrence of the x-th class event and the corresponding SVM decision output are fitted through a Sigmoid function, and the x-th class event can be obtained
Wherein y represents the intrusion event type finally judged by the SVM, x is the type of the intrusion event, f x As a decision function, (A) x ,B x ) Is a pair of fitting parameters.
As can be seen from the above description of the present invention, compared with the prior art, the present invention has the following advantages:
the invention relates to an intrusion event identification method for a fiber grating perimeter security system, which comprises the steps of calculating the zero crossing rate of a monitored intrusion signal and carrying out variational modal decomposition on the calculated zero crossing rate; calculating correlation coefficients of IMF components obtained by decomposition and the intrusion signals, selecting the IMF component with the highest correlation coefficient and extracting multi-scale fuzzy entropy of the IMF component; and inputting the signal zero crossing rate and the multi-scale fuzzy entropy serving as mixed feature vectors into a support vector machine fitted by a Sigmoid function to identify the intrusion events, and outputting the occurrence probability of each type of intrusion events while outputting the types of the intrusion events. The invention can effectively identify the man-made invasion event and the non-man-made invasion event, thereby reducing the false alarm rate of the man-made invasion event and the non-man-made invasion event.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic flow chart of the identification method of the present invention;
FIG. 2 is a schematic view of the security system of the present invention;
FIG. 3 is a graph of the raw signals of an intrusion event;
FIG. 4 is a VMD result diagram of an intrusion event (slosh);
FIG. 5 is a graph of zero-crossing rate results for intrusion events;
fig. 6 is a multi-scale fuzzy entropy result diagram of the highest correlation component of the intrusion signal VMD.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to verify the application effect of the method provided by the text in the fiber grating perimeter security system, the perimeter intrusion identification detection system is established by adopting a fiber grating sensing array and utilizing the optical time domain reflection technical principle. Referring to fig. 1-2, an intrusion event recognition method for a fiber grating perimeter security system includes a fiber grating array, a demodulator and a computer; the fiber bragg grating array is arranged on the inner side of the perimeter fence in an S-shaped winding mode in advance and is fixed, so that an intrusion signal can be monitored; the fiber grating array is connected with a demodulator through an optical switch, and the demodulator is connected with a computer; a light source in the demodulator provides an original optical signal, and the demodulator demodulates the optical signal reflected from the fiber bragg grating sensor along the optical fiber, converts the optical signal into an electric signal, converts the electric signal into a digital signal through high-speed A/D (analog/digital) and transmits the digital signal to a computer for processing; and extracting signals collected by the fiber grating array under the intrusion action through a demodulator, and judging the intrusion event based on the intrusion signal identification method.
The specific construction method comprises the following steps:
Wherein, y si Is the data point in the s-th IMF component,
is the mean of the magnitudes of the s-th IMF component, x i (t) is the amplitude of the ith data point,
is the average of the amplitudes of the original signal. And selecting the IMF component with the highest correlation coefficient to obtain the IMF component with the highest correlation of each intrusion signal.
In order to determine the number of IMF components with different time-frequency characteristics obtained by the variational modal decomposition, in step 1, the center frequencies of each IMF component under different decomposition numbers are compared, and finally the decomposition number is selected to be K =7.
And 2, extracting the features of the IMF components with different time-frequency characteristics, and forming a mixed feature vector with multiple features together with the signal zero crossing rate. The method comprises the following specific steps: the highest correlation IMF component of each intrusion signal is subjected to a feature extraction method of multi-scale fuzzy entropy, and is combined with the zero crossing rate of the intrusion signal to form a mixed feature vector, and the signal features are shown in figures 5-6.
In fig. 5, five signals represent, from left to right, shaking, shearing, wind blowing, rain, and no intrusion.
3 key parameters, namely an embedding dimension m, a scale factor s and a similar tolerance r, need to be determined when extracting the multi-scale fuzzy entropy, in step 2, multi-scale fuzzy entropy values of signals under different embedding dimensions m and different scale factors s are compared, finally, the embedding dimension m =2, the scale factor s =4, the similar tolerance r =0.15std and std is the standard deviation of the intrusion signal.
The finally obtained mixed feature vector F is:
f = [ FE1, FE2, FE3, FE4, ZCR ]. The mixed feature vector comprises time domain features of the signals and irregularity features of the signals, and accuracy of intrusion event pattern recognition is improved.
And 3, inputting the mixed feature vector obtained in the step 2 into a support vector machine fitted by a Sigmoid function for training to obtain a trained recognition model. The method comprises the following specific steps: the decision value of the SVM is converted into a probability value by Sigmoid function fitting, and the principle is that the output value of the SVM is mapped to [0,1] by adopting the Sigmoid function]Thereby achieving the purpose of probability output. For any x-th type intrusion event, the probability of the occurrence of the training sample and the corresponding SVM output decision f are determined x Fitting was performed using Sigmoid function.
Wherein y represents the intrusion event type finally judged by the SVM, x is the type of the intrusion event, f x As a decision function, (A) x ,B x ) Is a pair of fitting parameters。
And 4, testing the trained recognition model by using the test set to obtain the category of the intrusion event in the test set and the occurrence probability of each type of intrusion event.
In this embodiment, the model of the demodulator is AQ6370D, the wavelength of the light source is 1550nm, and the demodulation frequency is 1000Hz.
In this embodiment, the known types of intrusion signals include man-made intrusion signals, non-man-made intrusion signals, and non-intrusion interference signals; wherein, the artificial invasion signal comprises shaking and shearing; the non-human invasion signals comprise wind blowing and rain falling; the non-invasive interference is the acquired background noise signal.
In this example, each type of signal was acquired 100 times, and as a sample, there were 500 samples for each of the five types of signals. The samples are divided into a training set and a test set, wherein 300 samples are randomly selected as the training set, and the remaining 200 samples are used as the test set.
In this embodiment, a training set is used to train the support vector machine for Sigmoid function fitting, and a test set is used to test the trained recognition model.
Collecting intrusion signals in different environments, extracting mixed characteristic vectors of the intrusion signals, and inputting the mixed characteristic vectors into a support vector machine fitted by a Sigmoid function for testing. Through testing, the total recognition rate of the perimeter defense area of the fiber bragg grating is greater than 98%. The recognition rate of the shaking event is 100%, the recognition rate of the shearing event is 98.40%, the recognition rate of the wind blowing event is 99.34%, the recognition rate of the rain event is 95.96%, the recognition rate of the non-invasive interference is 97.76%, and specific recognition results are shown in table 1.
In this embodiment, the above-described recognition process is repeated 20 times, and the average value of the recognition results of 20 times is taken as the final recognition result.
In order to verify that the provided feature extraction method has better identification performance, the method of the invention is compared with empirical mode decomposition and integrated empirical mode decomposition extraction features under the same data set and identification model, and the identification results in table 1 are specifically referred.
TABLE 1
To further prove the superiority of the proposed recognition method, the recognition results of the four classifiers of K-neighbor classifier (KNN), probabilistic Neural Network (PNN), radial basis function neural network (RBF) and Support Vector Machine (SVM) for 5 events were compared under the same data set, and the specific results are shown in table 2.
TABLE 2
In addition, several examples of intrusion event misclassification are given, and the probability of judging other types of intrusion events when the intrusion event misclassification occurs is also given. See table 3 for specific results.
TABLE 3
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (6)
1. An intrusion event recognition method for a fiber grating perimeter security system comprises a fiber grating array, a demodulator and a computer, wherein the fiber grating array is arranged and fixed on the inner side of a perimeter fence in an S-shaped winding manner in advance and is used for monitoring an intrusion signal; the fiber grating array is connected with a demodulator through an optical switch, and the demodulator is connected with a computer; the light source in the demodulator provides original optical signal, and the demodulator demodulates the optical signal reflected from the fiber grating sensor along the optical fiber, and converts the optical signal into electric signal, and then converts the electric signal into digital signal through high-speed A/D, and transmits the digital signal to the computer for processing, and the optical fiber grating sensor is characterized in that: the identification method comprises the following steps:
step 1, extracting signals collected by a fiber grating array under the intrusion action through a demodulator, calculating the zero crossing rate of the intrusion signals, and carrying out variational modal decomposition on the intrusion signals to obtain IMF components with different time-frequency characteristics, so that the technical problem of mode aliasing in the signal decomposition process is solved;
step 2, performing feature extraction on IMF components with different time-frequency characteristics, and forming a mixed feature vector of multiple features with a signal zero crossing rate, wherein the mixed feature vector comprises time domain features of signals and irregularity features of the signals, and the mixed feature vector solves the technical problem of low signal recognition rate under a single feature;
step 3, inputting the mixed characteristic vector obtained in the step 2 into a support vector machine fitted by a Sigmoid function for training to obtain a trained recognition model, and converting a decision value of the SVM into a probability value through the Sigmoid function fitting, so that the technical problem of probability output in the traditional perimeter intrusion recognition scheme is solved;
and 4, testing the trained recognition model by using the test set.
2. The intrusion event identification method for the fiber grating perimeter security system according to claim 1, wherein the intrusion event identification method comprises the following steps: in step 1, performing variation modal decomposition on each extracted intrusion signal to obtain IMF components with different time-frequency characteristics and calculating a correlation coefficient between each IMF component and an original signal, wherein the correlation coefficient is a correlation coefficient between each IMF component and the original signal
Wherein, y si Is the data point in the s-th IMF component,
is the mean of the magnitudes of the s-th IMF component, x i (t) is the amplitude of the ith data point,
is the average of the amplitudes of the original signal. And selecting the IMF component with the highest correlation coefficient to obtain the highest correlation IMF component of each intrusion signal.
3. The intrusion event identification method for the fiber grating perimeter security system according to claim 1, wherein the intrusion event identification method comprises the following steps: in step 2, a multi-scale fuzzy entropy feature extraction method is used for the highest relevant IMF component in each intrusion signal, and the highest relevant IMF component and the zero crossing rate of the intrusion signals are combined to form a mixed feature vector.
4. The intrusion event recognition method for the fiber grating perimeter security system according to claim 1, wherein: when the signal is decomposed, the number of decomposition layers K =7; the MFE parameters were chosen as follows: embedding dimension m =2, scale factor s =4, similarity tolerance r =0.15std, std is the standard deviation of the intrusion signal.
5. The intrusion event recognition method for the fiber grating perimeter security system according to any one of claims 1 to 4, wherein: known types of intrusion signals include man-made intrusion signals, non-man-made intrusion signals, and non-intrusion interference signals; wherein, the artificial invasion signal comprises shaking and shearing, and the non-artificial invasion signal comprises wind blowing and rain. The non-invasive interference is the acquired background noise signal.
6. The intrusion event recognition method for the fiber grating perimeter security system according to claim 5, wherein: and mapping the output value of the SVM between [0,1] by adopting a Sigmoid function to realize the probability output of the intrusion event. For any x-th class event, the probability value of the occurrence of the x-th class event and the corresponding SVM decision output are fitted through a Sigmoid function, and the x-th class event can be obtained
Wherein y represents the intrusion event type finally judged by the SVM, x is the type of the intrusion event, f x As a decision function, (A) x ,B x ) Is a pair of fitting parameters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211015995.XA CN115374821A (en) | 2022-08-24 | 2022-08-24 | Intrusion event identification method for fiber grating perimeter security system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211015995.XA CN115374821A (en) | 2022-08-24 | 2022-08-24 | Intrusion event identification method for fiber grating perimeter security system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115374821A true CN115374821A (en) | 2022-11-22 |
Family
ID=84067503
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211015995.XA Pending CN115374821A (en) | 2022-08-24 | 2022-08-24 | Intrusion event identification method for fiber grating perimeter security system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115374821A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116504006A (en) * | 2023-06-21 | 2023-07-28 | 吉林省日月智感互联科技有限公司 | Micro-vibration unmanned on duty alarm system with environmental parameter compensation function |
| CN118051830A (en) * | 2024-04-16 | 2024-05-17 | 齐鲁工业大学(山东省科学院) | Perimeter security intrusion event identification method |
-
2022
- 2022-08-24 CN CN202211015995.XA patent/CN115374821A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116504006A (en) * | 2023-06-21 | 2023-07-28 | 吉林省日月智感互联科技有限公司 | Micro-vibration unmanned on duty alarm system with environmental parameter compensation function |
| CN116504006B (en) * | 2023-06-21 | 2023-09-19 | 吉林省日月智感互联科技有限公司 | Micro-vibration unmanned on duty alarm system with environmental parameter compensation function |
| CN118051830A (en) * | 2024-04-16 | 2024-05-17 | 齐鲁工业大学(山东省科学院) | Perimeter security intrusion event identification method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115374821A (en) | Intrusion event identification method for fiber grating perimeter security system | |
| CN102045120B (en) | Vibration signal identification method for optical fiber perimeter system | |
| CN105184319B (en) | Recognition methods, device and the circumference intrusion alarm system of optical fiber perimeter invasion signal | |
| Kankar et al. | Rolling element bearing fault diagnosis using wavelet transform | |
| CN103236127A (en) | Fiber fence intrusion monitoring system and pattern recognition method thereof | |
| CN112255538A (en) | Multi-feature optimization fusion high-voltage circuit breaker fault diagnosis method | |
| CN111436944B (en) | Falling detection method based on intelligent mobile terminal | |
| CN105971901A (en) | Centrifugal pump fault diagnosis method based on complete ensemble empirical mode decomposition and random forest | |
| WO2009139052A1 (en) | Signal judgment method, signal judgment apparatus, program, and signal judgment system | |
| Xiao et al. | Maximal overlap discrete wavelet transform and deep learning for robust denoising and detection of power quality disturbance | |
| KR100795227B1 (en) | Method and apparatus for analyzing signal pattern of sensor array | |
| CN113376516A (en) | Medium-voltage vacuum circuit breaker operation fault self-diagnosis and early-warning method based on deep learning | |
| Chin et al. | Learn to detect: Improving the accuracy of earthquake detection | |
| CN116912586A (en) | Substation intrusion event identification method, system, equipment and medium | |
| CN117437933A (en) | Feature cluster combination generation type learning-based unsupervised detection method for fault of voiceprint signal of transformer | |
| Mota et al. | A comparison of cycle spinning versus stationary wavelet transform for the extraction of features of partial discharge signals | |
| US20230024104A1 (en) | Identification of false transformer humming using machine learning | |
| CN114510960A (en) | Method for recognizing distributed optical fiber sensor system mode | |
| CN108801437A (en) | Distributed optical fiber vibration sensing localization method and device based on disturbing signal feature extraction | |
| CN116838955A (en) | Two-stage oil and gas pipeline line interference identification method | |
| CN114093106B (en) | Intrusion signal alarm method and system based on EfficientNET classification network | |
| CN111951505B (en) | Fence vibration intrusion positioning and mode identification method based on distributed optical fiber system | |
| CN115438765A (en) | Moving target identification method, device and equipment based on evolutionary neural network | |
| CN115137374A (en) | Sleep stage oriented electroencephalogram interpretability analysis method and related equipment | |
| Zhu et al. | A novel visual transformer for long-distance pipeline pattern recognition in complex environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |