CN103236127B - A kind of fiber fence system for monitoring intrusion and mode identification method thereof - Google Patents
A kind of fiber fence system for monitoring intrusion and mode identification method thereof Download PDFInfo
- Publication number
- CN103236127B CN103236127B CN201310163016.XA CN201310163016A CN103236127B CN 103236127 B CN103236127 B CN 103236127B CN 201310163016 A CN201310163016 A CN 201310163016A CN 103236127 B CN103236127 B CN 103236127B
- Authority
- CN
- China
- Prior art keywords
- sampled signal
- grader
- output
- layer
- typical case
- 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.)
- Active
Links
- 239000000835 fiber Substances 0.000 title claims abstract description 22
- 238000012544 monitoring process Methods 0.000 title claims abstract description 20
- 238000000034 method Methods 0.000 title claims abstract description 19
- 230000011218 segmentation Effects 0.000 claims abstract description 6
- 230000007935 neutral effect Effects 0.000 claims description 24
- 210000002569 neuron Anatomy 0.000 claims description 20
- 238000013461 design Methods 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 13
- 230000004913 activation Effects 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 7
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 230000008447 perception Effects 0.000 claims description 4
- 230000002123 temporal effect Effects 0.000 claims description 4
- 230000009545 invasion Effects 0.000 abstract description 5
- 238000001514 detection method Methods 0.000 abstract description 3
- 239000002360 explosive Substances 0.000 description 2
- 230000002265 prevention Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010292 electrical insulation Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/181—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using active radiation detection systems
- G08B13/183—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using active radiation detection systems by interruption of a radiation beam or barrier
- G08B13/186—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using active radiation detection systems by interruption of a radiation beam or barrier using light guides, e.g. optical fibres
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Burglar Alarm Systems (AREA)
Abstract
Disclosure one fiber fence system for monitoring intrusion and mode identification method thereof, its size according to sampled signal energy, set a dynamic threshold parameter, sampled signal is carried out segmentation by certain number, and statistical signal amplitude exceedes the LC number of threshold parameter in every section, thus obtaining LC distribution character, the distribution character calculating and adding up LC obtains a stack features value, this stack features value being inputted grader and makes decisions output, the event schema representated by the output result of grader decides whether to report to the police。The present invention is while intrusion detection and location, realize the identification to invasion signal type, the interference of noise can be removed well, size according to signal energy adjusts threshold value automatically, ensure the accuracy of the characteristic vector extracted, it is effectively increased the accuracy rate of identification, solves the wrong report problem under strong interference environments such as being applied to wind and rain well。
Description
Technical field
The present invention relates to circumference technical field of security and protection, particularly relate to a kind of fiber fence system for monitoring intrusion and mode identification method thereof。
Background technology
The basis that safety is a society and enterprise depends on for existence and development。In recent years, along with China's expanding economy, becoming stronger day by day of overall national strength, people's lives are day by day rich, and thus people are after solving living problem, day by day it is of concern that the whether safety of living。Owing to some scales are big, the appearance of the building structure that involves great expense, its circumference and property safety also exist certain potential safety hazard, and the early warning such as the complex environment perimeter security such as military area, power plant, airport, oilfield and important area particularly merits attention。Moreover, school's periphery, financial center, residential building community safety precaution also become the problem that residents become more concerned with。Therefore, one safely and effectively circumference security protection monitoring system to prevention with fight crime, maintaining public order, disaster prevention accident, reduce the aspect such as country, collective property and people's life and be just particularly important。And traditional security pre-warning system is based primarily upon electric sensor, owing to being limited by objective technique condition, conventional electrical sensor cannot be applied in the adverse circumstances such as inflammable, explosive, high temperature, strong electromagnetic, but Fibre Optical Sensor has electromagnetism interference, electrical insulation capability is good, corrosion-resistant, volume is little, the advantage such as lightweight so that it is be particularly well-suited to such as safety-protection systems such as military area, nuclear power station, airport, combustible and explosive area。
But the problem being only capable of solving intrusion alarm is inadequate, invasion signal can be divided into unwanted signals and harmless signal。Wherein, unwanted signals includes the invasion to optical fiber of people or apparatus, harmless signal includes the interference of the natural environment such as wind and rain, thunder and lightning and toy, therefore how accurately to differentiate the unwanted signals in intrusion event, by harmless signal shielding, reduce unnecessary false-alarm and false alarm, improve the intelligent level that sensory perceptual system is reported to the police, be the major issue of practical application urgent need solution。
Summary of the invention
It is an object of the invention to, by a kind of fiber fence system for monitoring intrusion and mode identification method thereof, solve the problem that background section above is mentioned。
For reaching this purpose, the present invention by the following technical solutions:
A kind of fiber fence system for monitoring intrusion, comprising: sensing optic cable, wavelength demodulation device, signal processing termination；
Described sensing optic cable is laid on fence, is in series with N number of Fiber Bragg Grating FBG (FBG) vibrating sensor, the vibration signal on perception fence in this sensing optic cable, output is to wavelength demodulation device, wherein, N is positive integer, can be actually needed setting according to user；
Described wavelength demodulation device is connected with fiber bragg grating vibration sensor, for described vibration signal is converted to sensor wavelength variable signal, and the signal of intrusion event is sampled；
Described signal processing termination is connected with wavelength demodulation device, for the sampled signal of wavelength demodulation device output is processed, intrusion event type is identified, and decides whether to report to the police。
The invention also discloses a kind of fiber fence system for monitoring intrusion mode identification method based on above-mentioned fiber fence system for monitoring intrusion, it comprises the steps:
Vibration signal on step S101, fiber bragg grating vibration sensor senses fence, output is to wavelength demodulation device；
Described vibration signal is converted to sensor wavelength variable signal by step S102, wavelength demodulation device, and with set time length system design is identified various typical case intrusion events be acquired, acquiring sampled signal a, output is to signal processing termination；
The sampled signal a process that wavelength demodulation device is exported by step S103, signal processing termination, the event schema exporting result representative according to grader decides whether to report to the police。
Especially, described step S103 specifically includes:
Step S1031, sampled signal a is carried out pretreatment；
Step S1032, temporal signatures to sampled signal a are analyzed, and build the characteristic vector of sampled signal a, and it can be used as one group of input of grader；
Step S1033, build based on the grader of neutral net and initialize, the various typical case's intrusion events being identified by system design carry out presorting and determining the grader target output of each typical case's intrusion event, the characteristic vector utilizing each typical case's intrusion event builds the training sample set of grader, and grader is trained；
The characteristic vector of the sampled signal a that step S1034, extraction wavelength demodulation device export, it is entered in the grader trained in step S1033 and carries out output judgement, the grader target output of the grader obtained output with each typical case's intrusion event is compared, if the error of the target output of grader output and some typical case's intrusion event is in a preset range, then this sampled signal a is classified as this kind of event schema, and report to the police accordingly。
Especially, described step S1031 specifically includes:
First sampled signal a is removed averaging operation, deducts the average of sampled signal a by each sample point of sampled signal, it is determined that a dynamic threshold parameter thresh,
Thresh=sum (abs (a))/length (a) * (1-sum (abs (a))/length (a))6
Wherein, a represents sampled signal, abs (a) represents that each sample point to sampled signal a takes absolute value, sum (abs (a)) represents that each sample point to sampled signal a takes absolute value and carries out sum operation, and length (a) represents the number of the sample point of sampled signal a；Then sampled signal a is carried out segmentation, every length (a)/one region Q of N number of division, thus can obtain Q1, Q2 ..., QN is N number of region altogether, and adds up the number of cross point LC respectively in regional according to below equation,
Wherein, M=length (a), Γ u} is function, when u is true time, Γ u}=1, and when u is fictitious time, Γ { u}=0。
LC is specifically defined as: total a (0) in one section of interval Q, a (1), ..., a (n-1) is n sampling point altogether, if for 2 adjacent a (i) and a (i+1), there is the value of a (i) less than dynamic threshold parameter thresh set in advance, and the value of a (i+1) is more than or equal to dynamic threshold parameter thresh set in advance, then think between these 2, there is a LC, the LC numerical value that in the interval Q of statistics, the total number of the LC that all two adjacent points obtain is in this siding-to-siding block length。
Especially, described step S1032 specifically includes:
Determine the distribution character of LC according to the number of LC in sampled signal a regional Q, and determine therefrom that a stack features value p1 of sampled signal a, p2, p3, p4, p5, and it can be used as one group of input of grader；
Wherein, p1 is the LC numerical value sum of sampled signal a first half section, p2 is the LC numerical value sum of sampled signal a second half section, p3=p1-p2, p4=p1+p2, p5=sum (abs (a)), abs (a) represent that each sample point to sampled signal a takes absolute value, and sum (abs (a)) represents that each sample point to sampled signal a takes absolute value and carries out sum operation。
Especially, the various typical case's intrusion events being identified by system design in described step S1033 carry out presorting and determine that the grader target of each typical case's intrusion event exports, and specifically include:
By N kind typical case intrusion event Si (i=1,2,3 ..., N) represent with vector (e1, e2), and thereby determine that two outputs of grader, represent with t1, t2 respectively；It is the three-layer neural network being made up of input layer, hidden layer, output layer by the classifier design based on BP neutral net, wherein, input layer has five neurons, corresponding five eigenvalue p1, p2, p3, p4, p5 that will input, hidden layer has five neurons, output layer has two neurons, respectively corresponding two outputs t1, t2；Input layer, hidden layer, output layer each neuron of each layer be responsible for receiving from the input information of last layer, and output is passed to each neuron of next layer；
Each neuronic output Y=F (W*X+b), wherein, X is neuronic input vector, and W is weight matrix, and b is offset parameter, and F is neuronic activation primitive；
Wherein, for hidden layer, all five neuronic activation primitives are tansig (), and its expression formula is tansig (n)=2/ (1+exp (-2*n))-1；For output layer, all two neuronic activation primitives are purelin (), and its expression formula is purelin (n)=n。
Especially, described step S1033 utilizes the characteristic vector of each typical case's intrusion event build the training sample set of grader, and grader is trained, specifically include:
Compose initial value to neuronic weight W all in neutral net and biasing b, the instruction sample that the characteristic vector of each typical case's intrusion event builds grader is practiced this collection P={Pi=[pi1,pi2,...,pi5] (i=1,2 ..., N) }, N is the number of samples gathered；Successively the characteristic vector Pi extracting from the signal of certain typical case intrusion event Si in training sample set P is input in network, and calculate output layer by layer according to each neuronic mathematic(al) representation, until obtaining the output layer output (e1 of neutral net, e2) value, then the value of t1, t2 is exported (e1 with the target corresponding to typical case intrusion event Si respectively, e2) compare, obtain mean square error；Adjusted every layer of neuronic weight W and biasing b layer by layer by back propagation, the error sum of squares making neutral net is minimum。
Especially, described step S1034 specifically includes:
Extract the characteristic vector of the sampled signal a of wavelength demodulation device output, be entered in the grader trained, obtain the output t of graderi1、ti2, by ti1、ti2Grader target output (e1 with each typical case's intrusion event, e2) compare, if the error of the target output of grader output and some typical case's intrusion event is in a preset range, then this sampled signal a is classified as this kind of event schema, and report to the police accordingly。
The present invention size according to sampled signal energy, set a dynamic threshold parameter thresh, sampled signal is carried out segmentation by certain number, and statistical signal amplitude exceedes the LC number of threshold parameter thresh in every section, thus obtaining LC distribution character, the distribution character calculating and adding up LC obtains a stack features value for characterizing signal, this stack features value being input to a grader based on neutral net and makes decisions output, the event schema representated by the output result of grader decides whether to report to the police。Compared with tradition fence system for monitoring intrusion, the present invention is while intrusion detection and location, realize the identification to invasion signal type, by arranging a dynamic threshold parameter higher than noise level, the interference of noise can be removed well, automatically adjust threshold value according to the size of signal energy, it is ensured that the accuracy of the characteristic vector extracted, it is effectively increased the accuracy rate of identification, solves the wrong report problem under strong interference environments such as being applied to wind and rain well。
Accompanying drawing explanation
The fiber fence system for monitoring intrusion structural representation that Fig. 1 provides for the embodiment of the present invention；
The system for monitoring intrusion mode identification method main flow chart that Fig. 2 provides for the embodiment of the present invention；
The system for monitoring intrusion mode identification method detail flowchart that Fig. 3 provides for the embodiment of the present invention；
The grader structural representation based on neutral net that Fig. 4 provides for the embodiment of the present invention。
Detailed description of the invention
Below in conjunction with drawings and Examples, the invention will be further described。It is understood that specific embodiment described herein is used only for explaining the present invention, but not limitation of the invention。It also should be noted that, for the ease of describing, accompanying drawing illustrate only part related to the present invention but not full content。
Refer to shown in Fig. 1, the fiber fence system for monitoring intrusion structural representation that Fig. 1 provides for the embodiment of the present invention。
In the present embodiment, fiber fence system for monitoring intrusion specifically includes: sensing optic cable 101, wavelength demodulation device 102, signal processing termination 103。
Described sensing optic cable 101 is laid on fence, is in series with N number of Fiber Bragg Grating FBG (FBG) vibrating sensor 104, the vibration signal on perception fence in this sensing optic cable 101, output is to wavelength demodulation device 102, wherein, N is positive integer, can be actually needed setting according to user。
Described wavelength demodulation device 102 is connected with fiber bragg grating vibration sensor 104, for described vibration signal is converted to sensor wavelength variable signal, and the signal of intrusion event is sampled。
Described signal processing termination 103 is connected with wavelength demodulation device 102, for the sampled signal of wavelength demodulation device 102 output is processed, intrusion event type is identified, and decides whether to report to the police。Signal processing termination described in the present embodiment 103 selects computer。
As in figure 2 it is shown, fiber fence system for monitoring intrusion mode identification method specifically includes following steps in the present embodiment:
Vibration signal on step S201, fiber bragg grating vibration sensor 104 perception fence, output is to wavelength demodulation device 102。
Described vibration signal is converted to sensor wavelength variable signal by step S202, wavelength demodulation device 102, and with set time length system design is identified various typical case intrusion events be acquired, acquiring sampled signal a, output is to signal processing termination 103。
The sampled signal a process that wavelength demodulation device 102 is exported by step S203, signal processing termination 103, the event schema exporting result representative according to grader decides whether to report to the police。
As it is shown on figure 3, signal processing termination 103 is as follows to the processing procedure of sampled signal:
Step S2031, sampled signal a is carried out pretreatment。
First sampled signal a being gone averaging operation, deduct the average of sampled signal a by each sample point of sampled signal, the sampled signal a to ensure various typical case's intrusion event all can be distributed in a more unified scope, is so easy to be uniformly processed。Determine a dynamic threshold parameter thresh that can do adjustment automatically according to sampled signal energy size,
Thresh=sum (abs (a))/length (a) * (1-sum (abs (a))/length (a))6
Wherein, a represents sampled signal, abs (a) represents that each sample point to sampled signal a takes absolute value, sum (abs (a)) represents that each sample point to sampled signal a takes absolute value and carries out sum operation, and length (a) represents the number of the sample point of sampled signal a。
Then sampled signal a is carried out segmentation, every length (a)/one region Q of N number of division, thus can obtain Q1, Q2 ..., QN is N number of region altogether, and adds up the number of cross point LC respectively in regional according to below equation,
Wherein, M=length (a), Γ u} is function, when u is true time, Γ u}=1, and when u is fictitious time, Γ { u}=0。
It should be noted that, LC is defined as the mathematical statistics of a kind of cross point number, it is specifically defined as: total a (0) in one section of interval Q, a (1), ..., a (n-1) is n sampling point altogether, if for 2 adjacent a (i) and a (i+1), there is the value of a (i) less than dynamic threshold parameter thresh set in advance, and the value of a (i+1) is more than or equal to dynamic threshold parameter thresh set in advance, then think between these 2, there is a LC, the LC numerical value that in the interval Q of statistics, the total number of the LC that all two adjacent points obtain is in this siding-to-siding block length。
Step S2032, temporal signatures to sampled signal a are analyzed, and build the characteristic vector of sampled signal a, and it can be used as one group of input of grader。
Determine the distribution character of LC according to the number of LC in sampled signal a regional Q, and determine therefrom that a stack features value p1 of sampled signal a, p2, p3, p4, p5, and it can be used as one group of input of grader；
Wherein, p1 is the LC numerical value sum of sampled signal a first half section, p2 is the LC numerical value sum of sampled signal a second half section, p3=p1-p2, p4=p1+p2, p5=sum (abs (a)), abs (a) represent that each sample point to sampled signal a takes absolute value, and sum (abs (a)) represents that each sample point to sampled signal a takes absolute value and carries out sum operation。
Step S2033, build based on the grader of neutral net and initialize, the various typical case's intrusion events being identified by system design carry out presorting and determining the grader target output of each typical case's intrusion event, the characteristic vector utilizing each typical case's intrusion event builds the training sample set of grader, and grader is trained。
The various typical case's intrusion events being identified by system design carry out presorting and determine that the grader target of each typical case's intrusion event exports, and specifically include:
By N kind typical case intrusion event Si (i=1,2,3 ..., N) with vector (e1, e2) represent, for instance typical case intrusion event S1 can use (0,1) to characterize, typical event S2 can use (1,0) characterizing, typical case intrusion event S3 characterizes with (1,1)。And thereby determine that and two outputs of grader represent respectively with t1, t2；As shown in Figure 4, it is the three-layer neural network being made up of input layer, hidden layer, output layer by the classifier design based on BP neutral net, wherein, input layer has five neurons, corresponding five eigenvalue p1, p2, p3, p4, p5 that will input, hidden layer has five neurons, and output layer has two neurons, respectively corresponding two outputs t1, t2；Input layer, hidden layer, output layer each neuron of each layer be responsible for receiving from the input information of last layer, and output is passed to each neuron of next layer。
Each neuronic output Y=F (W*X+b), wherein, X is neuronic input vector, and W is weight matrix, and b is offset parameter, and F is neuronic activation primitive。
Wherein, for hidden layer, all five neuronic activation primitives are tansig (), and its expression formula is tansig (n)=2/ (1+exp (-2*n))-1；For output layer, all two neuronic activation primitives are purelin (), and its expression formula is purelin (n)=n。
The characteristic vector utilizing each typical case's intrusion event builds the training sample set of grader, and grader is trained, and specifically includes:
Compose initial value to neuronic weight W all in neutral net and biasing b, the instruction sample that the characteristic vector of each typical case's intrusion event builds grader is practiced this collection P={Pi=[pi1,pi2,...,pi5] (i=1,2 ..., N) }, N is the number of samples gathered；Successively the characteristic vector Pi extracting from the signal of certain typical case intrusion event Si in training sample set P is input in network, and calculate output layer by layer according to each neuronic mathematic(al) representation, until obtaining the output layer output (e1 of neutral net, e2) value, then the value of t1, t2 is exported (e1 with the target corresponding to typical case intrusion event Si respectively, e2) compare, obtain mean square error；Adjusted every layer of neuronic weight W and biasing b layer by layer by back propagation, the error sum of squares making neutral net is minimum。
Often one Pi of input carries out once above-mentioned renewal process, and the error being performed until neutral net output reduces to acceptable degree。
So far, the training process of neutral net completes, and just the intrusion event signal of monitoring can be identified and be classified by the trained grader completed below。
The characteristic vector of the sampled signal a that step S2034, extraction wavelength demodulation device 102 export, it is entered in the grader trained in step S2033 and carries out output judgement, the grader target output of the grader obtained output with each typical case's intrusion event is compared, if the error of the target output of grader output and some typical case's intrusion event is in a preset range, then this sampled signal a is classified as this kind of event schema, and report to the police accordingly。
Extract the characteristic vector of the sampled signal a of wavelength demodulation device 102 output, be entered in the grader trained, obtain the output t of graderi1、ti2, by ti1、ti2Grader target output (e1 with each typical case's intrusion event, e2) compare, if the error of the target output of grader output and some typical case's intrusion event is in a preset range, then this sampled signal a is classified as this kind of event schema, and report to the police accordingly。Wherein, described preset range is a user-defined minimum scope, can flexible。
If occurring without intrusion event, then turn again to signals collecting step, carry out feature extraction and the classification of sampled signal next time。Thus, constantly circulation repeats the above link, can realize the real-time monitoring of intrusion event。
Technical scheme is while intrusion detection and location, realize the identification to invasion signal type, by arranging a dynamic threshold parameter higher than noise level, the interference of noise can be removed well, size according to signal energy adjusts threshold value automatically, ensure the accuracy of the characteristic vector extracted, be effectively increased the accuracy rate of identification, solve the wrong report problem under strong interference environments such as being applied to wind and rain well。
Note, above are only presently preferred embodiments of the present invention and institute's application technology principle。It will be appreciated by those skilled in the art that and the invention is not restricted to specific embodiment described here, various obvious change can be carried out for a person skilled in the art, readjust and substitute without departing from protection scope of the present invention。Therefore, although the present invention being described in further detail by above example, but the present invention is not limited only to above example, when without departing from present inventive concept, other Equivalent embodiments more can also be included, and the scope of the present invention is determined by appended right。
Claims (2)
1. a fiber fence system for monitoring intrusion, it is characterised in that including: sensing optic cable, wavelength demodulation device, signal processing termination；
Described sensing optic cable is laid on fence, is in series with N number of fiber bragg grating vibration sensor, the vibration signal on perception fence in this sensing optic cable, and output is to wavelength demodulation device, and wherein, N is positive integer, can be actually needed setting according to user；
Described wavelength demodulation device is connected with fiber bragg grating vibration sensor, for described vibration signal is converted to sensor wavelength variable signal, and the signal of intrusion event is sampled；Described signal processing termination is connected with wavelength demodulation device, for the sampled signal a process to wavelength demodulation device output, intrusion event type is identified, and decides whether to report to the police, and detailed process is as follows:
One, sampled signal a is carried out pretreatment, specifically includes: first sampled signal a is removed averaging operation, deduct the average of sampled signal a by each sample point of sampled signal, it is determined that a dynamic threshold parameter thresh,
Thresh=sum (abs (a))/length (a) * (1-sum (abs (a))/length (a))6
Wherein, a represents sampled signal, abs (a) represents that each sample point to sampled signal a takes absolute value, sum (abs (a)) represents that each sample point to sampled signal a takes absolute value and carries out sum operation, and length (a) represents the number of the sample point of sampled signal a；
Then sampled signal a is carried out segmentation, every length (a)/one region Q of N number of division, thus can obtain Q1, Q2 ..., QN is N number of region altogether, and adds up the number of cross point LC respectively in regional according to below equation,
Wherein, M=length (a), Γ u} is function, when u is true time, Γ u}=1, and when u is fictitious time, Γ { u}=0；
Two, the temporal signatures of sampled signal a is analyzed, build the characteristic vector of sampled signal a, and it can be used as one group of input of grader, specifically include: determine the distribution character of LC according to the number of LC in sampled signal a regional Q, and determine therefrom that a stack features value p1 of sampled signal a, p2, p3, p4, p5, and it can be used as one group of input of grader；
Wherein, p1 is the LC numerical value sum of sampled signal a first half section, p2 is the LC numerical value sum of sampled signal a second half section, p3=p1-p2, p4=p1+p2, p5=sum (abs (a)), abs (a) represent that each sample point to sampled signal a takes absolute value, and sum (abs (a)) represents that each sample point to sampled signal a takes absolute value and carries out sum operation；
Three, build the grader based on neutral net and initialize, the various typical case's intrusion events being identified by system design carry out presorting and determining the grader target output of each typical case's intrusion event, the characteristic vector utilizing each typical case's intrusion event builds the training sample set of grader, and grader is trained；Wherein, the described various typical case's intrusion events being identified by system design carry out presorting and determining the grader target output of each typical case's intrusion event, specifically include: by N kind typical case intrusion event Si (i=1,2,3 ..., N) with vector (e1, e2) represent, and thereby determine that two outputs of grader, represent with t1, t2 respectively；It is the three-layer neural network being made up of input layer, hidden layer, output layer by the classifier design based on BP neutral net, wherein, input layer has five neurons, corresponding five eigenvalue p1, p2, p3, p4, p5 that will input, hidden layer has five neurons, output layer has two neurons, respectively corresponding two outputs t1, t2；Input layer, hidden layer, output layer each neuron of each layer be responsible for receiving from the input information of last layer, and output is passed to each neuron of next layer；Each neuronic output Y=F (W*X+b), wherein, X is neuronic input vector, and W is weight matrix, and b is offset parameter, and F is neuronic activation primitive；Wherein, for hidden layer, all five neuronic activation primitives are tansig (), and its expression formula is tansig (n)=2/ (1+exp (-2*n))-1；For output layer, all two neuronic activation primitives are purelin (), and its expression formula is purelin (n)=n；
The described characteristic vector utilizing each typical case's intrusion event builds the training sample set of grader, and grader is trained, specifically include: compose initial value to neuronic weight W all in neutral net and biasing b, the instruction sample that the characteristic vector of each typical case's intrusion event builds grader is practiced this collection P={Pi=[pi1,pi2,...,pi5] (i=1,2 ..., N) }, N is the number of samples gathered；Successively the characteristic vector Pi extracting from the signal of certain typical case intrusion event Si in training sample set P is input in network, and calculate output layer by layer according to each neuronic mathematic(al) representation, until obtaining the output layer output (e1 of neutral net, e2) value, then the value of t1, t2 is exported (e1 with the target corresponding to typical case intrusion event Si respectively, e2) compare, obtain mean square error；Adjusted every layer of neuronic weight W and biasing b layer by layer by back propagation, the error sum of squares making neutral net is minimum；
Four, extract the characteristic vector of the sampled signal a of wavelength demodulation device output, be entered in the grader trained, obtain the output t of graderi1、ti2, by ti1、ti2Grader target output (e1 with each typical case's intrusion event, e2) compare, if the error of the target output of grader output and some typical case's intrusion event is in a preset range, then this sampled signal a is classified as this kind of event schema, and report to the police accordingly。
2. a fiber fence system for monitoring intrusion mode identification method, it is characterised in that comprise the steps:
Vibration signal on step S101, fiber bragg grating vibration sensor senses fence, output is to wavelength demodulation device；
Described vibration signal is converted to sensor wavelength variable signal by step S102, wavelength demodulation device, and with set time length system design is identified various typical case intrusion events be acquired, acquiring sampled signal a, output is to signal processing termination；
The sampled signal a process that wavelength demodulation device is exported by step S103, signal processing termination, the event schema exporting result representative according to grader decides whether to report to the police；Described step S103 specifically includes:
Step S1031, sampled signal a is carried out pretreatment, specifically includes: first sampled signal a is removed averaging operation, deduct the average of sampled signal a by each sample point of sampled signal, it is determined that a dynamic threshold parameter thresh,
Thresh=sum (abs (a))/length (a) * (1-sum (abs (a))/length (a))6
Wherein, a represents sampled signal, abs (a) represents that each sample point to sampled signal a takes absolute value, sum (abs (a)) represents that each sample point to sampled signal a takes absolute value and carries out sum operation, and length (a) represents the number of the sample point of sampled signal a；
Then sampled signal a is carried out segmentation, every length (a)/one region Q of N number of division, thus can obtain Q1, Q2 ..., QN is N number of region altogether, and adds up the number of cross point LC respectively in regional according to below equation,
Wherein, M=length (a), Γ u} is function, when u is true time, Γ u}=1, and when u is fictitious time, Γ { u}=0；
Step S1032, temporal signatures to sampled signal a are analyzed, build the characteristic vector of sampled signal a, and it can be used as one group of input of grader, specifically include: determine the distribution character of LC according to the number of LC in sampled signal a regional Q, and determine therefrom that a stack features value p1 of sampled signal a, p2, p3, p4, p5, and it can be used as one group of input of grader；
Wherein, p1 is the LC numerical value sum of sampled signal a first half section, p2 is the LC numerical value sum of sampled signal a second half section, p3=p1-p2, p4=p1+p2, p5=sum (abs (a)), abs (a) represent that each sample point to sampled signal a takes absolute value, and sum (abs (a)) represents that each sample point to sampled signal a takes absolute value and carries out sum operation；
Step S1033, build based on the grader of neutral net and initialize, the various typical case's intrusion events being identified by system design carry out presorting and determining the grader target output of each typical case's intrusion event, the characteristic vector utilizing each typical case's intrusion event builds the training sample set of grader, and grader is trained；Wherein, the various typical case's intrusion events being identified by system design carry out presorting and determine that the grader target of each typical case's intrusion event exports, and specifically include:
By N kind typical case intrusion event Si (i=1,2,3 ..., N) represent with vector (e1, e2), and thereby determine that two outputs of grader, represent with t1, t2 respectively；It is the three-layer neural network being made up of input layer, hidden layer, output layer by the classifier design based on BP neutral net, wherein, input layer has five neurons, corresponding five eigenvalue p1, p2, p3, p4, p5 that will input, hidden layer has five neurons, output layer has two neurons, respectively corresponding two outputs t1, t2；Input layer, hidden layer, output layer each neuron of each layer be responsible for receiving from the input information of last layer, and output is passed to each neuron of next layer；
Each neuronic output Y=F (W*X+b), wherein, X is neuronic input vector, and W is weight matrix, and b is offset parameter, and F is neuronic activation primitive；
Wherein, for hidden layer, all five neuronic activation primitives are tansig (), and its expression formula is tansig (n)=2/ (1+exp (-2*n))-1；For output layer, all two neuronic activation primitives are purelin (), and its expression formula is purelin (n)=n；
The described characteristic vector utilizing each typical case's intrusion event builds the training sample set of grader, and grader is trained, and specifically includes:
Compose initial value to neuronic weight W all in neutral net and biasing b, the instruction sample that the characteristic vector of each typical case's intrusion event builds grader is practiced this collection P={Pi=[pi1,pi2,...,pi5] (i=1,2 ..., N) }, N is the number of samples gathered；Successively the characteristic vector Pi extracting from the signal of certain typical case intrusion event Si in training sample set P is input in network, and calculate output layer by layer according to each neuronic mathematic(al) representation, until obtaining the output layer output (e1 of neutral net, e2) value, then the value of t1, t2 is exported (e1 with the target corresponding to typical case intrusion event Si respectively, e2) compare, obtain mean square error；Adjusted every layer of neuronic weight W and biasing b layer by layer by back propagation, the error sum of squares making neutral net is minimum；
The characteristic vector of the sampled signal a that step S1034, extraction wavelength demodulation device export, it is entered in the grader trained in step S1033 and carries out output judgement, the grader target output of the grader obtained output with each typical case's intrusion event is compared, if the error of the target output of grader output and some typical case's intrusion event is in a preset range, then this sampled signal a is classified as this kind of event schema, and report to the police accordingly；Described step S1034 specifically includes:
Extract the characteristic vector of the sampled signal a of wavelength demodulation device output, be entered in the grader trained, obtain the output t of graderi1、ti2, by ti1、ti2Grader target output (e1 with each typical case's intrusion event, e2) compare, if the error of the target output of grader output and some typical case's intrusion event is in a preset range, then this sampled signal a is classified as this kind of event schema, and report to the police accordingly。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310163016.XA CN103236127B (en) | 2013-05-06 | 2013-05-06 | A kind of fiber fence system for monitoring intrusion and mode identification method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310163016.XA CN103236127B (en) | 2013-05-06 | 2013-05-06 | A kind of fiber fence system for monitoring intrusion and mode identification method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103236127A CN103236127A (en) | 2013-08-07 |
| CN103236127B true CN103236127B (en) | 2016-06-22 |
Family
ID=48884165
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310163016.XA Active CN103236127B (en) | 2013-05-06 | 2013-05-06 | A kind of fiber fence system for monitoring intrusion and mode identification method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103236127B (en) |
Families Citing this family (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103605951A (en) * | 2013-09-11 | 2014-02-26 | 中科润程(北京)物联科技有限责任公司 | Novel behavior characteristic identification algorithm for vibration intrusion detection |
| CN103956013B (en) * | 2014-05-04 | 2016-05-25 | 中国电子科技集团公司第四十一研究所 | A kind of wind and rain disturbing signal real-time judgment method |
| CN104134303A (en) * | 2014-07-22 | 2014-11-05 | 上海光亮光电科技有限公司 | Intrusion signal identification method for optical fiber sensing systems |
| CN105374141A (en) * | 2014-08-21 | 2016-03-02 | 北京时代光华电子技术有限公司 | Radar, vibration and hit Internet of Things electronic wall and method thereof |
| CN104729667B (en) * | 2015-03-25 | 2017-11-07 | 北京航天控制仪器研究所 | A kind of disturbance kind identification method in distributed optical fiber vibration sensing system |
| CN104751580B (en) * | 2015-04-08 | 2017-07-11 | 武汉理工光科股份有限公司 | The recognition methods of distributing optical fiber sensing signal mode and system |
| CN105139560B (en) * | 2015-07-14 | 2017-08-15 | 山东康威通信技术股份有限公司 | Environmental disturbances filtering appts and its method based on prison separation net vibration monitoring |
| CN105184319B (en) * | 2015-09-02 | 2019-07-30 | 同方威视技术股份有限公司 | Recognition methods, device and the circumference intrusion alarm system of optical fiber perimeter invasion signal |
| CN105575024B (en) * | 2015-12-30 | 2018-05-08 | 杭州安远科技有限公司 | Jamproof fiber perimeter protection system and method |
| CN105469523B (en) * | 2015-12-30 | 2017-10-10 | 杭州安远科技有限公司 | The optical fiber perimeter means of defence of wind resistance rain interference |
| CN106485005B (en) * | 2016-10-14 | 2020-04-10 | 中国电力科学研究院 | Evaluation method and device for identification accuracy rate of damping ratio of power transmission tower |
| CN107507377A (en) * | 2017-08-08 | 2017-12-22 | 北京佳讯飞鸿电气股份有限公司 | The signal processing method and device of optical fiber perimeter system |
| CN108629932B (en) * | 2018-06-22 | 2019-11-12 | 苏州热工研究院有限公司 | A kind of fiber grating perimeter security system alarm method of bond area information |
| CN108877124A (en) * | 2018-06-29 | 2018-11-23 | 奇点新源国际技术开发(北京)有限公司 | Invader detection method and device applied to perimeter security system |
| EP3618019B1 (en) * | 2018-08-30 | 2021-11-10 | Infineon Technologies AG | Apparatus and method for event classification based on barometric pressure sensor data |
| CN109377691B (en) * | 2018-10-27 | 2021-05-07 | 北京控制与电子技术研究所 | Perimeter security early warning system based on fiber grating sensing |
| CN110631682A (en) * | 2019-09-26 | 2019-12-31 | 华北水利水电大学 | Strain safety monitoring method for cable tunnel bearing body |
| CN110631683A (en) * | 2019-09-26 | 2019-12-31 | 华北水利水电大学 | Building rigid part strain safety monitoring method |
| CN110853270A (en) * | 2019-10-10 | 2020-02-28 | 北京航天易联科技发展有限公司 | Optical fiber intrusion alarm processing method based on LabVIEW |
| CN111141412A (en) * | 2019-12-25 | 2020-05-12 | 深圳供电局有限公司 | Cable temperature and anti-theft dual-monitoring method and system and readable storage medium |
| CN116530063A (en) * | 2020-09-27 | 2023-08-01 | 瑞典爱立信有限公司 | Method and apparatus for detecting events in a communication network |
| CN112985508A (en) * | 2021-03-16 | 2021-06-18 | 四川通信科研规划设计有限责任公司 | Coding and communication method for multifunctional sensor output |
| CN113129530B (en) * | 2021-04-19 | 2022-05-31 | 深圳晶华相控科技有限公司 | Intelligent security electronic fence alarm system based on Internet of things and machine vision |
| CN114821997B (en) * | 2022-04-14 | 2023-11-17 | 天津大学 | Optical fiber early warning method and device based on feature extraction model |
| CN116189405B (en) * | 2022-12-05 | 2024-08-23 | 地卫二空间技术(杭州)有限公司 | Border safety monitoring and early warning device and method |
| CN115798131B (en) * | 2023-02-13 | 2023-04-28 | 成都陆迪盛华科技有限公司 | Multi-dimensional characteristic intrusion detection method based on distributed optical fibers |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101556724A (en) * | 2008-04-09 | 2009-10-14 | 中国科学院半导体研究所 | Safety management system of optical fiber perimeter and pattern recognition method thereof |
| CN102034330A (en) * | 2010-11-05 | 2011-04-27 | 电子科技大学 | Fire-prevention and invasion-prevention synchronous early warning system and signal processing method thereof |
| CN102663032A (en) * | 2012-03-26 | 2012-09-12 | 电子科技大学 | Fiber grating fence invasion event mode recognition method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4418376B2 (en) * | 2005-01-26 | 2010-02-17 | 株式会社クレヴァシステムズ | Intrusion detection sensor |
-
2013
- 2013-05-06 CN CN201310163016.XA patent/CN103236127B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101556724A (en) * | 2008-04-09 | 2009-10-14 | 中国科学院半导体研究所 | Safety management system of optical fiber perimeter and pattern recognition method thereof |
| CN102034330A (en) * | 2010-11-05 | 2011-04-27 | 电子科技大学 | Fire-prevention and invasion-prevention synchronous early warning system and signal processing method thereof |
| CN102663032A (en) * | 2012-03-26 | 2012-09-12 | 电子科技大学 | Fiber grating fence invasion event mode recognition method |
Non-Patent Citations (2)
| Title |
|---|
| Robust event classification for a fiber optic perimeter intrusion detection system using level crossing featuers and artificial neural networks;Seedahmed S. Mahmoud, Jim Katsifolis;《Proceeding of SPIE》;20100423;第7677卷;767708-1 - 767708-12 * |
| 光纤围栏扰动事件模式识别研究;王刚;黄星元;施明智;胡挺;张旭炜;李焰;《光通信技术》;20110731(第7期);51-53 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103236127A (en) | 2013-08-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103236127B (en) | A kind of fiber fence system for monitoring intrusion and mode identification method thereof | |
| CN104568122B (en) | A kind of distributed optical fiber vibration sensing system disturbance event identification and localization method | |
| CN101556724B (en) | Safety management system of optical fiber perimeter and pattern recognition method thereof | |
| Liu et al. | A high-efficiency multiple events discrimination method in optical fiber perimeter security system | |
| CN106874833A (en) | A kind of mode identification method of vibration event | |
| CN106507315B (en) | Urban traffic accident prediction technique and system based on network social intercourse media data | |
| CN103617684A (en) | Interference type optical fiber perimeter vibration intrusion recognition algorithm | |
| CN110233849A (en) | The method and system of network safety situation analysis | |
| CN106778841A (en) | The method for building up of abnormal electricity consumption detection model | |
| CN104729667A (en) | Method for recognizing disturbance type in a distributed optical fiber vibration sensing system | |
| CN103345808A (en) | Fiber Bragg grating perimeter intrusion pattern recognition method and system | |
| CN103116957B (en) | A kind of method of optical fiber perimeter security and protection system shielding climate impact | |
| CN107360152A (en) | A kind of Web based on semantic analysis threatens sensory perceptual system | |
| CN107391598A (en) | One kind threatens information automatic generation method and system | |
| CN103036745A (en) | Anomaly detection system based on neural network in cloud computing | |
| Sun et al. | Man-made threat event recognition based on distributed optical fiber vibration sensing and SE-WaveNet | |
| CN105023379A (en) | Signal identification method of fiber perimeter early-warning system of airport | |
| CN108506736A (en) | A kind of invasion signal diagnostic system and method | |
| CN105654645A (en) | Optical fiber security and protection signal processing method and system | |
| CN111597994B (en) | Optical fiber perimeter security intrusion event identification model construction method and security system | |
| CN107274324A (en) | A kind of method that accident risk assessment is carried out based on cloud service | |
| CN107730800A (en) | Anti-Interference Analysis method based on fiber-optic vibration safety pre-warning system | |
| CN101626270B (en) | Event pre-warning and classifying method by external safety pre-warning and positioning system of photoelectric composite cables | |
| Yan et al. | Event Identification for Phase-sensitive OTDR based on Boosting Ensemble Learning | |
| Jiang et al. | Multi-sensor fireproof alarm system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C41 | Transfer of patent application or patent right or utility model | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20161026 Address after: 610000 727-728, unit 2, 1, champagne international, No. 77, Tianmu Road, Chengdu hi tech Zone, Chengdu, Sichuan Patentee after: Sichuan Light Technology Co., Ltd. Address before: 214135 Jiangsu New District of Wuxi City Linghu Road No. 97 Pacific Garden of University Science and Technology Park industrial area five floor building D Patentee before: Wuxi Chengdian Optical Fiber Sensor Technology Co., Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200521 Address after: 610000 room 103, floor 1, unit 1, building 16, No. 18, East Shuxin Road, Pitong, Pixian County, Chengdu City, Sichuan Province Patentee after: Rao Yunjiang Address before: 610000 727-728, unit 2, 1, champagne international, No. 77, Tianmu Road, Chengdu hi tech Zone, Chengdu, Sichuan Patentee before: Sichuan Light Technology Co., Ltd. |