CN110033581A - Airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning - Google Patents
Airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning Download PDFInfo
- Publication number
- CN110033581A CN110033581A CN201910382594.XA CN201910382594A CN110033581A CN 110033581 A CN110033581 A CN 110033581A CN 201910382594 A CN201910382594 A CN 201910382594A CN 110033581 A CN110033581 A CN 110033581A
- Authority
- CN
- China
- Prior art keywords
- signal
- sequence
- alarm method
- frame
- intrusion alarm
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/02—Mechanical actuation
- G08B13/12—Mechanical actuation by the breaking or disturbance of stretched cords or wires
- G08B13/122—Mechanical actuation by the breaking or disturbance of stretched cords or wires for a perimeter fence
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Burglar Alarm Systems (AREA)
Abstract
The airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning that the invention discloses a kind of, which includes: the vibration signal for obtaining different intrusion behaviors;The temporal signatures of signal are extracted from each frame vibration signal;And spectrogram of the vibration signal on time-frequency domain is obtained by Hilbert-Huang transform, by spectrogram and corresponding statistical method, extract the time and frequency domain characteristics of signal;All features of extraction are merged, corresponding feature vector is formed, intrusion behavior type is judged by the output of machine learning algorithm as the input of machine learning algorithm, completes identification and alarm to intrusion behavior type.The characteristics of high reliablity of the present invention, the reason of generating wrong report for traditional airport circumference intrusion alarm method is with intrusion behavior, solves the technical issues of rate of false alarm height and bad adaptability.
Description
Technical field
The present invention relates to airport perimeter protection security technology areas, more particularly to one kind to be based on Hilbert-Huang transform and machine
The airport circumference intrusion alarm method of device study.
Background technique
With the development of social progress and science and technology, the trip of people becomes more and more convenient, in fast and efficiently theory
Promotion under, aircraft has been increasingly becoming the first choice of people's trip, and airport has the scale scale of construction as important public transportation hub
Greatly, investment is high, facilities and equipment is more, operation is complicated, densely populated place, political economy influence the features such as big, carries out highly effective and safe to it
Prevention be particularly important.On the other hand, international political situation is increasingly sophisticated, and counterterrorism issues have become in order to which countries in the world are total to
With a great problem faced, this kind of for airport crowded, the region that can not be evacuated in the short time, safety problem is even more in weight
Weight.
Airport circumference refers to the airport closed area boundary for needing to carry out entity protection or electronic protection, is Flying Area in Airport
First of safety curtain being isolated from the outside is the needs for maintaining the good order in movement area, is responsible for and ensures movement area safety
Important task.Airport circumference intrusion alarm system is using sensor technology, electronic information technology and depth learning technology, to illegal
The behavior for invading airport carries out real-time monitoring, and the system for generating alarm in time.
Detection Techniques used by circumference intrusion alarm system commonplace at present mainly include infrared acquisition, microwave spy
Survey, laser-correlation, buried cable, leaky cable, earth shock detection, pressure sensitivity detection, tension fence and video analysis etc..It is red
Outer detection, microwave sounding and laser-correlation are all made of emitter and reception device, emitter launch respectively infrared ray,
Microwave or laser can stop the propagation of infrared ray, microwave and laser, reception device caused not receive when there is target invasion
Infrared ray, microwave and laser, to generate alarm signal.Buried cable and leakage cable are by pre-buried " leakage " coaxial cable week
Enclose the electromagnetic field change triggers alarm of generation.Earth shock detection and pressure sensitivity detection sensor are embedded in underground, when there is invader to exist
When walking about, run on ground, creeping, the vibration of sensor and the variation of pressure can be caused, to generate alarm.Tension fence by
The steel wire tensed in parallel connects each detector composition, and cutting, climbing, separation steel wire can trigger alarm.Video analysis is according to anti-
The variation of image is to determine whether someone invades in area.
Technology and equipment used by above-mentioned circumference intrusion alarm system is by visitors such as animal, vegetation, strong wind, heavy rain, heavy snow
Sight condition is affected, it is easy to generate a large amount of wrong report phenomenon.Therefore, existing circumference intrusion alarm system also needs to supervise
Control personnel go the authenticity of scene review alarm after receiving warning message in person, and are pocessed to warning message, this makes
Alarm system largely still very dependent on the executive capability of staff, causes the reliability of alarm system extremely low
And the great work intensity of monitoring personnel, and intelligence truly is not implemented.
Summary of the invention
For overcome the deficiencies in the prior art, the purpose of the present invention is to provide one kind to be based on Hilbert-Huang transform and machine
The airport circumference intrusion alarm method of device study, the circumference intrusion alarm system for solving the problems, such as above-mentioned are also easy to produce wrong report.
The purpose of the present invention is implemented with the following technical solutions:
A kind of airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning, method includes following step
It is rapid:
Step (1) builds front-end collection equipment to provide the acquisition that suitable experimental situation carries out data;
Step (2), artificially simulates different intrusion behaviors, obtains the vibration signal of vibrating sensor under different intrusion behaviors;
Step (3), all vibration signals collected to institute carry out sub-frame processing;
Step (4) extracts the feature of each frame signal in the time domain;
Step (5) carries out Hilbert-Huang transform to each frame signal, obtains the Hilbert-Huang transform spectrum of vibration signal
Figure extracts statistical nature of the vibration signal on time-frequency domain with statistical method by obtained spectrogram;
Step (6) merges feature of the vibration signal extracted in time domain and time-frequency domain, forms corresponding feature
Vector judges inputted intrusion behavior class by the output of the machine learning algorithm as the input of machine learning algorithm
Type completes identification and alarm to intrusion behavior type.
Further, in the step (1), front-end collection equipment uses nine axis vibration sensors, and every circumference is taken on the net
3 vibrating sensors are built, every three throw the net is grouped for one, i.e., in each grouping, vibrating sensor carries out in a manner of 3 × 3
Arrangement.
Further, in the step (2), the different intrusion behaviors artificially simulated include normal condition, state of blowing, climb
Creep for, cut net behavior and saw net behavior.
Further, in the step (3), since vibration signal is time-varying, changing features are smaller in short time range,
Before handling data, sub-frame processing is carried out to it, the signal after framing is handled as stable state, framing method are as follows:
Formula (1) framing is pressed for the vibration signal of a length of L:
In formula, lap of the overlap between adjacent two frame, overlap=wlen-inc, data will be divided into fn
Frame, the position that each frame starts in data are
Startindex=(0:(nf-1)) * inc+1 (2)
Further, in the step (4), the feature of each frame of extraction in the time domain includes short-time energy, puts down in short-term
Peak value in equal zero-crossing rate and frame.
Further, the short-time energy of signal is the quadratic sum of the amplitude of every frame signal all the points;The short-time average of signal
Zero-crossing rate indicates that signal waveform is across the number of horizontal axis in a frame signal;Peak value indicates amplitude in a frame signal in the frame of signal
More than the number of the peak value of set threshold value, for reflecting the amplitude of signal and the information of frequency.
Further, in the step (5), carrying out Hilbert-Huang transform to signal includes two steps: empirical modal
Decomposition and Hilbert transform.
Further, the process of empirical mode decomposition is as follows:
1. finding out the minimum value sequence in each local maximum value sequence drawn game portion of original series;
2. carrying out interpolation to extreme value sequence with cubic spline function, the coenvelope sequence and lower envelope of former sequence are respectively obtained
Sequence;
3. the coenvelope sequence and lower envelope sequence to per moment are averaged, equal value sequence is obtained;
4. former sequence subtracts equal value sequence, the details sequence for removing low frequency is obtained;
5. being saved if obtained details sequence meets two conditions of intrinsic mode function;Otherwise, this is thin
Sequence is saved as former sequence, repeat step 1.~4., by k iteration, until obtained new sequence meets intrinsic mode function
Definition two conditions until, obtain first intrinsic mode function;
6. the intrinsic mode function subtracted with original series obtains remaining value sequence;
7. the remaining value sequence former sequence new as one is successively extracted the 2nd, the the 3rd ... ..., directly according to above step
To m-th intrinsic mode function.
Further, for arbitrary signal x (t), Hilbert transform is defined as:
Further, for obtained every Hilbert-Huang transform spectrogram, signal on different frequency is therefrom calculated
Energy and, as statistical nature of the signal on time-frequency domain.
Further, the feature in the time domain and time-frequency domain extracted is combined, forms corresponding spy
Vector is levied as the input of machine learning algorithm and carries out invasion type identification.
Further, the machine learning algorithm be BP neural network algorithm, including an input layer, three it is hidden
Containing layer and an output layer.
Further, the input layer of the BP neural network includes 30 neurons, respectively corresponds input feature value
In feature, output layer includes 5 neurons, is exported to the recognition results of different intrusion behaviors.
Further, algorithm model is obtained by network learning and training, and field conduct does not need to reconfigure parameter.
Compared with prior art, the beneficial effects of the present invention are: high reliablity of the invention can be effectively reduced monitoring
The working strength of personnel improves working efficiency.Meanwhile the airport circumference intrusion alarm system that the present invention is different from the past, not only
It is analyzed merely with the data-signal of single-sensor, but by point to line, by line to face, to the data letter in entire area
It number is uniformly processed and is analyzed, can greatly exclude to blow, the exceedingly odious weather such as heavy rain, severe snow is to the shadow of circumference net
It rings, to be effectively reduced false alarm rate.
Detailed description of the invention
Fig. 1 is that the present invention is based on the corresponding reports of the airport circumference intrusion alarm method of Hilbert-Huang transform and machine learning
Alert system;
Fig. 2 is that the present invention is based on the processes of Hilbert-Huang transform and the airport circumference intrusion alarm method of machine learning
Figure;
Fig. 3 is the grouping vibrational waveform figure under a kind of normal condition of acquisition of the embodiment of the present invention;
Fig. 4 is the grouping vibrational waveform figure under a kind of state of blowing of acquisition of the embodiment of the present invention;
Fig. 5 is grouping vibrational waveform figure when acquisition of the embodiment of the present invention a kind of has a climbing behavior;
Fig. 6 is a kind of grouping vibrational waveform figure having when cutting net behavior of acquisition of the embodiment of the present invention;
Fig. 7 is a kind of grouping vibrational waveform figure having when sawing net behavior of acquisition of the embodiment of the present invention;
Fig. 8 is the Xi Er that grouping sensor provided in an embodiment of the present invention is in low-frequency range when vibrating under five kinds of different conditions
Bert Huang statistical nature curve graph;
Fig. 9 is the Xi Er that grouping sensor provided in an embodiment of the present invention is in Mid Frequency when vibrating under five kinds of different conditions
Bert Huang statistical nature curve graph;
Figure 10 is the uncommon of high band when grouping sensor provided in an embodiment of the present invention is in vibration under five kinds of different conditions
That Bert Huang statistical nature curve graph;
Figure 11 is a kind of BP neural network architecture diagram provided in an embodiment of the present invention.
In figure: 1, circumference net;2, vibrating sensor.
Specific embodiment
With reference to the attached drawing in the embodiment of the present invention, technical solution in the embodiment of the present invention carries out clear, complete
Ground description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Based on this
Embodiment in invention, every other reality obtained by those of ordinary skill in the art without making creative efforts
Example is applied, shall fall within the protection scope of the present invention.
It should be noted that it can be directly on another component when component is referred to as " being fixed on " another component
Or there may also be components placed in the middle.When a component is considered as " connection " another component, it, which can be, is directly connected to
To another component or it may be simultaneously present component placed in the middle.When a component is considered as " being set to " another component, it
It can be and be set up directly on another component or may be simultaneously present component placed in the middle.Term as used herein is " vertical
", " horizontal ", "left", "right" and similar statement for illustrative purposes only.
Unless otherwise defined, all technical and scientific terms used herein and belong to technical field of the invention
The normally understood meaning of technical staff is identical.Term as used herein in the specification of the present invention is intended merely to description tool
The purpose of the embodiment of body, it is not intended that in the limitation present invention.Term " and or " used herein includes one or more phases
Any and all combinations of the listed item of pass.
Fig. 1 is the distribution situation for the vibrating sensor 2 that circumference intrusion alarm system in airport is individually grouped, wherein each grouping
There are three circumference net 1,3 vibrating sensors 2 are arranged in each circumference on the net, and the mode of formation 3 × 3 is distributed.
Fig. 2 is the flow chart of the airport circumference intrusion alarm method of one embodiment of the invention, including:
Step S101: the data for the various intrusion behaviors that acquisition is artificially simulated carry out label, every kind of intrusion behavior to data
Corresponding label is as shown in table 1.
The different intrusion behaviors of table 1 and corresponding label
Intrusion behavior | Sample label |
Normal condition | [1,0,0,0,0]T |
It blows state | [0,1,0,0,0]T |
Climbing behavior | [0,0,1,0,0]T |
Cut net behavior | [0,0,0,1,0]T |
Saw net behavior | [0,0,0,0,1]T |
It is respectively the vibrational waveform of nine vibrating sensors of a certain grouping there are when different intrusion behaviors shown in Fig. 3~7
Figure, wherein horizontal axis parameter is the time, and longitudinal axis parameter is amplitude.
Step S102: sub-frame processing is carried out to the collected vibration signal of institute.In framing, it is arranged between adjacent two frame
Some overlapping.The reason is that: the time-varying of vibration signal, changing features are smaller in short time range, so as stable state
To handle, but just changed beyond this short time range vibration signal.
Formula (1) framing is pressed for the vibration signal of a length of L:
In formula, lap of the overlap between adjacent two frame, overlap=wlen-inc, data will be divided into fn
Frame, the position that each frame starts in data are
Startindex=(0:(nf-1)) * inc+1. (2)
Step S103: the temporal signatures of every frame vibration signal are extracted.Including vibration signal short-time energy, in short-term put down
Peak value in equal zero-crossing rate and frame, calculation method are as follows:
Calculate the kth frame signal of i-th of sensorShort-time energy formula are as follows:
Short-time average zero-crossing rate indicates number of the signal waveform across horizontal axis, calculation method in a frame signal are as follows:
In formula, sgn [] is sign function, i.e.,
If the kth frame signal of i-th of sensorFrame in peak value beThe Rule of judgment of peak value such as formula in frame
(6) and shown in formula (7)
In formula, threshold value TH is taken as 300.IfMeet formula (6) and formula (7), be then denoted as a peak point,It is final
Meet conditionNumber.
Step S104: Hilbert-Huang transform is carried out to each frame signal, obtains Hilbert-Huang transform spectrogram.Wherein,
Hilbert-Huang transform consists of two parts, i.e. empirical mode decomposition (EMD) and Hilbert spectral analysis.
The process of empirical mode decomposition is as follows:
1, the minimum value sequence in each local maximum value sequence drawn game portion of original series x (t) is found out.
2, interpolation is carried out to extreme value sequence with cubic spline function, respectively obtains the coenvelope sequence u (t) of former sequence under
Envelope sequence l (t).
3, the u (t) and l (t) at per moment are averaged, obtain equal value sequence
m11(t)=[u (t)+l (t)]/2 (8)
4, former sequence subtracts equal value sequence, obtains the details sequence for removing low frequency
h11(t)=x (t)-m11(t) (9)
If the h 5, obtained11(t) two conditions for meeting IMF, then saved;Otherwise, h11(t) as former sequence
Column repeat step 1~4, by k iteration:
h1k(t)=h1(k-1)(t)-m1k(t) (10)
Until obtained new sequence meets two conditions of the definition of IMF, then c is enabled1(t)=h1k(t), thus
To first IMFc1(t).It is above exactly the process that EMD screens first IMF.
6, c next is subtracted with original series1(t), remaining value sequence is obtained
r1(t)=x (t)-c1(t) (11)
7, r1(t) the former sequence new as one successively extracts the 2nd, the the 3rd ... ..., until M according to above step
A IMFcM(t).If cM(t), rM(t) energy is too small or rM(t) become a monotonic sequence or only one extreme point
Sequence, just IMF new never again is extractable comes out.In this way, original series, which calculate x (t), to be expressed as followsin:
Hilbert transform is the important transformation of one of signal analysis, for arbitrary signal x (t), Hilbert
Transform definition are as follows:
Analytic signal q (t) corresponding to x (t) in this way are as follows:
Q (t)=x (t)+iy (t)=a (t) eiθ(t) (14)
Wherein a (t) and θ (t) is respectively the instantaneous amplitude and instantaneous phase of signal x (t), is calculated according to the following formula:
The instantaneous frequency of signal further is obtained to time derivation by instantaneous phase:
By above-mentioned empirical mode decomposition process and the definition of Hilbert transform, available signal is in time frequency
Amplitude distribution H (w, t) in rate plane, referred to as hilbert spectrum, as shown in formula (18):
In formula, T is the entire sample duration of signal, and H (ω, t) is the Hilbert spectrum of signal.
By formula (19) as it can be seen that boundary spectrum h (ω) is integral of the time-frequency spectrum to time shaft, boundary spectrum expresses each frequency and exists
Amplitude (or energy) contribution in the overall situation, it represents the cumulative amplitude of the total data in statistical significance, reflects probability
Accumulation amplitude of the amplitude on entire time span in meaning.
Step S105: by obtained Hilbert-Huang transform spectrogram, statistical nature of the signal on time-frequency domain is extracted.
The energy statistics feature of signal on different frequency is extracted from every spectrogram, wherein the division range of frequency be chosen to be (0,
20], (20,40] and (40,100].Energy statistics feature E of the signal within the scope of 0~20Hz0~20It is signal in this frequency model
Energy in enclosing and, can similarly obtain energy statistics feature E of the signal within the scope of 20~40Hz and 40~100Hz20~40With
E40~100。
It is respectively there are when different intrusion behaviors, according to the Hilbert-Huang transform of a certain packet signal shown in Fig. 8~10
The statistical nature of the obtained different frequency sections of spectrogram, wherein horizontal axis parameter be frame number, longitudinal axis parameter be energy and, for display
It is more intuitive, by the Conversion of measurement unit of energy sum be dB in figure, conversion formula is E (dB)=20 × lg (E).
Step S106: the feature vector of different intrusion behaviors is formed.Merge extracted vibration signal in the time domain and when
All features on frequency domain form the feature vector that corresponding size is 1 × 30.
Step S107: training BP neural network completes the identification to different invasion types.The BP neural network built has
One input layer, three hidden layers and an output layer composition.Wherein, input layer includes 30 neurons, and input is mentioned respectively
3 temporal signatures and the spectrogram statistical nature in different frequency scope of nine sensors in same grouping taken.Output
Layer includes 5 neurons, and output format is as shown in table 1.BP neural network overall structure is as shown in figure 11.
The present invention is different from the way that traditional circumference intrusion alarm system is only made a decision with the data of single sensor, needle
On blow, rain, the exceedingly odious weather such as severe snow influences circumference the characteristics of, the linkage of a certain range of vibrating sensor is got up
It is analyzed, the identification to intrusion behavior is expanded to the identification in region from the identification of single-point, significantly reduces the void of system
Alert rate.Meanwhile intrusion behavior knowledge is greatly improved by the training of a large amount of sample data with the algorithm of machine learning
Other accuracy rate.
The aforementioned description to specific exemplary embodiment of the invention is in order to illustrate and illustration purpose.These descriptions
It is not wishing to limit the invention to disclosed precise forms, and it will be apparent that can much be changed according to appeal introduction
And variation.The purpose of selecting and describing the exemplary embodiment is that explaining specific principle of the invention and its actually answering
With so that those skilled in the art can be realized and utilize a variety of different exemplary implementation schemes of the invention and
Various chooses and changes, i.e., according to the above description of the technical scheme and ideas, those skilled in the art can be done
Various other corresponding changes and deformation out, and all these change and deformation all should belong to the claims in the present invention
Protection scope within.
Claims (14)
1. the airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning, which is characterized in that method includes
Following steps:
Step (1) builds front-end collection equipment to provide the acquisition that suitable experimental situation carries out data;
Step (2), artificially simulates different intrusion behaviors, obtains the vibration signal of vibrating sensor under different intrusion behaviors;
Step (3), all vibration signals collected to institute carry out sub-frame processing;
Step (4) extracts the feature of each frame signal in the time domain;
Step (5) carries out Hilbert-Huang transform to each frame signal, obtains the Hilbert-Huang transform spectrogram of vibration signal,
Statistical nature of the vibration signal on time-frequency domain is extracted with statistical method by obtained spectrogram;
Step (6) merges the feature of the vibration signal that extracts in time domain and time-frequency domain, formed corresponding feature to
Amount judges inputted intrusion behavior class by the output of the machine learning algorithm as the input of machine learning algorithm
Type completes identification and alarm to intrusion behavior type.
2. circumference intrusion alarm method in airport according to claim 1, it is characterised in that: in the step (1), front end is adopted
Collect equipment and use nine axis vibration sensors, every circumference builds 3 vibrating sensors on the net, and every three throw the net is grouped for one, i.e.,
In each grouping, vibrating sensor is arranged in a manner of 3 × 3.
3. circumference intrusion alarm method in airport according to claim 1, it is characterised in that: in the step (2), artificial mould
Quasi- different intrusion behaviors include normal condition, state of blowing, climbing behavior, cut net behavior and saw net behavior.
4. circumference intrusion alarm method in airport according to claim 1, it is characterised in that: in the step (3), due to vibration
Dynamic signal is time-varying, and changing features are smaller in short time range, before handling data, sub-frame processing is carried out to it, by framing
Signal afterwards is handled as stable state, framing method are as follows:
Formula (1) framing is pressed for the vibration signal of a length of L:
In formula, lap of the overlap between adjacent two frame, overlap=wlen-inc, data will be divided into fnFrame, often
The position that one frame starts in data is
Startindex=(0:(nf-1)) * inc+1 (2).
5. circumference intrusion alarm method in airport according to claim 1, it is characterised in that: in the step (4), extraction
The feature of each frame in the time domain includes short-time energy, peak value in short-time average zero-crossing rate and frame.
6. circumference intrusion alarm method in airport according to claim 5, it is characterised in that: the short-time energy of signal is every frame
The quadratic sum of the amplitude of signal all the points;The short-time average zero-crossing rate of signal indicates that signal waveform passes through horizontal axis in a frame signal
Number;Amplitude is more than the number of the peak value of set threshold value in one frame signal of peak value expression in the frame of signal, for reflecting
The amplitude of signal and the information of frequency.
7. circumference intrusion alarm method in airport according to claim 1, it is characterised in that: in the step (5), to signal
Carrying out Hilbert-Huang transform includes two steps: empirical mode decomposition and Hilbert transform.
8. circumference intrusion alarm method in airport according to claim 7, it is characterised in that: the process of empirical mode decomposition is such as
Under:
1. finding out the minimum value sequence in each local maximum value sequence drawn game portion of original series;
2. carrying out interpolation to extreme value sequence with cubic spline function, the coenvelope sequence and lower envelope sequence of former sequence are respectively obtained
Column;
3. the coenvelope sequence and lower envelope sequence to per moment are averaged, equal value sequence is obtained;
4. former sequence subtracts equal value sequence, the details sequence for removing low frequency is obtained;
5. being saved if obtained details sequence meets two conditions of intrinsic mode function;Otherwise, this details sequence
Broomrape makees former sequence, repeat step 1.~4., by k iteration, until obtained new sequence meets determining for intrinsic mode function
Until two conditions of justice, first intrinsic mode function is obtained;
6. the intrinsic mode function subtracted with original series obtains remaining value sequence;
7. the remaining value sequence former sequence new as one is successively extracted the 2nd, the the 3rd ... ..., Zhi Dao according to above step
M intrinsic mode function.
9. circumference intrusion alarm method in airport according to claim 7, it is characterised in that: for arbitrary signal x (t),
Hilbert transform is defined as:
10. circumference intrusion alarm method in airport according to claim 1, it is characterised in that: uncommon for obtained every
Your Bert Huang spectrogram, therefrom calculate on different frequency the energy of signal and, as statistical nature of the signal on time-frequency domain.
11. circumference intrusion alarm method in airport according to claim 1, it is characterised in that: the time domain that will be extracted
It is combined with the feature on time-frequency domain, forms corresponding feature vector, as the input of machine learning algorithm, entered
Invade type identification.
12. circumference intrusion alarm method in airport according to claim 11, it is characterised in that: the machine learning algorithm
For BP neural network algorithm, including an input layer, three hidden layers and an output layer.
13. circumference intrusion alarm method in airport according to claim 12, it is characterised in that: the BP neural network
Input layer includes 30 neurons, respectively corresponds the feature in input feature value, and output layer includes 5 neurons, output pair
The recognition result of different intrusion behaviors.
14. according to circumference intrusion alarm method in airport described in claim 1, it is characterised in that: algorithm model is instructed by e-learning
It gets, field conduct does not need to reconfigure parameter.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910382594.XA CN110033581A (en) | 2019-05-09 | 2019-05-09 | Airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910382594.XA CN110033581A (en) | 2019-05-09 | 2019-05-09 | Airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110033581A true CN110033581A (en) | 2019-07-19 |
Family
ID=67241672
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910382594.XA Pending CN110033581A (en) | 2019-05-09 | 2019-05-09 | Airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110033581A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110570636A (en) * | 2019-09-12 | 2019-12-13 | 上海卓希智能科技有限公司 | Vibration alarm method based on nine-axis sensor and device and system thereof |
CN112687068A (en) * | 2021-03-19 | 2021-04-20 | 四川通信科研规划设计有限责任公司 | Intrusion detection method based on microwave and vibration sensor data |
CN112907869A (en) * | 2021-03-17 | 2021-06-04 | 四川通信科研规划设计有限责任公司 | Intrusion detection system based on multiple sensing technologies |
CN113781728A (en) * | 2021-08-02 | 2021-12-10 | 盐城市湛安智感科技有限公司 | Vibration sensing system and method based on group intelligent optimization |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101527698A (en) * | 2009-04-03 | 2009-09-09 | 北京理工大学 | Non-stationary interference suppression method based on Hilbert-Huang transformation and adaptive notch |
CN101988935A (en) * | 2009-12-18 | 2011-03-23 | 西安电子科技大学 | Digital down converter-Hilbert-Huang transform based instantaneous frequency measuring method |
GB2484196A (en) * | 2010-09-29 | 2012-04-04 | Secr Defence | Identifying sounds |
CN103617684A (en) * | 2013-12-12 | 2014-03-05 | 威海北洋电气集团股份有限公司 | Interference type optical fiber perimeter vibration intrusion recognition algorithm |
CN108961638A (en) * | 2018-05-23 | 2018-12-07 | 重庆科技学院 | Vibration optical fiber intrusion event detection method based on wavelet coefficient energy and algorithm |
-
2019
- 2019-05-09 CN CN201910382594.XA patent/CN110033581A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101527698A (en) * | 2009-04-03 | 2009-09-09 | 北京理工大学 | Non-stationary interference suppression method based on Hilbert-Huang transformation and adaptive notch |
CN101988935A (en) * | 2009-12-18 | 2011-03-23 | 西安电子科技大学 | Digital down converter-Hilbert-Huang transform based instantaneous frequency measuring method |
GB2484196A (en) * | 2010-09-29 | 2012-04-04 | Secr Defence | Identifying sounds |
CN103617684A (en) * | 2013-12-12 | 2014-03-05 | 威海北洋电气集团股份有限公司 | Interference type optical fiber perimeter vibration intrusion recognition algorithm |
CN108961638A (en) * | 2018-05-23 | 2018-12-07 | 重庆科技学院 | Vibration optical fiber intrusion event detection method based on wavelet coefficient energy and algorithm |
Non-Patent Citations (2)
Title |
---|
徐铖晋: "分布式光纤传感***的信号处理技术研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
赵亮: "HHT和CWT用于光纤振动信号分析的对比研究", 《激光技术》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110570636A (en) * | 2019-09-12 | 2019-12-13 | 上海卓希智能科技有限公司 | Vibration alarm method based on nine-axis sensor and device and system thereof |
CN112907869A (en) * | 2021-03-17 | 2021-06-04 | 四川通信科研规划设计有限责任公司 | Intrusion detection system based on multiple sensing technologies |
CN112687068A (en) * | 2021-03-19 | 2021-04-20 | 四川通信科研规划设计有限责任公司 | Intrusion detection method based on microwave and vibration sensor data |
CN113781728A (en) * | 2021-08-02 | 2021-12-10 | 盐城市湛安智感科技有限公司 | Vibration sensing system and method based on group intelligent optimization |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110033581A (en) | Airport circumference intrusion alarm method based on Hilbert-Huang transform and machine learning | |
CN105469523B (en) | The optical fiber perimeter means of defence of wind resistance rain interference | |
CN103968933B (en) | Fiber-optic vibration signal recognition method based on fuzzy field feature | |
WO2020138154A1 (en) | Method and system for specifying time-space, and for detecting and identifying vibration on basis of optical fiber signal feature | |
CN102045120B (en) | Vibration signal identification method for optical fiber perimeter system | |
CN106874833A (en) | A kind of mode identification method of vibration event | |
CN109616140B (en) | Abnormal sound analysis system | |
CN107527617A (en) | Monitoring method, apparatus and system based on voice recognition | |
CN103903008B (en) | A kind of method and system of the mist grade based on image recognition transmission line of electricity | |
CN105095624A (en) | Method for identifying optical fibre sensing vibration signal | |
CN104318503B (en) | A kind of system and method according to typhoon forecast rainfall | |
CN106127135B (en) | A kind of Ling Qu invasion vibration signal characteristics extract and classification and identification algorithm | |
CN111862538B (en) | Large wind early warning method and system for long-span arch bridge construction period | |
CN104900013A (en) | Mud-rock flow monitoring and alarming system based on infrasonic wave | |
CN106600869A (en) | Fence intrusion identification method for fiber fence security protection system | |
CN106382981B (en) | A kind of single station infrasound signal identification extracting method | |
CN106600870A (en) | Distributed optical fiber fence vibration invasion identifying system | |
CN105676287A (en) | Method for detecting nuclear explosion seismic event in specific area | |
CN114291081B (en) | Vehicle collision detection method based on artificial intelligence algorithm | |
CN103888725A (en) | Safe monitoring method and system | |
CN105575024B (en) | Jamproof fiber perimeter protection system and method | |
CN114742096A (en) | Intrusion alarm method and system based on vibration optical fiber detection and complete action extraction | |
CN105139869A (en) | Baby crying detection method based on interval difference features | |
KR20220146670A (en) | Traffic anomaly detection methods, devices, devices, storage media and programs | |
CN106651131A (en) | Power-transmission-line anti-typhoon early warning method and system thereof |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190719 |
|
RJ01 | Rejection of invention patent application after publication |