CN110335430A - Monitoring pipeline safety system, method and apparatus based on deep learning - Google Patents
Monitoring pipeline safety system, method and apparatus based on deep learning Download PDFInfo
- Publication number
- CN110335430A CN110335430A CN201910529579.3A CN201910529579A CN110335430A CN 110335430 A CN110335430 A CN 110335430A CN 201910529579 A CN201910529579 A CN 201910529579A CN 110335430 A CN110335430 A CN 110335430A
- Authority
- CN
- China
- Prior art keywords
- event
- signal
- deep learning
- module
- classification
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 35
- 238000013135 deep learning Methods 0.000 title claims abstract description 34
- 238000000034 method Methods 0.000 title claims description 50
- 238000012545 processing Methods 0.000 claims abstract description 29
- 239000013307 optical fiber Substances 0.000 claims abstract description 18
- 239000000523 sample Substances 0.000 claims abstract description 9
- 238000013528 artificial neural network Methods 0.000 claims abstract description 8
- 230000005540 biological transmission Effects 0.000 claims description 18
- 230000003287 optical effect Effects 0.000 claims description 18
- 238000001514 detection method Methods 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 12
- 238000010801 machine learning Methods 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 8
- 230000003993 interaction Effects 0.000 claims description 6
- 238000012417 linear regression Methods 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 2
- 230000015572 biosynthetic process Effects 0.000 abstract 1
- 238000004458 analytical method Methods 0.000 description 21
- 230000008569 process Effects 0.000 description 12
- 238000001228 spectrum Methods 0.000 description 11
- 239000000835 fiber Substances 0.000 description 9
- 238000010586 diagram Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 238000012549 training Methods 0.000 description 7
- 230000002547 anomalous effect Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 238000000354 decomposition reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000006378 damage Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 239000012530 fluid Substances 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000000739 chaotic effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012880 independent component analysis Methods 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000009527 percussion Methods 0.000 description 1
- 230000010287 polarization Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
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
- G06N3/08—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
- G08B13/124—Mechanical actuation by the breaking or disturbance of stretched cords or wires for a perimeter fence with the breaking or disturbance being optically detected, e.g. optical fibers in the perimeter fence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Alarm Systems (AREA)
Abstract
The invention belongs to Monitoring Pinpelines fields, and in particular to a kind of monitoring pipeline safety system based on deep learning, it is intended in order to solve existing interference-type optical fiber early warning system to event fail to report and rate of false alarm is high, can not steadily identify intrusion event classification problem.The present invention includes optical fiber probe module, for obtaining detectable signal and reference signal；Signal processing module, the time domain of the interference signal for obtaining detectable signal and reference signal formation, frequency domain statistical nature；Event category module is carried out the judgement of affair character classification for the event category model by being constructed based on deep neural network, obtains the classification of event.The present invention improves the accuracy of identification of event, reduces rate of failing to report and rate of false alarm, timely and accurately finds intrusion event.
Description
Technical field
The invention belongs to Monitoring Pinpelines fields, and in particular to a kind of monitoring pipeline safety system based on deep learning, side
Method and device.
Background technique
In recent years, the oil and gas pipelines that the domestic Gas Company in China is administered again and again by third-party destruction, thus
Serious economic loss and problem of environmental pollution are caused to state and society, and easily thus cause casualties and fire
The a series of disastrous effect such as calamity, explosion, influences economic security of the country and social stability.As oil-gas pipeline is laid with distance year by year
Increase, traditional manual inspection method has been unable to meet current monitoring requirements.And the leakage inspection propagated based on overpressure wave
Survey method can only be destroyed or be revealed and alarm after having occurred and that, prevention and protection can not be fundamentally played the role of.
Different from traditional manual inspection, mainstream detection method used by long distance oil-gas pipeline is predispersed fiber at present
It is alert, wherein the optical fiber early warning system of two kinds of principles of interference-type and scatter-type can be divided into again.Wherein, because interference-type early warning system has
The features such as transmission range is long, decays small, and positioning accuracy is high, high sensitivity, is particularly suitable for the early warning system as long-distance pipe
System.
Traditional interference-type optical fiber early warning system is classified using simple statistical nature, based on event effect of vibration in
Optical fiber and the optical signal for causing phase change carry out cross-correlation calculation in the time domain and are positioned, special in the enterprising line frequency domain of frequency domain
Sign statistics identification events classification.But in reality, the signal characteristic of intrusion event has nonlinearity, existing interference-type light
Fine early warning system to event fail to report and rate of false alarm is high, can not steadily identify intrusion event classification, therefore there is an urgent need to optimize
The precision of recognizer improves the practicability of system.
Summary of the invention
In order to solve the above problem in the prior art, in order to solve existing interference-type optical fiber early warning system to event
Fail to report with rate of false alarm height, the problem of can not steadily identifying intrusion event classification, the invention proposes one kind to be based on depth
The monitoring pipeline safety system of habit, the system include optical fiber probe module, signal processing module, event category module；
The optical fiber probe module is configured to the laser signal through remote optical cable transmission, obtains and phase tune occurs
The detectable signal of system and not phase modulated reference signal；
The signal processing module is configured to the interference signal of the detectable signal Yu the reference signal, passes through
Feature extraction obtains the Time-domain Statistics feature of the interference signal, frequency domain statistical nature as characteristic information；
The event category module is configured to the characteristic information that the signal processing module obtains, passes through event point
The classification of class model acquisition event；The event category model is constructed based on deep neural network, for the defeated of input data
Enter feature progress classification to judge to obtain the classification of event.
In some preferred embodiments, the monitoring pipeline safety system based on deep learning further includes state event location mould
Block, the state event location module are configured to the detectable signal of different transmission directions, by the detectable signal transmit when
Prolong the transmission range that difference calculates detectable signal, obtains event position information.
In some preferred embodiments, the monitoring pipeline safety system based on deep learning further includes human-computer interaction mould
Block, the human-computer interaction module is configured as output to the classification for the event that the event category module obtains and/or the event is determined
The event position information of position module output.
In some preferred embodiments, the event category model is by Adaboost method by multiple weak typings
Device combines the strong classifier to be formed；The multiple Weak Classifier is based respectively on deep learning algorithm, machine learning algorithm, non-machine
Learning algorithm is built.
In some preferred embodiments, the optical cable laying in pipeline vertical direction or with pipeline same level or
Pipeline oblique upper.
The second aspect of the present invention proposes a kind of monitoring pipeline safety method based on deep learning, this method include with
Lower step:
Based on the laser signal through remote optical cable transmission, the detectable signal that phase-modulation occurs is obtained and without phase tune
The reference signal of system；
Interference signal based on the detectable signal Yu the reference signal obtains the interference signal by feature extraction
Time-domain Statistics feature, frequency domain statistical nature is as characteristic information；
Based on the characteristic information, the classification of event is obtained by event category model；The event category model is based on
Deep neural network building carries out classification for the input feature vector to input data and judges to obtain the classification of event.
In some preferred embodiments, the event category model is by Adaboost method by multiple weak typings
Device combines the strong classifier to be formed；The multiple Weak Classifier is based respectively on deep learning algorithm, machine learning algorithm, non-machine
Learning algorithm is built.
In some preferred embodiments, the machine learning algorithm is support vector machines, linear regression, hidden Ma Erke
One of husband's model or many algorithms.
The third aspect of the present invention proposes a kind of storage device, wherein be stored with a plurality of program, described program be suitable for by
Processor is loaded and is executed to realize the above-mentioned monitoring pipeline safety method based on deep learning.
The fourth aspect of the present invention, including processor, storage device；Processor is adapted for carrying out each program；Storage dress
It sets, is suitable for storing a plurality of program；Described program is suitable for being loaded by processor and being executed above-mentioned based on deep learning to realize
Monitoring pipeline safety method.
Beneficial effects of the present invention: the present invention is using the interference of light as testing principle, based on deep neural network building
Event category model carries out the judgement of affair character classification, improves the accuracy of identification of event, reduces rate of failing to report and rate of false alarm,
Timely and accurately find intrusion event.
Detailed description of the invention
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, the application's is other
Feature, objects and advantages will become more apparent upon:
Monitoring pipeline safety service system block schematic illustration of the Fig. 1 based on deep learning；
Monitoring pipeline safety system light path schematic illustration of the Fig. 2 based on deep learning；
Monitoring pipeline safety system signal processing flow schematic diagram of the Fig. 3 based on deep learning；
Fig. 4 Digital Signal Processing terminal affair identification process schematic diagram；
Fig. 5 signal is with range attenuation amplitude-versus-frequency curve figure；
Fig. 6 background signal analysis chart；
Fig. 7 hand digging analysis of vibration signal figure；
Fig. 8 pipeline drilling analysis of vibration signal figure；
Fig. 9 saws pipe vibration signal analysis chart；
Figure 10 pipeline percussion vibration signal analysis chart.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached drawing to the embodiment of the present invention
In technical solution be clearly and completely described, it is clear that described embodiments are some of the embodiments of the present invention, without
It is whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not before making creative work
Every other embodiment obtained is put, shall fall within the protection scope of the present invention.
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to
Convenient for description, part relevant to related invention is illustrated only in attached drawing.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.
The present invention provides a kind of monitoring pipeline safety system 100 based on deep learning, as shown in Figure 1, the system includes
Optical fiber probe module 110, signal processing module 120, event category module 130.
Optical fiber probe module 110 is configured to the laser signal through remote optical cable transmission, obtains and phase-modulation occurs
Detectable signal and not phase modulated reference signal.
Signal processing module 120 is configured to the interference signal of detectable signal and reference signal, is obtained by feature extraction
The Time-domain Statistics feature that takes interference signal, frequency domain statistical nature are as characteristic information.
Event category module 130 is configured to the characteristic information of the acquisition of signal processing module 120, passes through event category
The classification of model acquisition event；Event category model is constructed based on deep neural network, for the input feature vector to input data
Classification is carried out to judge to obtain the classification of event.
In order to which more clearly the present invention will be described, below with reference to Fig. 1 to respectively being walked in a kind of embodiment of present system
Suddenly expansion detailed description is carried out.
1, optical fiber probe module 110
The present invention is using the interference of light as testing principle, using the variation of fiber middle light signal as detection means, therefore this
The optical transmission chain of optical fiber probe module is built in embodiment according to full fiber type Mach Zehnder interferometer principle.
As shown in Fig. 2, laser generator connects driving circuit, launch wavelength is in the light letter of sinusoidal variation at any time
Number, isolator is added in the one-way transmission for guaranteeing light in transmitting laser process in systems, eliminates polarization damage by scrambler
Evil, the signal peak for exporting laser generator keep stablizing, and coupler realizes the coupling of light.Laser signal passes through sensing respectively
Optical fiber and reference optical fiber transmission.When there is third party signalling, phase change occurs for the optical signal inside sensor fibre, with reference
Not phase modulated reference signal interferes in optical fiber, and interference signal is transmitted at signal processing module 120, passes through coupling
Device couples interference signal to the detector of signal processing module 120.
The optical cable for being used for transmission optical signal can use three kinds of different paving modes: optical cable laying is on pipeline is vertical
Side and pipeline same level or pipeline oblique upper.Under three kinds of paving modes, the present embodiment system can be worked normally.
Distributed Detection can be used in optical signal, and one-stop detection (i.e. Laser emission transposition and reception device position also can be used
At one).
When the length of some pipelines to be monitored is too long, can using Distributed Detection will test range be divided into uniformly etc.
Divide distance, and dispose detector and processor in each section, carries out data real-time storage, processing and transmission.
Only signal processing module 120 and event category module 130 are introduced at wherein one below, it is same suitable
Other websites for one-stop detection and Distributed Detection are arranged.
2, signal processing module 120
Signal processing module 120 includes photodetector, amplifier, filter, digital demodulator, Digital Signal Processing end
End, as shown in figure 3, photodetector converts optical signal into analog electrical signal, amplifier amplifies analog electrical signal, filter
For wave device by carrying out data cleansing to signal de-noising, filtering, the analog electrical signal after cleaning is converted number by digital demodulator
Signal.
Digital Signal Processing terminal extracts the Time-domain Statistics feature of interference signal, frequency domain statistical nature, construction algorithm layer
The input form of data set, data mode and algorithm layer defined in the data set matches.
3, event category module 130
Because deep learning algorithm has the ability in fitting nonlinearity space, and believe caused by third party's intrusion event
Number itself just there is nonlinearity and randomness, therefore event category model takes deep learning algorithm, to input data
Input feature vector carries out classification and judges to obtain the classification of event.The spy that Digital Signal Processing terminal is obtained based on signal processing module
Reference breath is carried out by data of the event category model to the data set comprising signal Time-domain Statistics feature, frequency domain statistical nature
Analysis and event classification.
Process flow of the Digital Signal Processing terminal after the generation of third party's intrusion event is as shown in figure 4, obtain third party
Analog signal caused by intrusion event amplifies, filters and analog-to-digital conversion, signal characteristic extracted later, to signal characteristic
Classification is carried out to judge finally to export result to obtain the classification of event.
4, state event location module
The signal of treated optical path 1 and optical path 2 is carried out cross-correlation training, signal processing stream by Digital Signal Processing terminal
Journey as shown in figure 3, obtaining the minimum period of change in signal strength, believe by the delay inequality calculating detection transmitted by two-way detectable signal
Number transmission range, precise positioning venue location point in the case where known fiber optic refractive index and launch time.
5, human-computer interaction module
Human-computer interaction service is provided on integrated platform, to carry out affair alarm, state event location information is shown.Event
Alarm: when having the generation of third party's intrusion event, system is reminded by modes such as text, sound, alarm lamps；State event location letter
Breath is shown: when there is intrusion event generation, the invasion more specific location information calculated is passed through image, sound or text etc. by system
One or more combined form outgoing event location informations.
In order to be preferably illustrated to the monitoring pipeline safety system embodiment the present invention is based on deep learning, to this reality
Example is applied to be described from the extraction of feature, building for event category model with two parts of training.
(1) extraction of feature
Pipeline is during fluid conveying, and due to the effect of fluid and tube wall, generation is unfavorable for analysis of vibration signal
Ambient noise.The ambient noise will study pipeline typical case accident detection identification system by sensor fibre inductive pick-up, therefore
The design of system, the analysis of ambient noise are the problem of must be taken into consideration before system designs.In addition, causing for typicalness anomalous event
Vibration signal its mechanism of production, route of transmission in the duct, attenuation characteristic be also asking of must be taken into consideration before system design
Topic.The vibration signal characteristics generated by understanding various typical anomalous events, study suitable signal characteristic extracting methods, make allusion quotation
The purpose of design of type anomalous event monitoring pipeline safety system is clear.Below to the analysis of pipe vibration signal and detection identification side
Method explains.
In one embodiment, pipeline abnormal transient vibration signal mechanism of production and propagation characteristic include: anomalous event vibration letter
Number frequency band is wider, and frequency is higher, and signal decaying is bigger.The amplitude and frequency content of signal are because between vibration source and sensor fibre
Distance it is different, vary widely.Signal major frequency components are equal are as follows: 0~1000Hz.It can be said that clear signal high frequency at
Deep fades, higher frequency signal energy are stronger than low-frequency component point with the increase of distance, and signal below for 0~1000Hz
Decay weaker, as shown in figure 5, the amplitude-versus-frequency curve figure that signal is decayed with distance.
Figure a), figure c), figure e), figure g) abscissa be distance, ordinate is amplitude, is the spacing of vibration source Yu sensor fibre
The waveform diagram of original signal when from respectively 2m, 9m, 4m, 7m；
Figure b), figure d), figure f), figure h) abscissa be frequency, it is between vibration source and sensor fibre that ordinate, which is amplitude,
The amplitude spectrum of original signal when apart from respectively 2m, 9m, 4m, 7m.
Pipeline is mainly 1kHz low frequency signal below by the useful vibration signal generated is destroyed, and the amplitude of signal exists
There are the variations of 11mv~110mv within the scope of 50~1000m, and as the increase of propagation distance exponentially decays.
Anomalous event vibration signal typical in one embodiment is analyzed below.
Background signal analysis, is illustrated in figure 6 an exemplary analysis result:
I) abscissa is distance to figure, and ordinate is amplitude, is background signal waveform diagram；
J) abscissa is frequency to figure, and ordinate is amplitude, is the amplitude spectrum of background signal；
K) abscissa is frequency to figure, and ordinate is amplitude, is the auto-power spectrum figure of background signal.
The analysis of hand digging signal, is illustrated in figure 7 an exemplary analysis result:
Figure is l), o) abscissa is distance to figure, and ordinate is amplitude, is hand digging signal 1 and hand digging signal 2 respectively
Waveform diagram；
Figure is m), p) abscissa is frequency to figure, and ordinate is amplitude, is hand digging signal 1 and hand digging signal 2 respectively
Amplitude spectrum；
Figure is n), q) abscissa is frequency to figure, and ordinate is amplitude, is hand digging signal 1 and hand digging signal 2 respectively
Auto-power spectrum figure.
The analysis of pipeline borehole signal, is illustrated in figure 8 an exemplary analysis result:
Figure is r), u) abscissa is distance to figure, and ordinate is amplitude, is pipeline borehole signal 1 and pipeline borehole signal 2 respectively
Waveform diagram；
Figure is s), v) abscissa is frequency to figure, and ordinate is amplitude, is pipeline borehole signal 1 and pipeline borehole signal 2 respectively
Amplitude spectrum；
Figure is t), w) abscissa is frequency to figure, and ordinate is amplitude, is pipeline borehole signal 1 and pipeline borehole signal 2 respectively
Auto-power spectrum figure.
Pipe signal analysis is sawed, an exemplary analysis result is illustrated in figure 9:
Figure x), figure x ') abscissa be distance, ordinate is amplitude, be respectively saw pipe signal 1 and saw pipe signal 2
Waveform diagram；
Figure y), figure y ') abscissa be frequency, ordinate is amplitude, be respectively saw pipe signal 1 and saw pipe signal 2
Amplitude spectrum；
Figure z), figure z ') abscissa be frequency, ordinate is amplitude, be respectively saw pipe signal 1 and saw pipe signal 2
Auto-power spectrum figure.
The analysis of pipeline knocking, it is an exemplary analysis result that the results are shown in Figure 10:
Scheme a1), figure d1) abscissa is distance, ordinate is amplitude, is that pipeline knocking 1 and pipeline tap letter respectively
Numbers 2 waveform diagram；
Scheme b1), figure e1) abscissa is frequency, ordinate is amplitude, is that pipeline knocking 1 and pipeline tap letter respectively
Numbers 2 amplitude spectrum；
Scheme c1), figure f1) abscissa is frequency, ordinate is amplitude, is that pipeline knocking 1 and pipeline tap letter respectively
Numbers 2 auto-power spectrum figure.
With continued reference to Figure 10, pipeline knocking signal between 0~40Hz is most strong, and is distributed between 40~1000Hz
It is more average.
The method of feature extraction mainly has: based on wavelet and wavelet package decompose signal energy characteristic vector pickup, be based on
The signal characteristic abstraction of empirical mode decomposition, the signal characteristic abstraction based on multi-scale chaotic characteristic analysis of detection, independent component analysis.
Found by calculating, the Wavelet Entropy of signal can extract the characteristic feature of signal, and directly to time-domain signal into
Row seeks entropy, and for the vibration signal of different event, Energy-Entropy maximum value distributed area is different, has certain regularity.
Based on wavelet and wavelet package decompose signal energy characteristic vector pickup, specially following four kinds of methods:
" energy-mode " signal characteristic extracting methods based on wavelet decomposition；
" energy-mode " signal characteristic extracting methods based on WAVELET PACKET DECOMPOSITION；
Signal characteristic extracting methods based on Wavelet Entropy；
Signal characteristic extracting methods based on Energy-Entropy.
Its effective frequency ingredient of pipeline typical case's anomalous event vibration signal be mainly distributed on 1000Hz hereinafter, signal analysis with
In identification process most important algorithm routine be have limit for length's unit impulse response filter and an Energy-Entropy seek program.
(2) model buildings and training
Event category model is constructed based on deep neural network, may be based on regular machinery learning algorithm (such as supporting vector
Machine, linear regression, hidden Markov model) and non-machine learning algorithm (such as cross-correlation coefficient), with Adaboost by several sides
Method is combined, training strong classifier.
Data set of the signal characteristic of signal processing module output as deep neural network model, feature are optical fiber early warning
The amplitude of system, phase, frequecy characteristic and auto-correlation coefficient, label are that hand digging, machine excavation, vehicle process etc. cause to shake
Dynamic third party's event.
By obtained characteristic parameter carry out deep learning training, training when using forward-propagating, obtain single iteration result,
Loss function is calculated, in conjunction with loss function Optimized model parameter, is repeated the above process, when cost loss function is reduced to ideal journey
When spending and training the maximum number of iterations for reaching required, event category model is generated.
It should be noted that the monitoring pipeline safety system provided by the above embodiment based on deep learning, only with above-mentioned
The division of each functional module carries out for example, in practical applications, can according to need and by above-mentioned function distribution by difference
Functional module complete, i.e., by the embodiment of the present invention module or step decompose or combine again, for example, above-mentioned implementation
The module of example can be merged into a module, multiple submodule can also be further split into, to complete whole described above
Or partial function.For module involved in the embodiment of the present invention, the title of step, it is only for distinguish modules or
Person's step, is not intended as inappropriate limitation of the present invention.
A kind of monitoring pipeline safety method based on deep learning of the embodiment of the present invention, comprising the following steps:
Based on the laser signal through remote optical cable transmission, the detectable signal that phase-modulation occurs is obtained and without phase tune
The reference signal of system；
Interference signal based on detectable signal and reference signal, the Time-domain Statistics for obtaining interference signal by feature extraction are special
Sign, frequency domain statistical nature are as characteristic information；
Based on characteristic information, the classification of event is obtained by event category model；Event category model is based on depth nerve
Network struction carries out classification for the input feature vector to input data and judges to obtain event category.
Person of ordinary skill in the field can be understood that, for convenience and simplicity of description, foregoing description
The specific work process of method and related explanation, can be with reference to the corresponding process in aforementioned system embodiment, and details are not described herein.
A kind of storage device of the embodiment of the present invention, wherein being stored with a plurality of program, described program is suitable for being added by processor
It carries and executes to realize the above-mentioned monitoring pipeline safety method based on deep learning.
A kind of processing unit of the embodiment of the present invention, including processor, storage device；Processor is adapted for carrying out each journey
Sequence；Storage device is suitable for storing a plurality of program；Described program is suitable for being loaded by processor and being executed above-mentioned based on depth to realize
Spend the monitoring pipeline safety method of study.
Person of ordinary skill in the field can be understood that, for convenience and simplicity of description, foregoing description
The specific work process and related explanation of storage device, processing unit can refer to the corresponding process in aforementioned system embodiment,
Details are not described herein.
Those skilled in the art should be able to recognize that, mould described in conjunction with the examples disclosed in the embodiments of the present disclosure
Block, method and step, can be realized with electronic hardware, computer software, or a combination of the two, software module, method and step pair
The program answered can be placed in random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electric erasable and can compile
Any other form of storage well known in journey ROM, register, hard disk, moveable magnetic disc, CD-ROM or technical field is situated between
In matter.In order to clearly demonstrate the interchangeability of electronic hardware and software, in the above description according to function generally
Describe each exemplary composition and step.These functions are executed actually with electronic hardware or software mode, depend on technology
The specific application and design constraint of scheme.Those skilled in the art can carry out using distinct methods each specific application
Realize described function, but such implementation should not be considered as beyond the scope of the present invention.
Term " includes " or any other like term are intended to cover non-exclusive inclusion, so that including a system
The process, method or equipment/device of column element not only include those elements, but also other are wanted including what is be not explicitly listed
Element either further includes these process, methods or the intrinsic element of equipment/device.
So far, it has been combined preferred embodiment shown in the drawings and describes technical solution of the present invention, still, this field
Technical staff is it is easily understood that protection scope of the present invention is expressly not limited to these specific embodiments.Without departing from this
Under the premise of the principle of invention, those skilled in the art can make equivalent change or replacement to the relevant technologies feature, these
Technical solution after change or replacement will fall within the scope of protection of the present invention.
Claims (10)
1. a kind of monitoring pipeline safety system based on deep learning, which is characterized in that the system includes optical fiber probe module, letter
Number processing module, event category module；
The optical fiber probe module is configured to the laser signal through remote optical cable transmission, obtains and phase-modulation occurs
Detectable signal and not phase modulated reference signal；
The signal processing module is configured to the interference signal of the detectable signal Yu the reference signal, passes through feature
The Time-domain Statistics feature for obtaining the interference signal, frequency domain statistical nature are extracted as characteristic information；
The event category module is configured to the characteristic information that the signal processing module obtains, passes through event category mould
The classification of type acquisition event；The event category model is constructed based on deep neural network, special for the input to input data
Sign carries out classification and judges to obtain the classification of event.
2. the monitoring pipeline safety system according to claim 1 based on deep learning, which is characterized in that the pipeline peace
Full monitoring system further includes state event location module；The state event location module is configured to the detection letter of different transmission directions
Number, the transmission range of detectable signal is calculated by the delay inequality that the detectable signal transmits, and obtains event position information.
3. the monitoring pipeline safety system according to claim 2 based on deep learning, which is characterized in that the pipeline peace
Full monitoring system further includes human-computer interaction module, and the human-computer interaction module is configured as output to what the event category module obtained
The event position information of the classification of event and/or state event location module output.
4. the monitoring pipeline safety system according to claim 1 based on deep learning, which is characterized in that the event point
Class model is multiple Weak Classifiers to combine the strong classifier to be formed by Adaboost method；The multiple Weak Classifier difference
It is built based on deep learning algorithm, machine learning algorithm, non-machine learning algorithm.
5. the monitoring pipeline safety system according to claim 1 based on deep learning, which is characterized in that the optical cable paving
Set on pipeline vertical direction or with pipeline same level or pipeline oblique upper.
6. a kind of monitoring pipeline safety method based on deep learning, which is characterized in that method includes the following steps:
Based on the laser signal through remote optical cable transmission, the detectable signal that phase-modulation occurs and not phase modulated is obtained
Reference signal；
Interference signal based on the detectable signal Yu the reference signal, by feature extraction obtain the interference signal when
Domain statistical nature, frequency domain statistical nature are as characteristic information；
Based on the characteristic information, the classification of event is obtained by event category model；The event category model is based on depth
Neural network building carries out classification for the input feature vector to input data and judges to obtain the classification of event.
7. the monitoring pipeline safety method according to claim 7 based on deep learning, which is characterized in that the event point
Class model is multiple Weak Classifiers to combine the strong classifier to be formed by Adaboost method；The multiple Weak Classifier difference
It is built based on deep learning algorithm, machine learning algorithm, non-machine learning algorithm.
8. the monitoring pipeline safety method according to claim 8 based on deep learning, which is characterized in that the engineering
Habit algorithm is one of support vector machines, linear regression, hidden Markov model or many algorithms.
9. a kind of storage device, wherein being stored with a plurality of program, which is characterized in that described program is suitable for being loaded and being held by processor
Row is to realize the monitoring pipeline safety method in claim 6-8 described in any claim based on deep learning.
10. a kind of processing unit, including processor, storage device；Processor is adapted for carrying out each program；Storage device is suitable for
Store a plurality of program；It is characterized in that, described program is any in claim 6-8 to realize suitable for being loaded by processor and being executed
Monitoring pipeline safety method described in claim based on deep learning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910529579.3A CN110335430A (en) | 2019-06-19 | 2019-06-19 | Monitoring pipeline safety system, method and apparatus based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910529579.3A CN110335430A (en) | 2019-06-19 | 2019-06-19 | Monitoring pipeline safety system, method and apparatus based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110335430A true CN110335430A (en) | 2019-10-15 |
Family
ID=68142810
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910529579.3A Pending CN110335430A (en) | 2019-06-19 | 2019-06-19 | Monitoring pipeline safety system, method and apparatus based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110335430A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110853268A (en) * | 2019-11-01 | 2020-02-28 | 江苏安防科技有限公司 | Method for realizing intrusion detection vulnerability real-time scanning based on analysis of mass data of pipe gallery system |
| CN111554061A (en) * | 2020-05-08 | 2020-08-18 | 武汉理工大学 | Intelligent perimeter security system fusing distributed optical fiber sensor and camera |
| CN112780951A (en) * | 2019-11-07 | 2021-05-11 | 中国石油化工股份有限公司 | Method, device and system for detecting storage tank and pipeline invasion event |
| CN113780204A (en) * | 2021-09-10 | 2021-12-10 | 西南石油大学 | Pipeline excavation vibration signal identification method based on convolutional neural network |
| CN113988170A (en) * | 2021-10-26 | 2022-01-28 | 国家石油天然气管网集团有限公司 | Event classification machine learning algorithm of optical fiber early warning system |
| CN117292706A (en) * | 2023-11-24 | 2023-12-26 | 天津声格智能科技有限公司 | Pipeline gas leakage sound diagnosis method, system, equipment and medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101566586A (en) * | 2008-04-22 | 2009-10-28 | 宁波诺可电子科技发展有限公司 | External safety early-warning and positioning system of optical cable and optical-electrical compound cable |
| US20100290734A1 (en) * | 2009-05-14 | 2010-11-18 | Celight, Inc. | Optical sensor for detecting and localizing events |
| CN104269006A (en) * | 2014-09-24 | 2015-01-07 | 天津大学 | Optical fiber early warning system and mode identification method |
| CN104751580A (en) * | 2015-04-08 | 2015-07-01 | 武汉理工光科股份有限公司 | Distributed optical fiber sensing signal mode identifying method and system |
| CN105374148A (en) * | 2015-12-11 | 2016-03-02 | 河南辉煌科技股份有限公司 | Optical fiber vibration-based railway perimeter protection method |
-
2019
- 2019-06-19 CN CN201910529579.3A patent/CN110335430A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101566586A (en) * | 2008-04-22 | 2009-10-28 | 宁波诺可电子科技发展有限公司 | External safety early-warning and positioning system of optical cable and optical-electrical compound cable |
| US20100290734A1 (en) * | 2009-05-14 | 2010-11-18 | Celight, Inc. | Optical sensor for detecting and localizing events |
| CN104269006A (en) * | 2014-09-24 | 2015-01-07 | 天津大学 | Optical fiber early warning system and mode identification method |
| CN104751580A (en) * | 2015-04-08 | 2015-07-01 | 武汉理工光科股份有限公司 | Distributed optical fiber sensing signal mode identifying method and system |
| CN105374148A (en) * | 2015-12-11 | 2016-03-02 | 河南辉煌科技股份有限公司 | Optical fiber vibration-based railway perimeter protection method |
Non-Patent Citations (7)
| Title |
|---|
| 孙茜等: "相关向量机在光纤预警系统模式识别中的应用", 《天津大学学报（自然科学与工程技术版）》 * |
| 张宇等: "基于动态压力信号的管道泄漏特征提取方法研究", 《石油学报》 * |
| 张景川等: "基于独立分量分析的管道异常振动时间定位方法", 《振动与冲击》 * |
| 曲志刚等: "基于经验模态分解的管道预警系统信号多尺度混沌特征分析方法", 《石油学报》 * |
| 李健等: "埋地管道典型异常事件信号特征提取方法研究", 《传感技术学报》 * |
| 葛道明: "分布式光纤油气长输管道泄漏检测及预警技术研究", 《硅谷》 * |
| 郎宪明等: "长输油气管道泄漏检测与定位技术研究进展", 《控制工程》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110853268A (en) * | 2019-11-01 | 2020-02-28 | 江苏安防科技有限公司 | Method for realizing intrusion detection vulnerability real-time scanning based on analysis of mass data of pipe gallery system |
| CN112780951A (en) * | 2019-11-07 | 2021-05-11 | 中国石油化工股份有限公司 | Method, device and system for detecting storage tank and pipeline invasion event |
| CN111554061A (en) * | 2020-05-08 | 2020-08-18 | 武汉理工大学 | Intelligent perimeter security system fusing distributed optical fiber sensor and camera |
| CN113780204A (en) * | 2021-09-10 | 2021-12-10 | 西南石油大学 | Pipeline excavation vibration signal identification method based on convolutional neural network |
| CN113988170A (en) * | 2021-10-26 | 2022-01-28 | 国家石油天然气管网集团有限公司 | Event classification machine learning algorithm of optical fiber early warning system |
| CN117292706A (en) * | 2023-11-24 | 2023-12-26 | 天津声格智能科技有限公司 | Pipeline gas leakage sound diagnosis method, system, equipment and medium |
| CN117292706B (en) * | 2023-11-24 | 2024-01-26 | 天津声格智能科技有限公司 | Pipeline gas leakage sound diagnosis method, system, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110335430A (en) | Monitoring pipeline safety system, method and apparatus based on deep learning | |
| US10620038B2 (en) | Distributed optical fiber sensing signal processing method for safety monitoring of underground pipe network | |
| Bai et al. | Detection and identification of external intrusion signals from 33 km optical fiber sensing system based on deep learning | |
| CN101183899B (en) | BP network based pipeline security identifying method for optical fiber pipeline leakage monitoring device | |
| CN104574742B (en) | Optical fiber perimeter security system based on phi-OTDR technology | |
| Fouda et al. | Pattern recognition of optical fiber vibration signal of the submarine cable for its safety | |
| CN201191221Y (en) | Disturbance signal determination module of distributed optical fiber vibration sensor system | |
| CN106225907A (en) | A kind of fiber-optic vibration identification system and method based on Φ OTDR technique | |
| Zhu et al. | Vibration pattern recognition and classification in OTDR based distributed optical-fiber vibration sensing system | |
| CN109374115A (en) | Broken vibration monitoring recognition methods outside oil-gas pipeline based on φ-OTDR | |
| CN103196465B (en) | Method for separating noise of sensing signal and extracting signal of phase-sensitive optical time domain reflectometer | |
| WO2016198683A1 (en) | Method and apparatus for monitoring pipeline using an optical fiber sensor system | |
| CN103824406A (en) | Optical fiber sensing-based tunnel cable invasive detection method and system | |
| CN206439635U (en) | A kind of Pipeline Leak monitoring system | |
| US12000729B2 (en) | Perpendicular distance prediction of vibrations by distributed fiber optic sensing | |
| Marie et al. | Pattern recognition algorithm and software design of an optical fiber vibration signal based on Φ-optical time-domain reflectometry | |
| CN101299635B (en) | Apparatus and method for discriminating front and back of wide field complete-optical fiber disturbing signal | |
| CN114613116B (en) | External damage prevention early warning method, device, equipment and storage medium | |
| CN105067101A (en) | Fundamental tone frequency characteristic extraction method based on vibration signal for vibration source identification | |
| CN109120336A (en) | Anti- false alarm false dismissal method based on phase sensitive optical time domain reflection sensor | |
| CN203799460U (en) | Tunnel cable intrusion detection system based on optical fiber sensing | |
| CN104916059B (en) | M Z interferometer types airport circumference early warning system mode identification method based on EEMD | |
| Wang et al. | Cumulative absolute velocity prediction for earthquake early warning with deep learning | |
| Wang et al. | RVFL-based optical fiber intrusion signal recognition with multi-level wavelet decomposition as feature | |
| Qiu et al. | A new detection method based on CFAR and DE for OFPS |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191015 |