CN201173390Y - Limit type signal processing and recognition module for distributed optical fibre disturbance sensing system - Google Patents
Limit type signal processing and recognition module for distributed optical fibre disturbance sensing system Download PDFInfo
- Publication number
- CN201173390Y CN201173390Y CNU2008200337617U CN200820033761U CN201173390Y CN 201173390 Y CN201173390 Y CN 201173390Y CN U2008200337617 U CNU2008200337617 U CN U2008200337617U CN 200820033761 U CN200820033761 U CN 200820033761U CN 201173390 Y CN201173390 Y CN 201173390Y
- Authority
- CN
- China
- Prior art keywords
- signal
- module
- storage
- identified
- output terminal
- 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.)
- Expired - Fee Related
Links
Images
Abstract
A limited signal processing module for a distributed optical fiber disturbance sensing system is mainly applied to monitoring pipelines which are used for conveying dangerous or high-valued chemicals. A data signal buffer (301) arranged at the frontal end of the pipeline is serially connected with a data reader (302), a difference spectrum noise-reducing module 303), a small wave noise-reducing module (304), an ICA signal separation module (305) and a feature extraction module (306) in series. Both output terminals of the feature extraction module are respectively connected with the input terminal of a memory (307) for signal feature value to be identified and the input terminal of a QDA recognizer (308); the output terminal of the QDA recognizer is respectively connected with the input terminal of a class feature database (309) and the input terminal of a memory (310) used for identifying results; the output terminal of a memory (313) for signals to be identified is sequentially connected with an audio player (312), an artificial detectophone (313), and the memory for signal feature value to be identified in series; the output terminal of the memory for signal feature value to be identified is connected with the input terminal of the class feature database.
Description
Technical field
The distribution type fiber-optic disturbance sensing network system is mainly used in the pipe monitoring of all kinds of conveying danger or high value chemicals pipe safety monitoring etc., belongs to Fibre Optical Sensor detection technique field.
Background technique
Along with science and technology and social development, the competition between each countries and regions becomes more and more fierce, and meanwhile, various dangerous terrorist activities are also more and more fiery.Therefore, for some critical facilitys or zone, as military restricted zone, high-risk forbidden zone and airdrome, nuclear power station or the like needs a kind of safety monitoring system that meets modern society's needs to guarantee being perfectly safe of these facilities or zone.This system must be efficient, real-time, accurate, intelligent.At present; the technology that is used for the safety check of building circumference has infrared, microwave and traditional technology such as iron wire electrical network, but they all have fatal separately shortcoming; and be difficult to signal is carried out accurately complicated processing, therefore the efficient that circumference safety protection, prevention are trespassed is very low.Show according to investigations, demand to the circumference safe examination system of a high efficiency smart in the society is extremely strong and huge, and that the whole world has only is external for the few companies of number can produce the circumference safe examination system that meets social demand, and therefore, the present invention has extremely wide application market.The present invention not only is applicable to critical facility and zone, also is applicable to the transport vehicle of crucial things such as oil and gas pipeline.Application area is such as having: airdrome, nuclear power station, oil transport pipeline, communication optical cable, means of transportation, historic reservation, armory, critical facilitys and zone such as emphasis office and essential industry plant area.
As the broad domain all-optical fiber disturbance sensing network system front end system, full fiber reflection interferes the distributed sensing circuitry will wrap up in after the mixed signal of taking each noise like and disturbance information receives, store in the data-signal buffer storage, can carry out real-time and effective processing and identification this moment to the received signal this network system is just seemed particularly important, especially in requiring than higher working environment recognition speed.
Summary of the invention
Technical problem: the purpose of this utility model provides a kind of distribution type fiber-optic perturbation sensing system limitation type signal processing identification module, and this device can identify real-time and accurately has dangerous incident, thereby takes appropriate measures.
Technological scheme: interfere on the distributed sensing circuitry basis at full fiber reflection, the utility model is according to the genesis mechanism and the characteristics of all kinds of input signals in the environment, proposed the signal recognition principle of the signal processing method comprehensive utilization of the denoising of employing small echo, ICA (independent component analysis) signal separation and feature extracting methods, artificial-intelligent QDA (Quadratic DiscriminantAnalysis), identified real-time and accurately and have dangerous incident.This principle implementation procedure comprises the content of two aspects, one, from the data-signal buffer storage, read out wrapping up in the mixed signal of taking each noise like and disturbance information, it is carried out Signal Pretreatment and separates, carry out the extraction of eigenvalue then, enter the QDA identification module at last, identify the type and the character of input signal; They are two years old, when being handled and discern, signal to be identified deposits it in signal storage to be identified, after the recognition result analysis for the treatment of input signal finishes, whether this input signal is manually monitored by the master controller decision, with the type and the character of further affirmation input signal.The signal characteristic value storage that whether will just have been discerned by monitoring personnel decision simultaneously is so that be used for the renewal in category feature storehouse later on.
Of the present utility modelly specifically constitute: the data-signal buffer storage order that is positioned at front end is connected in series with data reader, spectrum subtraction denoising module, small echo denoising module, ICA signal separation module, characteristic extracting module, two output terminals of characteristic extracting module are received the input end of identification signal eigenvalue storage and QDA recognizer respectively, and the output terminal of QDA recognizer connects the input end of category feature storehouse, recognition result storage respectively; The output terminal of recognition result storage, alarm connects the input end of master controller respectively, and the output terminal of master controller is received the input end of identification signal storage, audio player respectively; The input end of the output terminal reception identification signal storage of data reader; The output terminal order of signal storage to be identified is connected the input end in the output termination category feature storehouse of signal characteristic value storage to be identified with audio player, artificial audio monitor, signal characteristic value storage to be identified series connection.
Be shown as danger signal or user at recognition result and think that recognition result discerns when wrong, audio player can realize playing data in the signal storage to be identified by main controller controls, and concrete broadcast format and method are determined by master controller.Be together in series by signal storage to be identified, audio player, artificial audio monitor, signal characteristic value storage to be identified, category feature storehouse and formed the one and half artificial recognition feature updating devices that participate in, realize that the category feature storehouse upgrades.
Master controller is in charge of the running of whole system, comprise the realization of optical signal transmitting and reception control and navigation system, recognition result according to artificial-intelligent QDA identification module carries out different control operations, for example the analysis result of working as the environment input signal is shown as safety signal, master controller does not just carry out the corresponding subsequent control operation so, and continues work as usual; Be shown as danger signal as the analysis result to the environment input signal, master controller can select to start alarm so, controls signal storage to be identified simultaneously, and recognition data is reached audio player, the environment input signal is carried out audio frequency play.
Data reader reads data in the buffer memory, when it delivers to data spectrum subtraction denoising module, the input data can be deposited to signal storage to be identified, plays so that carry out audio data when finding danger signal later on.
Spectrum subtraction denoising module is earlier input signal to be carried out simple denoising earlier.
Small echo denoising module is input signal to be carried out the denoising of better effects if, employing be fixedly wavelet basis, i.e. Daubechies wavelet basis (Fig. 3).
The ICA signal separation module is to through denoising, still is that the data of mixed signal are carried out signal by ICA and separated, and separates laggard line data normalization, is convenient to ensuing feature extraction.
The signal that characteristic extracting module is come to biography carries out feature extraction, extracts and is characterized as LPCC (linear prediction cepstrum coefficient), utilizes ICA to carry out the secondary feature extraction after the extraction again.
Signal characteristic value storage to be identified can be stored the eigenvalue of the signal discerned, controls whether carry out longer-term storage by artificial audio monitor, is used for later category feature valve system and upgrades.
The QDA identification module calculates judgement with the eigenvalue extracted and the eigenvalue in the category feature storehouse by the QDA principle, and then provides recognition result.
The recognition result storage is used for the recognition result of QDA identification module is stored.
Data during audio player can be realized playing by control, concrete broadcast format and method are by decision.
Artificial audio monitor is that the signal sound of audio player plays is manually monitored, and then judges whether and will the intact signal characteristic value storage of firm identification be used for later category feature storehouse and upgrade.
Alarm is after the recognition result storage is shown as danger signal, started by master controller and realizes warning function.
Electrical schematic diagram of the present utility model constitutes as shown in Figure 2, and the master controller among the present invention, data reader, spectrum subtraction denoising module, small echo denoising module, ICA signal separation module, characteristic extracting module, QDA identification module, recognition result storage are included on the system host.The input of data-signal buffer storage one tunnel output welding system main frame, system host has two outputs to connect the input of audio signal playback equipment, a control signal that is output as master controller, a result who is output as data reader.In addition, system host also has the input of an output device taking alarm.
It is available that the hardware of most of module all has general market product in the native system.
Beneficial effect: by the limitation type signal processing recognition device of present networks system, can realize real-time processing fully, promptly the processing and identification time far is shorter than signal time length, and the recognition result of this method is felt simultaneously higher preparation rate.Self-teaching that this identifying method can also manually participate in and renewal can be satisfied the needs that the ambient signal that causes along with environment and seasonal variations alters a great deal.
Distribution type fiber-optic perturbation sensing system limitation type signal processing identification module is mainly used in the pipe monitoring of all kinds of conveying danger or high value chemicals pipe safety monitoring etc., and the monitoring in important warehouse, storehouse, important boat station, hangar, Explosive storehouse, national boundary, Aeronautics and Astronautics base, missile base, bank, museum, prison etc.; City tap-water, coal gas, rock gas, heat supply pipeline safety monitoring monitoring; The broad domain all-optical fiber disturbance sensing of circumference safety guards such as oil, Long-distance Transmission Pipeline safety monitoring monitoring and the disturbance framing signal treatment technology of navigation system, and the present invention is the limitation type signal processing recognition device that is specifically applied to this network system, and it relates to many technical fields such as measurement, collection, processing and identification of mechanical vibration and ambient sound.
The utility model proposes, utilize fixedly wavelet basis denoising, with ICA the mixed signal that receives being carried out signal simultaneously separates and feature extraction, the Classification and Identification function that achieves a butt joint and collect mail number by artificial-intelligent identification system QDA identification module at last, can tell danger signal and safety signal, and as required, can carry out signal plays and the eigenvalue storage update in real time.For danger signal, take corresponding action to go to clear immediately; For safety signal, then do not need to take action.The utility model can be simultaneously in conjunction with inventing the actual time safety early warning that realizes whole important area with the isonomic navigation system of the present invention.
To critical facility and zone, construct in the broad domain all-optical fiber disturbance sensing network system that circumference thing or buried pipeline have the monitoring early warning and safety function supporting in real time, signal processing recognition device accurately.The mixed signal of taking each noise like and disturbance information is wrapped up in utilization of the present invention, adopts signal processing method and artificial-intelligent QDA signal recognition principles such as small echo, ICA, and monitor critical facility in real time and disturb with all kinds of incidents that take place around the zone,
Description of drawings
Fig. 1 distribution type fiber-optic perturbation sensing system limitation type signal processing identification module theory diagram,
Fig. 2 distribution type fiber-optic perturbation sensing system limitation type signal processing identification module electrical schematic diagram,
The scaling function and the wavelet function of Fig. 3 db4 small echo,
Fig. 4 contains the tweedle signal of noise and the signal after the denoising,
Fig. 5 contains the gunslinging signal of noise and the signal after the denoising.
Wherein have:
100: system host; 101: master controller;
301: the data-signal buffer storage; 302: data reader;
303: spectrum subtraction denoising module; 304: small echo denoising module;
The 305:ICA signal separation module; 306: characteristic extracting module;
307: signal characteristic value storage to be identified; The 308:QDA identification module;
309: the category feature storehouse; 310: the recognition result storage;
311: signal storage to be identified; 312: audio player;
313: artificial audio monitor; 314: alarm;
Embodiment
Distribution type fiber-optic perturbation sensing system limitation type signal processing identification module of the present utility model can effectively be handled and discern all kinds of ambient signals of input, distinguishes user-defined danger signal and safety signal, and can make appropriate responsive to it.
As shown in Figure 1, what at first be arranged in this contrive equipment front end is the data-signal buffer storage 301 that full fiber reflection is interfered the distributed sensing circuitry, it links to each other with data reader 302, what link to each other with data reader is spectrum subtraction denoising module 303 and signal storage to be identified 311, what next link to each other with spectrum subtraction denoising module is small echo denoising module 304, what link to each other with small echo denoising module is ICA signal separation module 305, what link to each other with 305 is characteristic extracting module 306, what link to each other with characteristic extracting module is QDA identification module 308, what link to each other with the QDA identification module is recognition result storage 310, and 310 link to each other with master controller 101 more then.When signal to be identified is considered to danger signal, or the user suspects that signal to be identified is a danger signal, can by master controller 101 control reception identification signal storagies 311 with and audio player 312, make the data that audio player docks in the signal storage to be identified play, realize artificial the monitoring by artificial audio monitor 312,, the monitoring personnel are necessary that (the monitoring personnel think that this signal is known by mistake when thinking, or this signal is very typical) this signal is stored when being used for the category feature storehouse and upgrading, can be by control signal characteristic value storage 307 to be identified, the signal characteristic value to be identified of wherein storage is delivered to category feature storehouse 309, so that system upgrades in the future.In addition, master controller 101 also directly links to each other with alarm 314.
The denoising recognition process real-time that is made of small echo denoising mould 304, characteristic extracting module 306 and QDA identification module 308 is very strong, what small echo denoising module 205 adopted is that fixedly wavelet basis carries out denoising, and the QDA identification module is that each artificial intelligent identifying system medium velocity is a kind of faster.
Be shown as danger signal or user when thinking that institute's identification signal is wrong at recognition result, audio player 312 can be controlled the data that realize playing in the signal storage 311 to be identified by master controller 101, and concrete broadcast format and method are by master controller 101 decisions.
The broadcast result of audio player 312 realizes artificial the monitoring by artificial audio monitor 313, when thinking, the monitoring personnel are necessary that (the monitoring personnel think danger signal, but system identification is a safety signal, perhaps opposite) this signal characteristic value is stored and when being used for the category feature storehouse and upgrading, can be by control signal characteristic value storage 307 to be identified, the signal characteristic value that the firm identification of wherein storage is intact is delivered to category feature storehouse 309, so that system upgrades in the future.
Embodiment of the present utility model is described and the utility model is described further with this example.This example is an experimental prototype.Digital signal software pre-processing module (relating to signal denoising and all modules of separating), the characteristic signal extraction module, the artificial-intelligent identification system is all on system host, this machine model is joined 4CPU 8M buffer memory for the perfectly sound R630G5 of association, the 32G internal memory, I/O:NI PCI-E62511.25M 16channel16bit; Audio player: use be the diva series Hi-Fi sound-box and the earphone of Hui Wei company.Detailed process is: one tunnel output of data-signal buffer storage connects the input of this main frame, read and store for the input data, carry out the spectrum subtraction denoising then, next carry out small echo denoising (seeing Fig. 4 and Fig. 5), what use is Daubechies wavelet basis (see figure 3), carry out the ICA signal later on earlier and separate entering the identification link, carry out the LPCC feature extraction then, can use ICA to carry out the secondary feature extraction, the eigenvalue of extracting is sent into the QDA identification module, then the recognition result storage just can return recognition result to master controller, if the gained recognition result is that danger signal then starts alarm, the pedestrian worker that goes forward side by side monitors, if safety signal then continues proper functioning, waits for the arrival of identification requirement next time.
100: system host: adopt the perfectly sound R630G5 of association to join 4CPU 8M buffer memory, 32G internal memory;
I/O：NI?PCI-E62511.25M?16channel?16bit。
101: master controller: be included in the system host.
301: data-signal buffer storage: be included in the system host.
302: data reader: be included in the system host.
303: spectrum subtraction denoising module: adopt enhancement mode dynamic spectrum subtractive method, be included in the system host.
304: small echo denoising module: adopt the Daubechies wavelet basis, be included in the system host.
305:ICA signal separation module: adopt the FastICA processing method, be included in the system host.
306: characteristic extracting module: adopt the LPCC characteristic parameter, be included in the system host.
307: signal characteristic value storage to be identified: be included in the system host.
308:QDA identification module: adopt the QDA identifying method, be included in the system host.
309: the category feature storehouse: adopt the data of QDA identifying method training, leave in the system host storage.
310: recognition result storage: be included in the system host.
311: signal storage to be identified: be included in the system host.
312: audio player: the diva series Hi-Fi sound-box and the earphone that adopt Hui Wei company.
313: artificial audio monitor: by artificial use audio player.
314: alarm: adopt golden rich star alarm (model: JFX-2002-DL-4).
Therefore, the utility model is used the signal analysis recognition principle of means such as the signal processing method comprehensive utilization of small echo denoising, the separation of ICA signal and feature extracting methods, artificial-intelligent QDA, can effective monitoring major facility and zone, identify the dangerous situation that occurs around critical facility and the zone real-time and accurately, be with a wide range of applications.
Claims (1)
1. distribution type fiber-optic perturbation sensing system limitation type signal processing identification module, it is characterized in that being positioned at data-signal buffer storage (301) order and data reader (302) of front end, spectrum subtraction denoising module (303), small echo denoising module (304), ICA signal separation module (305), characteristic extracting module (306) is connected in series, two output terminals of characteristic extracting module (306) are received the input end of identification signal eigenvalue storage (307) and QDA recognizer (308) respectively, and the output terminal of QDA recognizer (308) connects category feature storehouse (309) respectively, the input end of recognition result storage (310); The output terminal of recognition result storage (310), alarm (314) connects the input end of master controller (101) respectively, and the output terminal of master controller (101) is received the input end of identification signal storage (311), audio player (312) respectively; The input end of the output terminal reception identification signal storage (311) of data reader (302); The output terminal order of signal storage to be identified (311) is connected in series the input end in the output termination category feature storehouse (309) of signal characteristic value storage to be identified (307) with audio player (312), artificial audio monitor (313), signal characteristic value storage to be identified (307).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNU2008200337617U CN201173390Y (en) | 2008-04-03 | 2008-04-03 | Limit type signal processing and recognition module for distributed optical fibre disturbance sensing system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNU2008200337617U CN201173390Y (en) | 2008-04-03 | 2008-04-03 | Limit type signal processing and recognition module for distributed optical fibre disturbance sensing system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN201173390Y true CN201173390Y (en) | 2008-12-31 |
Family
ID=40200511
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNU2008200337617U Expired - Fee Related CN201173390Y (en) | 2008-04-03 | 2008-04-03 | Limit type signal processing and recognition module for distributed optical fibre disturbance sensing system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN201173390Y (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103244829A (en) * | 2013-04-27 | 2013-08-14 | 天津大学 | Distributed optical fiber sensor-based pipeline safety event grading early warning method |
| GB2539254A (en) * | 2015-06-12 | 2016-12-14 | Pimon Gmbh | Method and apparatus for monitoring pipeline |
| CN112893427A (en) * | 2021-01-14 | 2021-06-04 | 农业农村部环境保护科研监测所 | Intelligent decision-making method for heavy metal polluted farmland restoration treatment |
-
2008
- 2008-04-03 CN CNU2008200337617U patent/CN201173390Y/en not_active Expired - Fee Related
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103244829A (en) * | 2013-04-27 | 2013-08-14 | 天津大学 | Distributed optical fiber sensor-based pipeline safety event grading early warning method |
| CN103244829B (en) * | 2013-04-27 | 2015-05-13 | 天津大学 | Distributed optical fiber sensor-based pipeline safety event grading early warning method |
| GB2539254A (en) * | 2015-06-12 | 2016-12-14 | Pimon Gmbh | Method and apparatus for monitoring pipeline |
| CN112893427A (en) * | 2021-01-14 | 2021-06-04 | 农业农村部环境保护科研监测所 | Intelligent decision-making method for heavy metal polluted farmland restoration treatment |
| CN112893427B (en) * | 2021-01-14 | 2022-04-12 | 农业农村部环境保护科研监测所 | Intelligent decision-making method for heavy metal polluted farmland restoration treatment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103836347B (en) | Leakage monitoring device and method for crude oil gathering pipelines | |
| CN100582679C (en) | Broad domain all-optical fiber disturbance sensing network system self-adapting disturbance signal processing recognition device | |
| CN103500579B (en) | Audio recognition method, Apparatus and system | |
| CN107527617A (en) | Monitoring method, apparatus and system based on voice recognition | |
| CN112735473B (en) | Method and system for identifying unmanned aerial vehicle based on voice | |
| CN201173390Y (en) | Limit type signal processing and recognition module for distributed optical fibre disturbance sensing system | |
| CN104729667A (en) | Method for recognizing disturbance type in a distributed optical fiber vibration sensing system | |
| CN101251950B (en) | Wide field full optical fibre destabilization sensing network system limit type disturbance signal processing recognition device | |
| CN109712612A (en) | A kind of voice keyword detection method and device | |
| CN101098195A (en) | Optical fiber safety early-warning system | |
| CN103863784B (en) | A kind of image capturing system for belt connector fault under monitor well | |
| CN106781152A (en) | A kind of fiber grating fence intrusion alarm detecting system and method | |
| CN106383862A (en) | Violation short message detection method and system | |
| CN109065036A (en) | Method, apparatus, electronic equipment and the computer readable storage medium of speech recognition | |
| CN109557583B (en) | Seismic attribute extraction method and system | |
| CN201212819Y (en) | Adaptive signal processing recognition module for distributed optical fiber disturbing sensing system | |
| CN101900810A (en) | Method for fusing multi-probe end sonar information by using submersible as carrier | |
| CN106454255A (en) | Video monitoring system and method | |
| CN202209524U (en) | Airport aviation fuel pipe network leak detecting system based on pressure temperature analysis | |
| CN103886033A (en) | Intelligent vertical searching device and method for safety industry chain | |
| US20230024104A1 (en) | Identification of false transformer humming using machine learning | |
| CN110263283A (en) | Website detection method and device | |
| CN109272052A (en) | A kind of anti-intrusion intelligent identifying system of underground structure periphery pile foundation construction and method | |
| CN105574038B (en) | Content of text discrimination test method and device based on anti-identification rendering | |
| CN101976186A (en) | Voice recognition method of computer and computer |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Assignee: Yongda Technology Group Co., Ltd. Assignor: Southeast University Contract record no.: 2010320001084 Denomination of utility model: Limit type signal processing and recognition module for distributed optical fibre disturbance sensing system Granted publication date: 20081231 License type: Exclusive License Record date: 20100813 |
|
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20081231 Termination date: 20120403 |
|
| EE01 | Entry into force of recordation of patent licensing contract |
Assignee: Yongda Technology Group Co., Ltd. Assignor: Southeast University Contract record no.: 2010320001084 Denomination of utility model: Limit type signal processing and recognition module for distributed optical fibre disturbance sensing system Granted publication date: 20081231 License type: Exclusive License Record date: 20100813 |
|
| LICC | Enforcement, change and cancellation of record of contracts on the licence for exploitation of a patent or utility model |