CN208147865U - A kind of sound recognition system applied to robot used for intelligent substation patrol - Google Patents
A kind of sound recognition system applied to robot used for intelligent substation patrol Download PDFInfo
- Publication number
- CN208147865U CN208147865U CN201820676476.0U CN201820676476U CN208147865U CN 208147865 U CN208147865 U CN 208147865U CN 201820676476 U CN201820676476 U CN 201820676476U CN 208147865 U CN208147865 U CN 208147865U
- Authority
- CN
- China
- Prior art keywords
- sound
- signal
- processor
- storage unit
- substation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000003595 spectral effect Effects 0.000 claims abstract description 9
- 230000010358 mechanical oscillation Effects 0.000 claims description 11
- 238000007599 discharging Methods 0.000 claims description 10
- 238000004891 communication Methods 0.000 claims description 6
- 238000012544 monitoring process Methods 0.000 claims description 4
- 239000000203 mixture Substances 0.000 abstract description 8
- 230000008901 benefit Effects 0.000 abstract description 3
- 238000000034 method Methods 0.000 description 11
- 238000012545 processing Methods 0.000 description 11
- 238000005516 engineering process Methods 0.000 description 7
- 238000007689 inspection Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 239000000284 extract Substances 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000009432 framing Methods 0.000 description 3
- 238000013139 quantization Methods 0.000 description 3
- 230000035939 shock Effects 0.000 description 3
- 230000005236 sound signal Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011946 reduction process Methods 0.000 description 2
- 206010068150 Acoustic shock Diseases 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 230000002459 sustained effect Effects 0.000 description 1
Abstract
The utility model discloses a kind of sound recognition system applied to robot used for intelligent substation patrol, and microphone array is equipped in robot, and the source of sound that the microphone array will acquire is delivered to processor, and the processor connects memory；The processor includes：Sound source signal processor unit：It will switch to digital signal after source of sound；DSP unit：Digital signal is switched into live sound spectral characteristic；Taxon：Live sound spectral characteristic is classified as impingement signal and leveling style signal；The memory includes the first storage unit for storing impingement feature templates and the second storage unit for storing leveling style feature templates；The utility model has the advantage of be able to use substation's environment-identification, acquisition substation field sound rapidly and efficiently, and it is promptly and accurately effectively identified, after judging substation field sound composition, it can take timely measure and hidden danger is rectified and improved, further ensure that substation safety, stable operation.
Description
Technical field
The utility model relates to voice recognition technology fields.
Background technique
Inspecting substation equipment is the operating condition for supervision equipment, lacks limit to find and eliminate equipment in time, in advance
Accident prevention occurs, it is ensured that equipment safety operation.When Wuhu power supply company power transformation O&M room carries out inspection to substation equipment, usually
Way be the face such as integrated use vision, the sense of hearing, tactile, smell function and some corresponding instruments to substation equipment into
Row walkaround inspection.However as the development of power grid, substation's quantity increases sharply, and Wuhu power supply company electricity O&M room carries out inspection
Not only need to expend a large amount of manpower and material resources when operation, but also monitoring time is long, security risk is high, factors can all influence power transformation
The operational safety of station equipment.Inspecting substation equipment still relies on manual type to carry out at present, and in place always whether manual patrol
It is the main bugbear of power transformation O&M room work, the increase bring routing inspection efficiency of unattended operation transformer station and inspection quality problems day
Benefit is prominent, needs to develop a kind of robot suitable for substation equipment patrol inspection.
Current crusing robot can obtain live sound, by voice recognition technology to noise, electric discharge in substation
The spectrum analysis of sound, chatter, and judge the composition of live sound.Sound signal processing technology is built upon the number of sound
On the basis of learning model and Digital Signal Processing, it is related to multiple and different field such as Digital Signal Processing, pattern-recognition, is
One comprehensive branch of learning.
So-called voice recognition technology is exactly the process that machine can be identified and be understood, it refers in this process sound
Sound signal is converted into the high-tech of corresponding text or instruction, that is, machinery equipment is allowed to can be appreciated that the information in voice signal.
For sound recognition system, if computer is configured with the program groups of identification function, when sound is inputted by acquisition device
After storing inside computer and in digital form, the voice signal that recognizer just starts just to have inputted is performed some processing, and passes through
Specific algorithm extracts the voice signal property of input, will extract obtained feature and the multiple sound being previously stored later
The feature of sample compares.Sound characteristic comparison work complete after, computer will export one it considers that most " as " sound
The serial number of sound sample characteristics, we have any meaning it is known that voice signal was just now in this way, and then execute subsequent
Operation.
Experts and scholars all over the world have been successfully found a variety of methods in terms of signal processing, these sides up to now
Method has become essential data processing tools in each sound recognition system, as framing, short-time energy, zero-crossing rate,
Fourier transformation etc..They improve the performance of sound recognition system well, are the bases that voice recognition technology continues to develop
Plinth.Each sound recognition system is similar, although sometimes concrete implementation form is different in some step, it
All have following identical step：Sound transducer collected sound signal is used first and passes through certain compressed encoding
Mode carries out quantification treatment and obtains digital signal, secondly carries out pretreatment operation to obtained digital signal, then uses feature
Extraction algorithm extracts the feature vector of voice signal, provides recognition result eventually by algorithms of template matching recognition.
But be directed to patrol formula robot, current sound processing method be difficult to quickly timely to pass by equipment carry out
Effectively identification, the practicability that discrimination lowly causes crusing robot sound recognition system to use reduce, it is difficult to popularize.
Utility model content
Technical problem to be solved in the utility model is to realize that a kind of sound for being able to use substation's environment-identification is known
Other method.
To achieve the goals above, the technical solution adopted in the utility model is：One kind being applied to intelligent substation patrol
The sound recognition system of robot, including robot, the microphone array being mounted in robot, the microphone array will obtain
The source of sound taken is delivered to processor, and the processor connects memory；
The processor includes：
Sound source signal processor unit：It will switch to digital signal after source of sound；
DSP unit：Digital signal is switched into live sound spectral characteristic；
The memory includes the first storage unit for storing impingement feature templates and for storing leveling style spy
Levy the second storage unit of template；
The processor further includes taxon：Live sound spectral characteristic is classified as impingement signal and leveling style letter
Number.
The processor further includes subclassification unit：It is electric discharge acoustical signal and mechanical oscillation sound letter by leveling style Modulation recognition
Number, second storage unit includes for storing the first of discharging sound feature templates the sub- storage unit and for mechanical oscillation sound
The sub- storage unit of the second of feature templates.
The processor connection communication unit, the processor pass through communication unit to remote monitoring and/or warning device
Send signal.
The utility model has the advantage of substation's environment-identification is able to use, acquisition substation field sound rapidly and efficiently
Sound, and promptly and accurately effectively identified, after judging substation field sound composition, it can take timely measure to hidden danger
It is rectified and improved, further ensures that substation safety, stable operation.
Detailed description of the invention
The content of width attached drawing every in the utility model specification expression is briefly described below：
Fig. 1 is sound recognition system composition schematic diagram；
Fig. 2-7 is the common sound audio spectrogram of substation.
Specific embodiment
For sound recognition system composition as shown in Figure 1, in sound recognition system, acquisition and quantization are entire voice recognition systems
The basis of system, microphone are the most important components of collection process, it plays a part of " ear ", it hears we are daily
Sound by being converted into the electric signal of analog quantity.Quantization plays a part of analog-to-digital conversion, it can be by this analoglike of electric signal
Signal is converted into the digital signal that computer can identify and handle.Pre-treatment step is the important composition portion of sound recognition system
Point.In complete system, in order to reduce the calculation amount of system, the real-time of whole system is improved, is often examined using endpoint
Survey technology intercepts useful signal segment, calculates useless signal segment without handling.In order to facilitate processing, obtain
To signal segment be often divided into a bit of signal, here it is sub-frame processings.The foundation of sub-frame processing is voice signal tool
There is short-term stationarity, ordinary circumstance takes the length that 10-30 milliseconds are a frame signal.But compulsory sub-frame processing is carried out to signal
Afterwards, signal tends to occur the phenomenon that spectrum leakage, at this moment need windowing process.Difference is often obtained using different window functions
Effect, under normal circumstances using the window functions such as Hamming window, Hanning window.In practice by acquiring in the signal that quantization obtains often
It is mingled with many noise signals, in order to improve the signal-to-noise ratio of signal, to improve the recognition performance of entire identifying system, needs pair
Signal carries out some noise reduction process.Here end-point detection technology, framing windowing process and noise reduction process belongs to pretreatment step
Suddenly.
Pattern matching is the core of entire sound recognition system.Feature extraction can extract to obtain sound letter
Characteristic parameter in number, obtains feature vector.In the world of computer, voice signal can be indicated with eigen vector, side
Just calculation processing.Computer is matched the feature of extraction with the feature templates in feature database one by one, is finally provided system and is recognized
For a most like template number, which is exactly template matching.
Substation's sound is broadly divided into general noise, discharging sound, chatter, and robot obtains source of sound by microphone array
After switch to digital signal, live sound spectral characteristic, and same feature database are obtained after amplification, framing, discrete Fourier transform
In feature templates matched, thus judge scene sound composition (wherein feature templates need at the scene collection analysis obtain
Out), as illustrated in figs. 2-7.
Discharging sound, mechanical oscillation sound these two types signal time-domain diagram have apparent pulse-like signal, sometime can
Amount has apparent mutation.And the comparison that the energy of general noise signal just shows is steady.One can be found according to this characteristic
Three classes signal can be divided into two classes by this amount to reflect the catastrophe of energy by characteristic quantity.One kind is impingement
Signal, i.e. discharging sound, mechanical oscillation sound, one kind are leveling style signal, i.e. noise.
The mechanical oscillation of equipment have periodically, by between shock pulse moment energy duration and adjacent pulse
Interval, it is estimated that the period of a certain signal is (between shock pulse moment energy duration and adjacent pulse
Every), i.e., customized in advance to the period, by the period, we can identify mechanical chatter and discharging sound, mechanical oscillation acoustic shock
Pulse time energy is sustained for longer than discharging sound shock pulse moment energy duration, mechanical oscillation sound adjacent pulse
Between interval be less than discharging sound adjacent pulse between interval.By these features, can design a classifier by this two
Class impingement signal distinguishing.
According to the Special use environment and specific use of identifying system, by presorting to sound, and classify
Matching, the raising matching spent time that can be increased, improves the efficiency and accuracy rate of identification, quickly judges substation in this way
After live sound composition, it can take timely measure and hidden danger is rectified and improved, further ensure that substation safety, stable operation.
Above-mentioned recognition methods is applied to the sound recognition system of robot used for intelligent substation patrol, which includes machine
People, the microphone array being mounted in robot, the source of sound that microphone array will acquire are delivered to processor, and the processor connects
Connect memory；Wherein processor includes：
Sound source signal processor unit：It will switch to digital signal after source of sound；
DSP unit：Digital signal is switched into live sound spectral characteristic；
Taxon：Live sound spectral characteristic is classified as impingement signal and leveling style signal；
Subclassification unit：It is electric discharge acoustical signal and mechanical oscillation acoustical signal by leveling style Modulation recognition；
Memory includes for storing the first storage unit of impingement feature templates, for storing leveling style feature templates
The second storage unit, wherein the second storage unit includes for storing the first of discharging sound feature templates the sub- storage unit and use
In the second sub- storage unit of mechanical oscillation acoustic signature template.
Processor connection communication unit, processor send signal to remote monitoring and/or warning device by communication unit.
The utility model is exemplarily described above in conjunction with attached drawing, it is clear that the utility model specific implementation not by
The limitation of aforesaid way, as long as the various unsubstantialities carried out using the method concept and technical solution of the utility model change
Into, or it is not improved the conception and technical scheme of the utility model are directly applied into other occasions, in the utility model
Protection scope within.
Claims (3)
1. a kind of sound recognition system applied to robot used for intelligent substation patrol, including robot, are mounted in robot
Microphone array, the source of sound that the microphone array will acquire is delivered to processor, and the processor connects memory；
The processor includes：
Sound source signal processor unit：It will switch to digital signal after source of sound；
DSP unit：Digital signal is switched into live sound spectral characteristic；
It is characterized in that：The memory includes the first storage unit for storing impingement feature templates and flat for storing
Second storage unit of steady type feature templates；
The processor further includes taxon：Live sound spectral characteristic is classified as impingement signal and leveling style signal.
2. the sound recognition system according to claim 1 applied to robot used for intelligent substation patrol, it is characterised in that：
The processor further includes subclassification unit：It is acoustical signal and the mechanical oscillation acoustical signal of discharging by leveling style Modulation recognition, it is described
Second storage unit includes for storing the first of discharging sound feature templates the sub- storage unit and for mechanical oscillation acoustic signature mould
The sub- storage unit of the second of plate.
3. the sound recognition system according to claim 1 or 2 applied to robot used for intelligent substation patrol, feature exist
In：The processor connection communication unit, the processor send letter to remote monitoring and/or warning device by communication unit
Number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201820676476.0U CN208147865U (en) | 2018-05-08 | 2018-05-08 | A kind of sound recognition system applied to robot used for intelligent substation patrol |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201820676476.0U CN208147865U (en) | 2018-05-08 | 2018-05-08 | A kind of sound recognition system applied to robot used for intelligent substation patrol |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN208147865U true CN208147865U (en) | 2018-11-27 |
Family
ID=64389841
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201820676476.0U Active CN208147865U (en) | 2018-05-08 | 2018-05-08 | A kind of sound recognition system applied to robot used for intelligent substation patrol |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN208147865U (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108501003A (en) * | 2018-05-08 | 2018-09-07 | 国网安徽省电力有限公司芜湖供电公司 | A kind of sound recognition system and method applied to robot used for intelligent substation patrol |
| CN112014774A (en) * | 2020-03-26 | 2020-12-01 | 中国铁建电气化局集团第二工程有限公司 | Transformer fault inspection system and method based on sound processing |
-
2018
- 2018-05-08 CN CN201820676476.0U patent/CN208147865U/en active Active
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108501003A (en) * | 2018-05-08 | 2018-09-07 | 国网安徽省电力有限公司芜湖供电公司 | A kind of sound recognition system and method applied to robot used for intelligent substation patrol |
| CN112014774A (en) * | 2020-03-26 | 2020-12-01 | 中国铁建电气化局集团第二工程有限公司 | Transformer fault inspection system and method based on sound processing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102426835B (en) | Method for identifying local discharge signals of switchboard based on support vector machine model | |
| CN112201260B (en) | Transformer running state online detection method based on voiceprint recognition | |
| CN110208022A (en) | Power equipment multiple features audio-frequency fingerprint fault diagnosis method and system based on machine learning | |
| CN108169639B (en) | Method for identifying switch cabinet fault based on parallel long-time and short-time memory neural network | |
| Negi et al. | Event detection and its signal characterization in PMU data stream | |
| CN109949823B (en) | DWPT-MFCC and GMM-based in-vehicle abnormal sound identification method | |
| CN102623009B (en) | Abnormal emotion automatic detection and extraction method and system on basis of short-time analysis | |
| CN102005070A (en) | Voice identification gate control system | |
| CN106328166A (en) | Man-machine dialogue anomaly detection system and method | |
| CN104198899B (en) | Partial discharge type identifying method for transformer under multi-discharge source situation | |
| CN107153150A (en) | A kind of power distribution network over-voltage fault type recognition method and device | |
| CN208147865U (en) | A kind of sound recognition system applied to robot used for intelligent substation patrol | |
| CN104614069A (en) | Voice detection method of power equipment failure based on combined similar diagonalizable blind source separation algorithm | |
| CN108538757A (en) | A kind of solar cell processing procedure retroactive method | |
| CN102982347A (en) | Method for electric energy quality disturbance classification based on KL distance | |
| CN108196164B (en) | Method for extracting cable fault point discharge sound signal under strong background noise | |
| CN110488675A (en) | A kind of substation's Abstraction of Sound Signal Characteristics based on dynamic time warpping algorithm | |
| CN116778964A (en) | Power transformation equipment fault monitoring system and method based on voiceprint recognition | |
| CN104089699A (en) | Substation equipment sound reconstruction algorithm | |
| CN108872743A (en) | A kind of electric energy quality real-time monitoring analysis system and method based on wavelet transformation | |
| CN112207631A (en) | Method for generating tool detection model, method, system, device and medium for detecting tool detection model | |
| CN113345399A (en) | Method for monitoring sound of machine equipment in strong noise environment | |
| CN115376526A (en) | Power equipment fault detection method and system based on voiceprint recognition | |
| CN112348052A (en) | Power transmission and transformation equipment abnormal sound source positioning method based on improved EfficientNet | |
| CN105354170A (en) | CEEMD and singular value decomposition based recognition method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GR01 | Patent grant | ||
| GR01 | Patent grant |