CN108593282A - A kind of breaker on-line monitoring and fault diagonosing device and its working method - Google Patents
A kind of breaker on-line monitoring and fault diagonosing device and its working method Download PDFInfo
- Publication number
- CN108593282A CN108593282A CN201810728487.3A CN201810728487A CN108593282A CN 108593282 A CN108593282 A CN 108593282A CN 201810728487 A CN201810728487 A CN 201810728487A CN 108593282 A CN108593282 A CN 108593282A
- Authority
- CN
- China
- Prior art keywords
- breaker
- point
- digital
- parameter
- voice
- 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
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000012544 monitoring process Methods 0.000 title claims abstract description 27
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000006073 displacement reaction Methods 0.000 claims abstract description 20
- 238000004458 analytical method Methods 0.000 claims abstract description 15
- 238000012549 training Methods 0.000 claims description 16
- 238000006243 chemical reaction Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 11
- 238000005070 sampling Methods 0.000 claims description 10
- 238000001228 spectrum Methods 0.000 claims description 10
- 238000001514 detection method Methods 0.000 claims description 9
- 230000003595 spectral effect Effects 0.000 claims description 9
- 230000001133 acceleration Effects 0.000 claims description 6
- 238000009432 framing Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000005055 memory storage Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 241000208340 Araliaceae Species 0.000 claims description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 239000004744 fabric Substances 0.000 claims description 3
- 230000014509 gene expression Effects 0.000 claims description 3
- 235000008434 ginseng Nutrition 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 229920003023 plastic Polymers 0.000 claims description 3
- 239000004033 plastic Substances 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000005299 abrasion Methods 0.000 claims description 2
- 230000003190 augmentative effect Effects 0.000 claims description 2
- 230000005540 biological transmission Effects 0.000 claims description 2
- 230000004069 differentiation Effects 0.000 claims description 2
- 238000005315 distribution function Methods 0.000 claims description 2
- 230000008451 emotion Effects 0.000 claims description 2
- 230000000977 initiatory effect Effects 0.000 claims description 2
- 239000000203 mixture Substances 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims description 2
- 238000010183 spectrum analysis Methods 0.000 claims description 2
- 230000003068 static effect Effects 0.000 claims description 2
- 239000013589 supplement Substances 0.000 claims description 2
- 238000012546 transfer Methods 0.000 claims description 2
- 238000003745 diagnosis Methods 0.000 abstract description 3
- 238000012512 characterization method Methods 0.000 abstract description 2
- 238000005259 measurement Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/18—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being spectral information of each sub-band
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/24—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being the cepstrum
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Computational Linguistics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The present invention relates to a kind of breaker on-line monitoring and fault diagonosing device and its working methods, including central processing unit and coupled power supply, RTC timers, reset circuit, displacement sensor, analog/digital converter, voice module, jtag interface, switch value input interface, warning circuit etc.；Comprehensive mechanical Characterization method of the present invention and voice messaging method analysis the High Voltage Circuit Breaker Condition thinking be, in characteristic parameter layer, the characteristic parameter of mechanical property method and voice messaging analysis method is combined into a new characteristic parameter, this new characteristic parameter is input in trained gauss hybrid models again, the model closer under which kind of operating status judged, judges the operating status of breaker accordingly；This thinking had not only remained the parameter of most direct reflection circuit-breaker status, but also referred to the speech characteristic parameter for being not easy to be affected by the external environment, and guarantee is provided to the online accurate measurements of the High Voltage Circuit Breaker Condition and fault diagnosis.
Description
Technical field
The present invention relates to a kind of embedded speech information analysis method and devices, belong to technical field of voice recognition, specifically
It is to be related to a kind of breaker on-line monitoring and fault diagonosing device and its working method.
Background technology
Breaker application scenario and its extensively, working condition affects the job stability of entire electric system.Nowadays,
The maintenance mode of power equipment has occurred that change, no longer using the action frequency of equipment as foundation, but with the state of equipment
For foundation, this means that carrying out on-line monitoring to circuit-breaker status becomes most important.
Statistical result shows that the operating mechanism of breaker is the common place of generation problem, and mechanical property is to cause
Most of circuit breaker failure main causes.By the research and development of decades, the monitoring means of breaker is constantly innovated, wherein
What is be most widely used is stroke/time graph analytic approach, because it can detect the mechanical property of breaker, but there is also
The shortcomings of wiring is complicated, cumbersome, sensor is difficult to install；Image measuring method just starts the failure for being used for breaker
Diagnosis research, this method is still under development.
The conjunction of high-voltage circuitbreaker (divides) lock operation mainly to be completed by its mechanical part, and mechanical part reliability of operation is straight
The normal operation for being related to breaker is connect, mechanical property status monitoring is significant.But mechanical property on-line monitoring and failure
Diagnostic method there are problems that being affected by strong-electromagnetic field around, and the otherness of the difference and environment between breaker
So that the working characteristics of breaker is not identical, the variation of breaker mechanic property, it is desirable that processing data are obtained
Algorithm is improved, and can be only achieved the effect to the stability monitoring of breaker.
More to accurately acquire the operating status of breaker, the present invention proposes to increase a kind of real based on voice messaging analysis
Mechanical property parameter, is combined with voice messaging characteristic parameter in characteristic parameter layer, obtains by the method for existing circuit breaker failure detection
The characteristic parameter new to one, then new characteristic parameter is input in trained gauss hybrid models, judge closer to which
The model of a state makes the judgement to the High Voltage Circuit Breaker Condition accordingly.The on-line monitoring and fault diagonosing of breaker can be
Failure sends out warning before occurring, to improve breaker reliability of operation.
Invention content
Problems solved by the invention is：It is affected to breaker mechanic property on-line monitoring result to make up environmental factor
Defect, while proposing a kind of completely new the High Voltage Circuit Breaker Condition on-line monitoring method based on speech signal analysis, in conjunction with
Upper background and demand, the present invention provide a kind of based on speech signal analysis and the breaker of mechanical Characteristic on-line monitoring and event
Hinder diagnostic device and its working method.This system can either take into account the on-line monitoring of breaker mechanic property, and can pass through
Voice messaging analysis is monitored on-line to make up error caused by the otherness of mechanical property under varying environment, is reached more accurately
On-line monitoring and breakdown judge result.
To solve the technical problem of the present invention, used technical solution is：
A kind of breaker on-line monitoring and fault diagonosing device, the device include central processing unit and coupled electricity
Source, RTC timers, reset circuit, displacement sensor, analog/digital converter, voice module, RS485 interfaces, RS232 connect
Mouth, jtag interface, switch value input interface, warning circuit, SRAM and EEPROM；
Central processing unit is the control chip of signal acquisition and processing, the TMS320F28335 cores produced using TI companies
Piece；
Displacement sensor measures breaker trip-time response, using WDL25-2 type direct slide conductive plastic potentiometers；
It exports the input terminal of termination analog/digital converter；
Analog/digital converter completes the analog/digital conversion of sensor acquisition signal；Analog/digital converter is parallel
Interface is connected with the data/address bus of processor, and wherein chip select terminal is connected with the XZCS0 of processor, i.e., stores the data of AD conversion
In the areas piece external memory storage area XINTF 0 of processor, read-write control signal end is connected with XRD the and XWE signals of processor；
Voice module is connect with processor by SPI interface, including memory module and processing module, pre- in memory module
There is speech recognition modeling, processing module is used to judge current breaker based on speech recognition modeling and current voice messaging
Operation conditions；
SRAM and EEPROM, is the memory of piece external expansion, and the data and parameter when respectively preserving system operation do standard
It is standby；SRAM parallel interfaces are connected with the data/address bus of processor, and chip select terminal is connected with the XZCS6 signals of processor, indicate external
RAM bit is in the piece external memory storage area XINTF6 of processor；The IIC host controllers that EEPROM is provided with processor are connected directly；
Switch value input interface, the i.e. monitoring of breaker switching on and off signal and warning circuit；
RS485 interfaces, for the serial line interface communicated with host computer；The 3.3V power supply chips produced using TI companies
SN65HVD12 is connected to processor as 485 transceiving chips, transmitting-receiving control terminal by GPIO mouthfuls, by the height for controlling GPIO mouthfuls
Level switches 485 reiving/transmitting state, receives transmitting terminal and is connected by SCIC interfaces with processor；
RS232 interface can be used as debugging or connect other serial equipments for meeting 232 level standards；It is given birth to using MAXIM company
The MAX3232 chips of production complete level conversion, and are connected by SCIB interfaces with processor；
Jtag interface is artificial debugging interface and program burn writing interface；
Warning circuit is connected by GPIO mouthfuls with processor.
A kind of working method of breaker on-line monitoring and fault diagonosing device, includes the following steps：
Step A：It obtains breaker switching on and off and signal occurs；
Step B：Voice snippet input is received, speech characteristic parameter F1 is extracted, this feature parameter is that 12 Wei Meier frequencies are fallen
Spectral coefficient；
Step C：Displacement sensor output signal is received, vacuum breaker mechanical features parameter F2 is sought：When splitting or integrating lock
Between, contact travel, open away from, overtravel, splitting or integrating lock average speed and just closed just-off speed, and be transferred to by RS485 upper
Machine；
Step D：Obtain the characteristic parameter of 18 dimensionsCharacteristic parameter F3 is input to trained state
In identification model, to determine which state recognition model is a best match, selected according to matched state recognition model
The corresponding working state of circuit breaker of the model is selected, and the results are shown in host computers.
The advantageous effect of the present invention compared with the existing technology is：
(1) present invention increases voice messaging on the basis of common mechanical property on-line monitoring and fault diagonosing method
The method of analysis, mechanical property on-line monitoring and fault diagonosing method there are problems that being affected by strong-electromagnetic field around,
And the otherness of the difference and environment between breaker so that the working characteristics of breaker is not identical, breaker
The variation of mechanical property, it is desirable that algorithm is obtained to processing data and is improved, the effect to the stability monitoring of breaker is can be only achieved
The method of fruit, voice messaging analysis will not then be influenced by environmental factor, compensate for use mechanical property merely to a certain extent
Error caused by method.
(2) comprehensive mechanical Characterization method of the present invention and voice messaging method analysis the High Voltage Circuit Breaker Condition thinking be,
The characteristic parameter of mechanical property method and voice messaging analysis method is combined into a new characteristic parameter by characteristic parameter layer,
This new characteristic parameter is input in trained gauss hybrid models again, the mould closer under which kind of operating status judged
Type judges the operating status of breaker accordingly；This thinking had not only remained the parameter of most direct reflection circuit-breaker status, but also drew
With the speech characteristic parameter for being not easy to be affected by the external environment, carried to the online accurate measurements of the High Voltage Circuit Breaker Condition and fault diagnosis
It has supplied to ensure.
Description of the drawings
The present invention is made with attached drawing with reference to embodiments and being discussed further.
Fig. 1 is the system structure diagram of the present invention；
Fig. 2 is the structure diagram of voice module；
Fig. 3 is the mechanical property acquisition module software the general frame of the present invention；
Fig. 4 is the mechanical property acquisition software flow chart of the present invention；
Fig. 5 is the feature extraction flow chart of the present invention.
Specific implementation mode
Embodiment 1
Please refer to Fig.1 with shown in 2, a kind of breaker on-line monitoring and fault diagonosing device, including central processing unit 101 and
Coupled power supply 102, RTC timers 103, reset circuit 104, displacement sensor 105, analog/digital converter 106,
Voice module 107, RS485 interfaces 108, RS232 interface 109, jtag interface 110, switch value input interface 111, warning circuit
112, SRAM113 and EEPROM114.
Central processing unit 101 is the control chip of signal acquisition and processing, the TMS320F28335 produced using TI companies
Chip.
Displacement sensor 105 measures breaker trip-time response, using WDL25-2 type direct slide conductive plastics current potentials
Device.It exports the input terminal of termination analog/digital converter 106.
Analog/digital converter 106 completes the analog/digital conversion of sensor acquisition signal.Analog/digital converter
106 parallel interfaces are connected with the data/address bus of processor 101, and wherein chip select terminal is connected with the XZCS0 of processor 101, i.e., by AD
The data of conversion are stored in the areas piece external memory storage area XINTF 0 of processor 101, the XRD at read-write control signal end and processor 101
It is connected with XWE signals.
Voice module 107 is connect with processor 101 by SPI interface, including memory module 1071 and processing module
1072, prestore speech recognition modeling in memory module 1071, processing module 1072 is used for based on speech recognition modeling and current
Voice messaging judge the operation conditions of current breaker.
SRAM113 and EEPROM114, is the memory of piece external expansion, respectively preserves data and ginseng when system operation
Number is prepared.SRAM113 parallel interfaces are connected with the data/address bus of processor 101, and chip select terminal and the XZCS6 of processor 101 believe
Number it is connected, indicates that external RAM is located at the piece external memory storage area XINTF6 of processor 101.EEPROM114 is provided with processor 101
IIC host controllers are connected directly.
Switch value input interface 111, the i.e. monitoring of breaker switching on and off signal and warning circuit.
RS485 interfaces 108, for the serial line interface communicated with host computer.The 3.3V power supply chips produced using TI companies
SN65HVD12 is connected to processor 101 as 485 transceiving chips, transmitting-receiving control terminal by GPIO mouthfuls, by the height for controlling GPIO mouthfuls
Low level switches 485 reiving/transmitting state, receives transmitting terminal and is connected by SCIC interfaces with processor 101.
RS232 interface 109 can be used as debugging or connect other serial equipments for meeting 232 level standards.Using MAXIM company
The MAX3232 chips of production complete level conversion, and are connected by SCIB interfaces with processor 101.
Jtag interface 110 is artificial debugging interface and program burn writing interface.
Warning circuit 112 is connected by GPIO mouthfuls with processor 101.
Embodiment 2
Also referring to Fig. 3-5, the working method of breaker on-line monitoring and fault diagonosing device includes the following steps：
Step A：It obtains breaker switching on and off and signal occurs.
Step B：Voice snippet input is received, speech characteristic parameter F1 is extracted, this feature parameter is that 12 Wei Meier frequencies are fallen
Spectral coefficient.
Step C：Displacement sensor output signal is received, vacuum breaker mechanical features parameter F2 is sought：When splitting or integrating lock
Between, contact travel, open away from, overtravel, splitting or integrating lock average speed and just closed just-off speed, and be transferred to by RS485 upper
Machine.
Step D：Obtain the characteristic parameter of 18 dimensionsCharacteristic parameter F3 is input to trained state
In identification model, to determine which state recognition model is a best match, selected according to matched state recognition model
The corresponding working state of circuit breaker of the model is selected, and the results are shown in host computers.
In the present embodiment, speech characteristic parameter F1 is extracted with the following method in step B：
Step 1：Receive voice snippet input to be identified.
Step 2：Voice snippet to be identified is digitized to provide audio digital signals.
Step 3：Audio digital signals X (n) to be identified is pre-processed, including preemphasis, framing, adding window, endpoint
Detection：
Step 3.1：Following carry out preemphasis is pressed to audio digital signals X (n) to be identified：
α=0.9375 in formula, n indicate emotion digital speech discrete point serial number to be identified.
Step 3.2：Framing is carried out using the method for overlapping segmentation, has overlapping part between former frame and a later frame, claims
It is moved for frame, frame pipettes 7ms herein, i.e., 80 points, each frame length is taken to take 23ms, that is, take 256 under 11.025kHz sample rates
Point.
Step 3.3：Hamming window is selected to carry out windowing process to voice signal, window function is as follows：
The each frame of digital voice discrete point serial number of n ' expressions in formula, N indicate each frame of digital voice discrete point points, herein
N=256.
Step 3.4：End-point detection is completed using well known energy zero-crossing rate dual-threshold judgement method, i.e., according to ambient noise
Energy and zero-crossing rate be below the short-time energy of voice signal and the principle of short-time zero-crossing rate, first with short-time energy make first
Grade differentiates, then makees second level differentiation with short-time zero-crossing rate again on this basis, calculates the short-time energy upper limit, lower limit and zero passage
The value of rate thresholding, then judges every frame data, and each frame of digital voice X (n ') is obtained after end-point detection.
Step 4：To extracting speech characteristic parameter F1 by pretreated digital speech, this feature parameter is 12 Wei Meier frequencies
Rate cepstrum coefficient：
Step 4.1：0 is augmented afterwards in time-domain signal X (n ') so that the length of the sequence after supplement 0 is N ', and it is 2 to make N '
Integral number power, then obtains linear spectral X (k) after Discrete Fourier Transform DFT, and conversion formula is：
Step 4.2：Above-mentioned linear spectral X (k) is passed through into Mei Er frequency filter groups Hm(k) Mei Er frequency spectrums are obtained, and are led to
The processing for crossing logarithmic energy, obtains log spectrum S (m), and the overall transfer function by linear spectral X (k) to log spectrum S (m) is：
Wherein for having the filter group of M bandpass filter, m=1,2 ..., M, the transmission letter of each bandpass filter
Number is：
Step 4.3：Above-mentioned log spectrum S (m) is passed through into discrete cosine transform, transforms to cepstrum frequency domain to get to Mei Er
Frequency cepstral coefficient c (n ')：
In the present embodiment, vacuum breaker mechanical features parameter F2 is extracted with the following method in step C：
Step 1：Receive displacement sensor output signal.
Step 2：Sampling analogue voltage signal is converted to digital signal by analog/digital converter.
Step 3：Sampled signal is pre-processed, i.e., digital filtering is carried out to it：
Step 3.1：Spectrum analysis is made to stroke-time graph, spectrogram is obtained using Fast Fourier Transform (FFT).
Step 3.2：Analysis spectrum figure obtains the design objective of digital filter.
Step 3.3：According to index, analysis needs digital filter types and relevant parameter to be used.Final determination makes
With FIR lowpass digital filters, window function metht is selected to carry out adding Hanning window N=80, Fs=to ideal filter sequence
40800HZ, Fc=3000HZ.
The window function of Hanning window is：
Step 4：Double extremum methods are taken to seek splitting or integrating lock exchange point to point (conjunction) lock curve：
Step 4.1：The information of 4000 points of setting sampling, primary every 20us samplings in sampling routine, through overtesting
Understand that preceding 3000 points contain all moving contact of breaker travel informations of breaker.
Step 4.2：After obtaining displacement data, carry out difference twice to it, obtain speed-time curve and acceleration-when
Half interval contour.
Step 4.3：Maximum point adjacent on speed-time curve and minimum point pair can be found by searching for algorithm
(i.e. adjacent wave crest and trough), and judge according to this position of exchange point.
Step 5：Vacuum breaker mechanical features parameter F2 is sought according to obtained splitting or integrating lock exchange point：When splitting or integrating lock
Between, contact travel, open away from, overtravel, splitting or integrating lock average speed and just closed just-off speed.
Judge that the method for exchange point position is in above-mentioned steps 4.3：
In making process, start from scratch, if extreme point is to the energy following two conditions of first fit, you can judge the extreme value
The maximum point of point centering is combined floodgate exchange point.
1) continuously the points less than 0 are more than to set in advance to acceleration in extreme point is to corresponding acceleration-time graph
The threshold values K1 set.
2) extreme point is more than pre-set threshold values to the absolute value of the difference of corresponding speed-time curve medium velocity
K2。
During separating brake, since zero moment, this can determine that when extreme point condition following two to first fit
Position corresponding to the speed minimum point of extreme point centering is separating brake exchange point.
1) in the acceleration corresponding to extreme point-time graph section, the time that acceleration is continuously more than zero is more than advance
The threshold values M1 of setting.
2) extreme point is more than preset threshold values M2 to the absolute value of the difference of corresponding speed.
It is as follows to the circular of F2 parameters in above-mentioned steps 5：
The splitting or integrating lock time：There is electric current by since splitting or integrating brake cable circle, until moving contact moves to exchange point
Time, as splitting or integrating lock time.Whether the splitting or integrating lock time, which stablizes, reflects breaker mechanism flexibility and abrasion condition.
Moving contact stroke：Calculate switch on of moving contact before stable position to combined floodgate after stable position between displacement it
Difference, you can obtain switch on of moving contact stroke.
Moving contact open away from：By combined floodgate stroke subtract overtravel can be obtained corresponding moving contact open away from.
Moving contact overtravel：Moving contact and static contact have just touched the displacement between stable position.
Breaker average speed：Breaker average speed includes combined floodgate average speed and separating brake average speed.Calculate dynamic touch
Head speed of 6ms before combined floodgate exchange point can be obtained combined floodgate average speed.Calculate moving contact 6ms after separating brake exchange point
Speed can be obtained separating brake average speed.
Rigid sum velocity：Rigid sum velocity can be obtained by calculating the speed of moving contact 2ms before combined floodgate exchange point.
Just-off speed：Just-off speed can be obtained by calculating the speed of moving contact 2ms after separating brake exchange point.
In the present embodiment, the training method of state recognition model includes the following steps in step D：
Step 1：Receive the training voice snippet input of two kinds of operating statuses.
Step 2：To training voice snippet digitlization to provide audio digital signals X (n1), wherein n1Expression state trains number
Word voice discrete point serial number.
Step 3：To audio digital signals X (n1) pre-processed, including preemphasis, framing, adding window, end-point detection, it obtains
To operating status training audio digital signals X (n1′)。
Step 4：To passing through pretreated digital speech X (n1') extraction speech characteristic parameter F1, this feature parameter is 12 dimensions
Mel frequency cepstrum coefficient.
Step 5：Two kinds of operating status bottom displacement sensor output signals are received respectively.
Step 6：Sampling analogue voltage signal is converted to digital signal by analog/digital converter.
Step 7：Sampled signal is pre-processed, i.e., digital filtering is carried out to it.
Step 8：Double extremum methods are taken to seek splitting or integrating lock exchange point to point (conjunction) lock curve.
Step 9：Vacuum breaker mechanical features parameter F2 is sought according to obtained splitting or integrating lock exchange point.
Step 10：Obtain the characteristic parameter of 18 dimensions
Step 11：Using characteristic parameter F3 come physical training condition identification model, it is as follows：
Step 11.1：The exponent number that the gauss hybrid models of state recognition model are arranged is 4.
Step 11.2：With K Mean Methods (kmeans) init state identification model, the initialization of each Gaussian Profile is obtained
Parameter：Mean vector μk, covariance matrix ∑k, mixed components weights ck, indicate the corresponding initial beggar of k-th of operating status
Model parameter.
Step 11.3：If t-th of characteristic parameter of c-th of state training voiceFor
Wherein TcIndicate the frame number of c-th of state training voice, C indicates the sum of training sample, according to following formula to Gauss point
The initiation parameter of cloth is reevaluated, and is enabledWhereinIt indicates corresponding operating status, obtains each
Running state recognition submodule shape parameter：
Step 11.4：State recognition model is gauss hybrid models, by each running state recognition submodel obtained above
Parameter substitutes into following formula, forms trained each running state recognition submodel, these trained submodel set are
For final running state recognition model：
The gauss hybrid models describe point of the frame feature in feature space with the linear combination of 4 single Gaussian Profiles
Cloth is described in detail below：
Wherein,
Wherein, D is characterized dimension, herein D=12, bk(x) it is known as kernel function, is that mean vector isCovariance matrix
ForGauss of distribution function, Gaussian Mixture distribution weighting coefficientMeet：
State recognition gauss hybrid models parameter set λ1It is exactly by above-mentioned each mean value component, covariance matrix and mixing point
The weights of amount form, and are expressed as the form of triple：
The above content is just an example and description of the concept of the present invention, affiliated those skilled in the art
It makes various modifications or additions to the described embodiments or substitutes by a similar method, without departing from invention
Design or beyond the scope defined by this claim, be within the scope of protection of the invention.
Claims (7)
1. a kind of breaker on-line monitoring and fault diagonosing device, which is characterized in that the device include central processing unit (101) and
Coupled power supply (102), RTC timers (103), reset circuit (104), displacement sensor (105), analog/digital turn
Parallel operation (106), voice module (107), RS485 interfaces (108), RS232 interface (109), jtag interface (110), switching value are defeated
Incoming interface (111), warning circuit (112), SRAM (113) and EEPROM (114)；
Central processing unit (101) is the control chip of signal acquisition and processing, the TMS320F28335 cores produced using TI companies
Piece；
Displacement sensor (105) measures breaker trip-time response, using WDL25-2 type direct slide conductive plastics current potentials
Device；It exports the input terminal of termination analog/digital converter (106)；
Analog/digital converter (106) completes the analog/digital conversion of sensor acquisition signal；Analog/digital converter
(106) parallel interface is connected with the data/address bus of processor (101), and wherein chip select terminal is connected with the XZCS0 of processor (101),
The data of AD conversion are stored in the areas piece external memory storage area XINTF 0 of processor (101), read-write control signal end and processor
(101) XRD with XWE signals are connected；
Voice module (107) is connect with processor (101) by SPI interface, including memory module (1071) and processing module
(1072), speech recognition modeling is prestored in memory module (1071), processing module (1072) is used to be based on speech recognition modeling
And current voice messaging judges the operation conditions of current breaker；
SRAM (113) and EEPROM (114), is the memory of piece external expansion, respectively preserves data and ginseng when system operation
Number is prepared；SRAM (113) parallel interfaces are connected with the data/address bus of processor (101), chip select terminal and processor (101)
XZCS6 signals are connected, and indicate that external RAM is located at the piece external memory storage area XINTF6 of processor (101)；EEPROM (114) and processing
The IIC host controllers that device (101) provides are connected directly；
Switch value input interface (111), the i.e. monitoring of breaker switching on and off signal and warning circuit；
RS485 interfaces (108), for the serial line interface communicated with host computer；The 3.3V power supply chips produced using TI companies
SN65HVD12 is as 485 transceiving chips, and transmitting-receiving control terminal is connected to processor (101) by GPIO mouthfuls, by controlling GPIO mouthfuls
Low and high level switches 485 reiving/transmitting state, receives transmitting terminal and is connected by SCIC interfaces with processor (101)；
RS232 interface (109) can be used as debugging or connect other serial equipments for meeting 232 level standards；It is given birth to using MAXIM company
The MAX3232 chips of production complete level conversion, and are connected by SCIB interfaces with processor (101)；
Jtag interface (110) is artificial debugging interface and program burn writing interface；
Warning circuit (112) is connected by GPIO mouthfuls with processor (101).
2. a kind of working method of breaker on-line monitoring and fault diagonosing device as described in claim 1, which is characterized in that packet
Include following steps：
Step A：It obtains breaker switching on and off and signal occurs；
Step B：Voice snippet input is received, speech characteristic parameter F1 is extracted, this feature parameter is 12 dimension Mel frequency cepstrum systems
Number；
Step C：Displacement sensor output signal is received, vacuum breaker mechanical features parameter F2 is sought：The splitting or integrating lock time touches
Head stroke, open away from, overtravel, splitting or integrating lock average speed and just closed just-off speed, and host computer is transferred to by RS485；
Step D：Obtain the characteristic parameter of 18 dimensionsCharacteristic parameter F3 is input to trained state recognition mould
In type, to determine which state recognition model is a best match, which is selected according to matched state recognition model
The corresponding working state of circuit breaker of type, and the results are shown in host computers.
3. working method as claimed in claim 2, which is characterized in that extract speech characteristic parameter F1 in step B using as follows
Method：
Step 1：Receive voice snippet input to be identified；
Step 2：Voice snippet to be identified is digitized to provide audio digital signals；
Step 3：Audio digital signals X (n) to be identified is pre-processed, including preemphasis, framing, adding window, end-point detection：
Step 3.1：Following carry out preemphasis is pressed to audio digital signals X (n) to be identified：
α=0.9375 in formula, n indicate emotion digital speech discrete point serial number to be identified；
Step 3.2：Framing is carried out using the method for overlapping segmentation, there is overlapping part, referred to as frame between former frame and a later frame
It moves, frame pipettes 7ms herein, i.e., takes 80 points, each frame length to take 23ms under 11.025kHz sample rates, that is, take 256 points；
Step 3.3：Hamming window is selected to carry out windowing process to voice signal, window function is as follows：
The each frame of digital voice discrete point serial number of n ' expressions in formula, N indicate each frame of digital voice discrete point points, herein N=
256；
Step 3.4：End-point detection is completed using well known energy zero-crossing rate dual-threshold judgement method, i.e., according to the energy of ambient noise
Amount and zero-crossing rate are below the short-time energy of voice signal and the principle of short-time zero-crossing rate, are sentenced first as the first order with short-time energy
Not, then make second level differentiation with short-time zero-crossing rate again on this basis, calculate the short-time energy upper limit, lower limit and zero-crossing rate door
The value of limit, then judges every frame data, and each frame of digital voice X (n ') is obtained after end-point detection；
Step 4：To extracting speech characteristic parameter F1 by pretreated digital speech, this feature parameter is that 12 Wei Meier frequencies are fallen
Spectral coefficient：
Step 4.1：0 is augmented afterwards in time-domain signal X (n ') so that the length of the sequence after supplement 0 is N ', and it is 2 integer to make N '
Power, then obtains linear spectral X (k) after Discrete Fourier Transform DFT, and conversion formula is：
Step 4.2：Above-mentioned linear spectral X (k) is passed through into Mei Er frequency filter groups Hm(k) Mei Er frequency spectrums are obtained, and pass through logarithm
The processing of energy, obtains log spectrum S (m), and the overall transfer function by linear spectral X (k) to log spectrum S (m) is：
Wherein for thering is the filter group of M bandpass filter, the transmission function of m=1,2 ..., M, each bandpass filter to be：
Step 4.3：Above-mentioned log spectrum S (m) is passed through into discrete cosine transform, transforms to cepstrum frequency domain to get to Mei Er frequencies
Cepstrum coefficient c (n ')：
4. working method as claimed in claim 2, which is characterized in that extract vacuum breaker mechanical features parameter in step C
F2 is with the following method：
Step 1：Receive displacement sensor output signal；
Step 2：Sampling analogue voltage signal is converted to digital signal by analog/digital converter；
Step 3：Sampled signal is pre-processed, i.e., digital filtering is carried out to it：
Step 3.1：Spectrum analysis is made to stroke-time graph, spectrogram is obtained using Fast Fourier Transform (FFT)；
Step 3.2：Analysis spectrum figure obtains the design objective of digital filter；
Step 3.3：According to index, analysis needs digital filter types and relevant parameter to be used；Final determine uses FIR
Lowpass digital filter selects window function metht to carry out adding Hanning window N=80, Fs=40800HZ, Fc=to ideal filter sequence
3000HZ；
The window function of Hanning window is：
Step 4：Double extremum methods are taken to seek splitting or integrating lock exchange point to point (conjunction) lock curve：
Step 4.1：The information of 4000 points of setting sampling, primary every 20us samplings in sampling routine, through known to overtesting
Preceding 3000 points contain all moving contact of breaker travel informations of breaker；
Step 4.2：After obtaining displacement data, difference twice is carried out to it, obtains speed-time curve and acceleration-time is bent
Line；
Step 4.3：Maximum point adjacent on speed-time curve and minimum point can be found to (i.e. phase by searching for algorithm
Adjacent wave crest and trough), and judge according to this position of exchange point；
Step 5：Vacuum breaker mechanical features parameter F2 is sought according to obtained splitting or integrating lock exchange point：The splitting or integrating lock time touches
Head stroke, open away from, overtravel, splitting or integrating lock average speed and just closed just-off speed.
5. working method as claimed in claim 4, which is characterized in that judge that the method for exchange point position is in step 4.3：
In making process, start from scratch, if extreme point is to the energy following two conditions of first fit, you can judge the extreme point pair
In maximum point be combined floodgate exchange point；
1) acceleration is more than continuously pre-set less than 0 points in extreme point is to corresponding acceleration-time graph
Threshold values K1；
2) extreme point is more than pre-set threshold values K2 to the absolute value of the difference of corresponding speed-time curve medium velocity；
During separating brake, since zero moment, this extreme value can determine that when extreme point condition following two to first fit
Position corresponding to the speed minimum point of point centering is separating brake exchange point；
1) in the acceleration corresponding to extreme point-time graph section, the time that acceleration is continuously more than zero is more than to preset
Threshold values M1；
2) extreme point is more than preset threshold values M2 to the absolute value of the difference of corresponding speed.
6. working method as claimed in claim 4, which is characterized in that the circular of F2 parameters in step 5
It is as follows：
The splitting or integrating lock time：There is electric current by since splitting or integrating brake cable circle, until moving contact moves to the time of exchange point,
The as splitting or integrating lock time；Whether the splitting or integrating lock time, which stablizes, reflects breaker mechanism flexibility and abrasion condition；
Moving contact stroke：The difference of the displacement between stable position of the stable position before switch on of moving contact to after closing a floodgate is calculated,
It can be obtained switch on of moving contact stroke；
Moving contact open away from：By combined floodgate stroke subtract overtravel can be obtained corresponding moving contact open away from；
Moving contact overtravel：Moving contact and static contact have just touched the displacement between stable position；
Breaker average speed：Breaker average speed includes combined floodgate average speed and separating brake average speed；Moving contact is calculated to exist
The speed of 6ms can be obtained combined floodgate average speed before combined floodgate exchange point；Calculate the speed of moving contact 6ms after separating brake exchange point
Degree can be obtained separating brake average speed；
Rigid sum velocity：Rigid sum velocity can be obtained by calculating the speed of moving contact 2ms before combined floodgate exchange point；
Just-off speed：Just-off speed can be obtained by calculating the speed of moving contact 2ms after separating brake exchange point.
7. working method as claimed in claim 2, which is characterized in that the training method of state recognition model includes in step D
Following steps：
Step 1：Receive the training voice snippet input of two kinds of operating statuses；
Step 2：To training voice snippet digitlization to provide audio digital signals X (n1), wherein n1The digital language of expression state training
Sound discrete point serial number；
Step 3：To audio digital signals X (n1) pre-processed, including preemphasis, framing, adding window, end-point detection, it is run
State trains audio digital signals X (n1′)；
Step 4：To passing through pretreated digital speech X (n1') extraction speech characteristic parameter F1, this feature parameter is 12 Wei Meier
Frequency cepstral coefficient；
Step 5：Two kinds of operating status bottom displacement sensor output signals are received respectively；
Step 6：Sampling analogue voltage signal is converted to digital signal by analog/digital converter；
Step 7：Sampled signal is pre-processed, i.e., digital filtering is carried out to it；
Step 8：Double extremum methods are taken to seek splitting or integrating lock exchange point to point (conjunction) lock curve；
Step 9：Vacuum breaker mechanical features parameter F2 is sought according to obtained splitting or integrating lock exchange point；
Step 10：Obtain the characteristic parameter of 18 dimensions
Step 11：Using characteristic parameter F3 come physical training condition identification model, it is as follows：
Step 11.1：The exponent number that the gauss hybrid models of state recognition model are arranged is 4；
Step 11.2：With K Mean Methods (kmeans) init state identification model, the initialization ginseng of each Gaussian Profile is obtained
Number：Mean vector μk, covariance matrix ∑k, mixed components weights ck, indicate the corresponding initialization submodule of k-th of operating status
Shape parameter；
Step 11.3：If t-th of characteristic parameter of c-th of state training voiceFor
Wherein TcIndicate the frame number of c-th of state training voice, C indicates the sum of training sample, according to following formula to Gauss point
The initiation parameter of cloth is reevaluated, and is enabledWhereinIt indicates corresponding operating status, obtains each
Running state recognition submodule shape parameter：
Step 11.4：State recognition model is gauss hybrid models, by each running state recognition submodule shape parameter obtained above
Following formula is substituted into, trained each running state recognition submodel is formed, these trained submodel set are most
Whole running state recognition model：
The gauss hybrid models describe distribution of the frame feature in feature space with the linear combination of 4 single Gaussian Profiles, have
Body is described as follows：
Wherein,
Wherein, D is characterized dimension, herein D=12, bk(x) it is known as kernel function, is that mean vector isCovariance matrix is
Gauss of distribution function, Gaussian Mixture distribution weighting coefficientMeet：
State recognition gauss hybrid models parameter set λ1Exactly by above-mentioned each mean value component, covariance matrix and mixed components
Weights form, and are expressed as the form of triple：
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810728487.3A CN108593282A (en) | 2018-07-05 | 2018-07-05 | A kind of breaker on-line monitoring and fault diagonosing device and its working method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810728487.3A CN108593282A (en) | 2018-07-05 | 2018-07-05 | A kind of breaker on-line monitoring and fault diagonosing device and its working method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108593282A true CN108593282A (en) | 2018-09-28 |
Family
ID=63635305
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810728487.3A Pending CN108593282A (en) | 2018-07-05 | 2018-07-05 | A kind of breaker on-line monitoring and fault diagonosing device and its working method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108593282A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109870299A (en) * | 2019-04-02 | 2019-06-11 | 国网安徽省电力有限公司培训中心 | A kind of breaker failure detection device and a variety of data fusion fault detection methods |
| CN110097881A (en) * | 2019-04-26 | 2019-08-06 | 深圳德微电技术有限公司 | A kind of intelligent breaker of speech recognition controlled |
| CN110702401A (en) * | 2019-11-15 | 2020-01-17 | 广州供电局有限公司 | 10kV circuit breaker mechanical characteristic on-line monitoring device based on acceleration sensor |
| CN112177865A (en) * | 2020-12-02 | 2021-01-05 | 南京智谷人工智能研究院有限公司 | Method for solving marking noise and insufficient marks in fan fault detection |
| CN112365901A (en) * | 2020-11-03 | 2021-02-12 | 武汉工程大学 | Mechanical audio fault detection method and device |
| CN113314142A (en) * | 2021-05-26 | 2021-08-27 | 山东大学 | Engine fault detection method, system, storage medium and equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH09215178A (en) * | 1995-11-27 | 1997-08-15 | Matsushita Electric Works Ltd | Circuit breaker device with earthquake sensing function and residential distribution board with it |
| CN102324232A (en) * | 2011-09-12 | 2012-01-18 | 辽宁工业大学 | Method for recognizing sound-groove and system based on gauss hybrid models |
| CN102721919A (en) * | 2012-06-25 | 2012-10-10 | 南京南自西电电气有限责任公司 | Online monitoring method and online monitoring device of high voltage vacuum circuit breaker |
| CN102737629A (en) * | 2011-11-11 | 2012-10-17 | 东南大学 | Embedded type speech emotion recognition method and device |
| CN103487749A (en) * | 2013-09-18 | 2014-01-01 | 国家电网公司 | On-line monitoring and diagnosing system and method for mechanical state of high-voltage circuit breaker |
| CN205619953U (en) * | 2016-01-19 | 2016-10-05 | 华北电力大学（保定） | A multiple signal synchronization collection system for high voltage circuit breaker operating characteristic test |
| CN206362901U (en) * | 2017-01-09 | 2017-07-28 | 华北电力大学（保定） | A kind of breaker mechanic property intelligence test and status monitoring robot |
-
2018
- 2018-07-05 CN CN201810728487.3A patent/CN108593282A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH09215178A (en) * | 1995-11-27 | 1997-08-15 | Matsushita Electric Works Ltd | Circuit breaker device with earthquake sensing function and residential distribution board with it |
| CN102324232A (en) * | 2011-09-12 | 2012-01-18 | 辽宁工业大学 | Method for recognizing sound-groove and system based on gauss hybrid models |
| CN102737629A (en) * | 2011-11-11 | 2012-10-17 | 东南大学 | Embedded type speech emotion recognition method and device |
| CN102721919A (en) * | 2012-06-25 | 2012-10-10 | 南京南自西电电气有限责任公司 | Online monitoring method and online monitoring device of high voltage vacuum circuit breaker |
| CN103487749A (en) * | 2013-09-18 | 2014-01-01 | 国家电网公司 | On-line monitoring and diagnosing system and method for mechanical state of high-voltage circuit breaker |
| CN205619953U (en) * | 2016-01-19 | 2016-10-05 | 华北电力大学（保定） | A multiple signal synchronization collection system for high voltage circuit breaker operating characteristic test |
| CN206362901U (en) * | 2017-01-09 | 2017-07-28 | 华北电力大学（保定） | A kind of breaker mechanic property intelligence test and status monitoring robot |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109870299A (en) * | 2019-04-02 | 2019-06-11 | 国网安徽省电力有限公司培训中心 | A kind of breaker failure detection device and a variety of data fusion fault detection methods |
| CN110097881A (en) * | 2019-04-26 | 2019-08-06 | 深圳德微电技术有限公司 | A kind of intelligent breaker of speech recognition controlled |
| CN110702401A (en) * | 2019-11-15 | 2020-01-17 | 广州供电局有限公司 | 10kV circuit breaker mechanical characteristic on-line monitoring device based on acceleration sensor |
| CN112365901A (en) * | 2020-11-03 | 2021-02-12 | 武汉工程大学 | Mechanical audio fault detection method and device |
| CN112177865A (en) * | 2020-12-02 | 2021-01-05 | 南京智谷人工智能研究院有限公司 | Method for solving marking noise and insufficient marks in fan fault detection |
| CN113314142A (en) * | 2021-05-26 | 2021-08-27 | 山东大学 | Engine fault detection method, system, storage medium and equipment |
| CN113314142B (en) * | 2021-05-26 | 2022-04-01 | 山东大学 | Engine fault detection method, system, storage medium and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108593282A (en) | A kind of breaker on-line monitoring and fault diagonosing device and its working method | |
| CN109870299A (en) | A kind of breaker failure detection device and a variety of data fusion fault detection methods | |
| CN102426835B (en) | Method for identifying local discharge signals of switchboard based on support vector machine model | |
| CN103310789B (en) | A kind of sound event recognition method of the parallel model combination based on improving | |
| Kounoudes et al. | The DYPSA algorithm for estimation of glottal closure instants in voiced speech | |
| CN106646205A (en) | Random big-disturbance signal removing algorithm for analyzing circuit breaker fault through sound and vibration combination | |
| CN107993648A (en) | A kind of unmanned plane recognition methods, device and electronic equipment | |
| Manjula et al. | Comparison of empirical mode decomposition and wavelet based classification of power quality events | |
| Özseven et al. | SPeech ACoustic (SPAC): A novel tool for speech feature extraction and classification | |
| CN109360572A (en) | Call separation method, device, computer equipment and storage medium | |
| Yang et al. | Acoustics recognition of construction equipments based on LPCC features and SVM | |
| CN114325256A (en) | Power equipment partial discharge identification method, system, equipment and storage medium | |
| CN110808068A (en) | Sound detection method, device, equipment and storage medium | |
| CN104089699A (en) | Substation equipment sound reconstruction algorithm | |
| Li et al. | Speech emotion recognition using novel hht-teo based features. | |
| CN111639583A (en) | Method and system for identifying power quality disturbance of power grid | |
| CN103207931A (en) | Sub-synchronous oscillation mode identification method based on milli molar (MM) and autoregressive moving average (ARMA) algorithm | |
| CN107644651A (en) | Circuit and method for speech recognition | |
| Honggang et al. | Mechanical fault diagnosis of GIS based on MFCCs of sound signals | |
| XinXing et al. | Speech recognition based on efficient DTW algorithm and its DSP implementation | |
| CN108629087A (en) | Disturbance event model building method, device and disturbance event recognition methods, device | |
| Song et al. | Research on rolling bearing fault diagnosis method based on improved LMD and CMWPE | |
| Yang et al. | Transient fault diagnosis of track circuit based on MFCC-DTW | |
| CN115420977A (en) | Electric appliance fault detection method, training method, computer equipment and storage medium | |
| CN103778914A (en) | Anti-noise voice identification method and device based on signal-to-noise ratio weighing template characteristic matching |
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 |