CN108182950A - The abnormal sound in public places feature decomposition and extracting method of improved experience wavelet transformation - Google Patents
The abnormal sound in public places feature decomposition and extracting method of improved experience wavelet transformation Download PDFInfo
- Publication number
- CN108182950A CN108182950A CN201711462639.1A CN201711462639A CN108182950A CN 108182950 A CN108182950 A CN 108182950A CN 201711462639 A CN201711462639 A CN 201711462639A CN 108182950 A CN108182950 A CN 108182950A
- Authority
- CN
- China
- Prior art keywords
- voice signal
- abnormal sound
- public places
- original anomaly
- experience wavelet
- 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.)
- Granted
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 48
- 238000000034 method Methods 0.000 title claims abstract description 26
- 230000009466 transformation Effects 0.000 title claims abstract description 21
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 12
- 238000001228 spectrum Methods 0.000 claims abstract description 10
- 238000000605 extraction Methods 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 3
- 238000006467 substitution reaction Methods 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000005236 sound signal Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 239000002360 explosive Substances 0.000 description 2
- 239000011521 glass Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 206010011224 Cough Diseases 0.000 description 1
- 230000002547 anomalous effect Effects 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000192 social effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
Classifications
-
- 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
- G10L15/00—Speech recognition
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Soundproofing, Sound Blocking, And Sound Damping (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The present invention proposes the abnormal sound in public places feature decomposition and extracting method of improved experience wavelet transformation, specifically the Fourier spectrum of original anomaly voice signal is divided by the method for analog equivalent rectangular bandwidth scale, the equivalent rectangular bandwidth experience wavelet transform filter group of fixed boundary is obtained according to the boundary after division, with the wave filter component solution original anomaly voice signal, its different frequency component is obtained；The different frequency component of original anomaly voice signal is calculated relative to the Energy-Entropy of original anomaly voice signal, using the Energy-Entropy after normalized as the feature vector of original anomaly voice signal.
Description
Technical field
The invention belongs to audio signal characteristic extraction and identification technology fields.
Background technology
Public place such as square, bus stop, subway etc. have the characteristics that crowd massing is intensive, region is wide, and public
The safety precaution in place is constantly subjected to national governments and the extensive concern of the people.Monitoring technology based on video monitoring at present
Safety precaution for public place plays positive effect, however Video Supervision Technique has monitoring dead angle, monitoring rainy days
The problems such as fuzzy.It is well known that shriek, shot, glass breaking sound, explosive sound etc. are frequently accompanied by when anomalous event occurs
Abnormal sound, therefore the cooperating operation of Voice Surveillance and video monitoring has become the development side in public place security monitoring field
To.At present, video monitoring is had been widely used in reality relative to Voice Surveillance comparative maturity, and existing audio is supervised
Control system only comprising simple sound collection, transmission etc., lacks effective identification to abnormal sound.It is public involved by this patent
Place exceptional sound recognition technology is the core technology of Audio Monitoring System.Therefore, have to the research of the technology prior
Social effect and researching value.
Experience wavelet transformation (Empirical Wavelet Transform, EWT) handles abnormal sound in public places feature
During extraction there are the problem of：1. experience wavelet transformation is made by the minimum value between detecting the adjacent maximum of signal Fourier spectrum
One group of wave filter group is built for boundary, and with such one group of boundary, as shown in Figure 1.This detection signal Fourier spectrum extreme value
The method of point is suitable for the simple signal of frequency content, and after signal adds certain noise, testing result is by larger
It influences.And abnormal sound in public places is made of two parts：Abnormal sound signal and ambient noise signal.Abnormal sound signal
Fourier spectrum is inherently containing more extreme point, and due to the presence of ambient noise, so when using EWT to public field
When institute's abnormal sound signal is decomposed, since the extreme point of the Fourier spectrum of signal is complex, lead to the filter constructed
Wave device component cloth is comparatively concentrated, so that the distribution of wave filter group constructed is equally more concentrated.When using this
When abnormal sound in public places is identified in method, undesirable result has been obtained.2. and since public place background is made an uproar
Sound is more complicated, and the signal decomposition of same type can be caused to obtain result there are larger differences, can equally obtain less desirable point
Class result.Simultaneously as being vulnerable to the interference of noise, the reconstructed error of signal is larger.
Invention content
For more than the deficiencies in the prior art, the present invention proposes a kind of public place based on improved experience wavelet transformation
Abnormal sound feature decomposition and extraction, with improved experience wavelet transformation (Developed Empirical Wavelet
Transform, DEWT) abnormal sound in public places is decomposed, by introducing equivalent rectangular bandwidth in this process
(Equivalent Rectangular Bandwidth, ERB) solves EWT and is applied to abnormal sound in public places feature extraction
Key theory and technical barrier obtain feature of the abnormal sound in public places under different scale.
The present invention proposes that the abnormal sound in public places feature decomposition of improved experience wavelet transformation includes with extracting method
Following steps：
Step 1, to the Fourier spectrum of original anomaly voice signal by the method for analog equivalent rectangular bandwidth scale into
Row divides, and obtains the equivalent rectangular bandwidth experience wavelet transform filter group of fixed boundary according to the boundary after division, use is such
Rectangular bandwidth experience wavelet transform filter component solution original anomaly voice signal is imitated, obtains its different frequency component
Step 2 calculates the different frequency component of original anomaly voice signal relative to the energy of original anomaly voice signal
Entropy, using the Energy-Entropy after normalized as the feature vector of original anomaly voice signal.
The specific method of step 1 is：
(1) the centre frequency f of equivalent rectangular bandwidth wave filter is calculatedc, enable the center of experience wavelet transformation median filter frequently
Rate fcnEqual to the centre frequency f of the equivalent rectangular bandwidth wave filterc。
(2) experience wavelet transformation median filter up-and-down boundary calculating formula f is utilizedcn=(wn+wn+1)/2 obtain a series of
Boundary value wn, by boundary value wnIn the low-pass filter and bandpass filter group of substitution experience Construction of Wavelets, equivalent rectangular band is formed
Wide experience wavelet transform filter group.
(3) by each original anomaly voice signal by the equivalent rectangular bandwidth experience wavelet transform filter group into
Row decomposes, and obtains the different frequency component f of original anomaly voice signalj。
The specific method of step 2 is：
By improved experience wavelet transform filter group for decomposing abnormal sound in public places, obtain a series of different
Mode fk.Define Energy-Entropy HjAs abnormal sound in public places feature vectorIn formula, HjRepresent energy
Entropy, parameter εj=Ej/ E, E represent the energy of original anomaly voice signal, Ej=| fj|2, represent j-th of frequency component fjEnergy
Amount.Energy-Entropy of the different frequency component for original anomaly voice signal of signal is calculated, and is combined into feature vector and is returned
One change is handled, the feature vector as original signal.
The advantage of the invention is that：
1st, the present invention fully considers the key theory and skill for facing EWT methods applied to abnormal sound in public places identification
Art problem, the characteristics of in combination with abnormal sound and ambient noise respectively, research EWT is decomposed existing for abnormal sound in public places
The problem of discrimination is low by simulating ERB scales, forms improved experience wavelet transform filter group.
2nd, the present invention is a kind of method for being more suitable for the description of abnormal sound in public places signal characteristic, and the thought of this method is
A wave filter group is established, using improved experience wavelet transform filter group, obtains the different frequencies point of a series of signal
Amount, then using the different frequency component of signal relative to the Energy-Entropy of original signal as feature vector, be normalized
Afterwards as the feature vector of original signal, thus extraction obtains abnormal sound in public places.
Description of the drawings
Fig. 1：Signal Fourier spectrum is divided and wave filter group establishes schematic diagram；
Fig. 2：ERB-EWT wave filter class frequency response diagrams；
Fig. 3：Shriek decomposition result figure；
Fig. 4：The present invention proposes improved experience Feature Extraction of Wavelet Transform flow diagram；
Fig. 5：The present invention proposes abnormal sound in public places identification process block diagram.
Specific embodiment
Below by taking the whole process of abnormal sound in public places identification as an example, with reference to the attached drawing think of that the present invention is further explained
Think：
As shown in figure 5, abnormal sound in public places identification process block diagram mainly includes two parts：Abnormal sound in public places
Feature extraction and abnormal sound in public places identification are carried out, wherein feature extraction is using the present invention is based on improved experiences
The method of wavelet transformation obtains.
First, abnormal sound in public places feature extraction
1. simulate the boundary that ERB scales obtain wave filter
Experience wavelet transformation is a kind of method for analyzing nonlinear and non local boundary value problem, if time domain discrete signal is f, EWT will
Boundary of the minimum value as wave filter between the continuous maximum of the Fourier spectrum of signal, boundary result note wn, n is whole
Number, n ∈ [0, N].Assuming that signal is made of N number of mode, then Fourier spectrum range [0, π] can be divided into N number of continuous section,
Then it needs to be determined that number of boundary be N-1.w0=0, wn=π, each divided section is denoted as Λn=[wn,wn+1].SoObtain boundary wnLater, with wnCentered on 2 τnRegion on build wave filter experience scaling function and warp
Test wavelet function.As shown in Figure 1, it can obtain a low-pass filterWith the bandpass filter of N-1So as in frequency
Domain forms one group of wave filter group.One low-pass filter and N-1 bandpass filter.However original experience small echo is utilized to become
It is low there are discrimination when changing method processing abnormal sound, the shortcomings that decomposition easily by noise jamming.By reading document, find to draw
Entering ERB scales can preferably solve the problems, such as.
ERB scales can accurately describe the frequency selective characteristic of human ear basilar memebrane, can preferably reflect voice signal
Essence.Moore points out that the centre frequency of human auditory system wave filter only covers 50-15000Hz.Public place abnormal sound in this example
The sample frequency of sound is 16KHz, and frequency range is in 0-8KHz.
The centre frequency f of human auditory system wave filtercRelationship with the scale r of human auditory system wave filter ERB is：
The calculating formula of the centre frequency of human auditory system wave filter is：
Since the frequency domain of abnormal sound is [0,8KHz], then ERB scales r can use to integer 31.Calculate equivalent rectangular
The centre frequency f of bandwidth filterc, pass through the centre frequency and equivalent rectangular bandwidth of the wave filter in experience wavelet transformation
The centre frequency f of auditory filtercnIt is equal；
I.e.：fcn=fc (3)
Centre frequency is calculated by up-and-down boundary by wave filter using in experience wavelet transformation, to be calculated one
The boundary value w of seriesn。
fcn=(wn+wn+1)/2 (4)
Thus it calculates and obtains a series of boundary value wn, this value is updated in the wave filter group of experience Construction of Wavelets.
First wave filter is low-pass filter in the wave filter group of structure, remaining is bandpass filter.Since EWT builds wave filter group
Beginning boundary for 0Hz, terminations boundary is 8KHz, covers the entire frequency range of abnormal sound.
Wherein low-pass filter is：
Bandpass filter is：
Wherein τn=γ ωn, function β (x) is an arbitrary Ck([0,1]) function.β (x)=x4(35-84x+70x2-
20x3).WhenWhen,In L2(R) compact schemes frame is spatially constituted.
Approximation coefficient can be obtained by the inner product of signal and experience scaling function
Detail coefficients can be obtained by the inner product of signal and experience wavelet function
Wherein,WithIt is φ respectively1(t) and ψn(t) Fourier transformation.
For this purpose, reconstruction signal is obtained by formula (7)：
Wherein, * represents convolution,WithIt is respectivelyWithFourier transformation.
2. improved experience wavelet transform filter group
By simulating ERB scales, improved experience wavelet transform filter group is formed, constructs wave filter group such as Fig. 2 institutes
Show, the improved experience wavelet transform filter group of pending signal, signal f is decomposed and obtains empirical modal fk, such as
Shown in formula (8).
By taking shriek as an example, shriek is decomposed, and result is carried out using improved experience wavelet transform filter group
Signal, decomposition result are as shown in Figure 3.
3. feature extraction
Improved experience wavelet transform filter group be used for abnormal sound in public places feature extraction flow as shown in figure 4,
By improved experience wavelet transform filter group for decomposing abnormal sound in public places, a series of different mode f are obtainedk。
The energy of original signal is E, j-th of mode fjCorresponding energy is Ej=| fj|2.Enable εj=Ej/ E defines Energy-Entropy HjAs
Abnormal sound in public places feature vectorThe different frequency component of signal is calculated for original anomaly sound
The Energy-Entropy of sound signal, and be combined into feature vector and be normalized, the feature vector as original signal.
2nd, abnormal sound in public places identifies
Two steps are divided into abnormal sound in public places identification：
1. establish disaggregated model：Establish abnormal sound database, the abnormal sound database include explosive sound, shriek,
(similar normal sound includes ATM voice sound, bank calls out the numbers for shot, glass breaking sound and similar five class of normal sound
Sound, boof, brouhaha, cough, switch gate sound, knock, Conversation Voice), half quantity conduct is randomly selected from every class
Training sample, another half-sample are distinguished as test sample to be identified with aforementioned abnormal sound in public places extracting method
The feature vector of training sample and test sample is asked for, support vector machines (Support is established according to the feature vector of training sample
Vector Machine, SVM) disaggregated model.
2. Classification and Identification：The feature vector obtained using training sample trains svm classifier model, and model is tested according to this
Sample to be tested identifies the classification of test sample abnormal sound, exports the classification results per class abnormal sound, is finally known
Other result.
Claims (3)
1. the abnormal sound in public places feature decomposition and extracting method of improved experience wavelet transformation, which is characterized in that including
Step is as follows：
Step 1 draws the Fourier spectrum of original anomaly voice signal by the method for analog equivalent rectangular bandwidth scale
Point, the equivalent rectangular bandwidth experience wavelet transform filter group of fixed boundary is obtained according to the boundary after division, with the equivalent square
Shape bandwidth experience wavelet transform filter component solution original anomaly voice signal, obtains its different frequency component；
Step 2, calculate original anomaly voice signal different frequency component relative to original anomaly voice signal Energy-Entropy, and
It is combined into feature vector to be normalized, the feature vector as original anomaly voice signal.
2. the abnormal sound in public places feature decomposition of improved experience wavelet transformation according to claim 1 and extraction side
Method, which is characterized in that the specific method of step 1 is：
(1) the centre frequency f of equivalent rectangular bandwidth wave filter is calculatedc, enable the centre frequency f of experience wavelet transformation median filtercn
Equal to the centre frequency f of equivalent rectangular bandwidth wave filterc；
(2) experience wavelet transformation median filter up-and-down boundary calculating formula f is utilizedcn=(wn+wn+1)/2 obtain a series of boundary
Value wn, by boundary value wnIn the low-pass filter and bandpass filter group of substitution experience Construction of Wavelets, composition equivalent rectangular bandwidth warp
Test wavelet transform filter group；
(3) each original anomaly voice signal is divided by the equivalent rectangular bandwidth experience wavelet transform filter group
Solution, obtains the different frequency component f of original anomaly voice signalj。
3. the abnormal sound in public places feature decomposition and extracting method of the small transformation of improved experience according to claim 1,
It is characterized in that, step 2 specifically includes：
(1) improved experience wavelet transform filter group is obtained a series of different for decomposing abnormal sound in public places
Mode fk；
(2) Energy-Entropy H is definedjAs abnormal sound in public places feature vectorIn formula, HjRepresent energy
Entropy, parameter εj=Ej/ E, E represent the energy of original anomaly voice signal, Ej=| fj|2, represent j-th of frequency component fjEnergy
Amount；
(3) calculate signal different frequency component for original anomaly voice signal Energy-Entropy, and be combined into feature vector into
Row normalized, the feature vector as original anomaly voice signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711462639.1A CN108182950B (en) | 2017-12-28 | 2017-12-28 | Improved method for decomposing and extracting abnormal sound characteristics of public places through empirical wavelet transform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711462639.1A CN108182950B (en) | 2017-12-28 | 2017-12-28 | Improved method for decomposing and extracting abnormal sound characteristics of public places through empirical wavelet transform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108182950A true CN108182950A (en) | 2018-06-19 |
| CN108182950B CN108182950B (en) | 2021-05-28 |
Family
ID=62548499
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711462639.1A Active CN108182950B (en) | 2017-12-28 | 2017-12-28 | Improved method for decomposing and extracting abnormal sound characteristics of public places through empirical wavelet transform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108182950B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109034127A (en) * | 2018-08-31 | 2018-12-18 | 中国电子科技集团公司第三十六研究所 | A kind of frequency spectrum method for detecting abnormality, device and electronic equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1207170A (en) * | 1996-09-13 | 1999-02-03 | 株式会社东芝 | Unsteady signal analyzer and medium for recording unsteady signal analyzer program |
| US20040083094A1 (en) * | 2002-10-29 | 2004-04-29 | Texas Instruments Incorporated | Wavelet-based compression and decompression of audio sample sets |
| CN103366734A (en) * | 2012-03-31 | 2013-10-23 | 佳能株式会社 | Method and device for checking voice recognition results, voice recognition system and audio monitoring system |
| US20160078883A1 (en) * | 2013-04-26 | 2016-03-17 | Nec Corporation | Action analysis device, action analysis method, and action analysis program |
| CN106228979A (en) * | 2016-08-16 | 2016-12-14 | 重庆大学 | A kind of abnormal sound in public places feature extraction and recognition methods |
| CN106683687A (en) * | 2016-12-30 | 2017-05-17 | 杭州华为数字技术有限公司 | Abnormal voice classifying method and device |
| CN107393555A (en) * | 2017-07-14 | 2017-11-24 | 西安交通大学 | A kind of detecting system and detection method of low signal-to-noise ratio abnormal sound signal |
-
2017
- 2017-12-28 CN CN201711462639.1A patent/CN108182950B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1207170A (en) * | 1996-09-13 | 1999-02-03 | 株式会社东芝 | Unsteady signal analyzer and medium for recording unsteady signal analyzer program |
| US20040083094A1 (en) * | 2002-10-29 | 2004-04-29 | Texas Instruments Incorporated | Wavelet-based compression and decompression of audio sample sets |
| CN103366734A (en) * | 2012-03-31 | 2013-10-23 | 佳能株式会社 | Method and device for checking voice recognition results, voice recognition system and audio monitoring system |
| US20160078883A1 (en) * | 2013-04-26 | 2016-03-17 | Nec Corporation | Action analysis device, action analysis method, and action analysis program |
| CN106228979A (en) * | 2016-08-16 | 2016-12-14 | 重庆大学 | A kind of abnormal sound in public places feature extraction and recognition methods |
| CN106683687A (en) * | 2016-12-30 | 2017-05-17 | 杭州华为数字技术有限公司 | Abnormal voice classifying method and device |
| CN107393555A (en) * | 2017-07-14 | 2017-11-24 | 西安交通大学 | A kind of detecting system and detection method of low signal-to-noise ratio abnormal sound signal |
Non-Patent Citations (5)
| Title |
|---|
| A.K.M FAZLUL HAQUE: "FFT and Wavelet-Based Feature Extraction for Acoustic Audio Classification", 《INTERNATIONAL JOURNAL OF ADVANCE INNOVATIONS, THOUGHTS & IDEAS》 * |
| FEI CHEN ET AL: "Fault diagnosis of motorized spindle via modified empirical wavelet transform-kernel PCA and optimized support vector machine", 《JVE JOURNALS》 * |
| 李伟红等: "改进的ESMD用于公共场所异常声音特征提取", 《仪器仪表学报》 * |
| 栾少文: "智能监控系统中公共场所异常声音检测的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 栾少文等: "公共场所典型异常声音的特征提取", 《计算机工程》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109034127A (en) * | 2018-08-31 | 2018-12-18 | 中国电子科技集团公司第三十六研究所 | A kind of frequency spectrum method for detecting abnormality, device and electronic equipment |
| CN109034127B (en) * | 2018-08-31 | 2022-04-05 | 中国电子科技集团公司第三十六研究所 | Frequency spectrum anomaly detection method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108182950B (en) | 2021-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109616140B (en) | Abnormal sound analysis system | |
| CN106909784B (en) | Epileptic electroencephalogram (eeg) identification device based on two-dimentional time-frequency image depth convolutional neural networks | |
| CN114201988B (en) | Satellite navigation composite interference signal identification method and system | |
| CN110534118A (en) | Transformer/reactor method for diagnosing faults based on Application on Voiceprint Recognition and neural network | |
| CN107305774A (en) | Speech detection method and device | |
| CN110490071A (en) | A kind of substation's Abstraction of Sound Signal Characteristics based on MFCC | |
| CN106782505A (en) | A kind of method based on electric discharge voice recognition high-tension switch cabinet state | |
| CN106683687B (en) | Abnormal sound classification method and device | |
| CN106874833A (en) | A kind of mode identification method of vibration event | |
| CN110398647B (en) | Transformer state monitoring method | |
| CN110636048B (en) | Vehicle-mounted intrusion detection method and system based on ECU signal characteristic identifier | |
| CN106546818A (en) | A kind of harmonic signal detection method based on DNL Mode Decomposition | |
| CN110718235A (en) | Abnormal sound detection method, electronic device and storage medium | |
| CN103048593A (en) | Identification method of insulation defect type of gas-insulated switchgear | |
| Sharan et al. | Cochleagram image feature for improved robustness in sound recognition | |
| CN107037486A (en) | The Time-frequency Spectrum Analysis method and system of earth natural pulses electromagnetic field data processing | |
| CN106991773A (en) | A kind of safety-protection system vibration signal recognition based on S-transformation feature extraction | |
| CN108836322B (en) | Naked eye 3D display vision-induced motion sickness detection method | |
| Kotani et al. | Unsupervised scanning behavior detection based on distribution of network traffic features using robust autoencoders | |
| CN106019102A (en) | Signal de-noising method and apparatus | |
| CN114742096A (en) | Intrusion alarm method and system based on vibration optical fiber detection and complete action extraction | |
| CN113707175B (en) | Acoustic event detection system based on feature decomposition classifier and adaptive post-processing | |
| CN108182950A (en) | The abnormal sound in public places feature decomposition and extracting method of improved experience wavelet transformation | |
| CN110146922A (en) | The double seismometer interference identification methods of high speed rail earthquake pre-warning system list | |
| CN111933186B (en) | Method, device and system for fault identification of on-load tap-changer |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |