CN109142523A - A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging - Google Patents
A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging Download PDFInfo
- Publication number
- CN109142523A CN109142523A CN201810922185.XA CN201810922185A CN109142523A CN 109142523 A CN109142523 A CN 109142523A CN 201810922185 A CN201810922185 A CN 201810922185A CN 109142523 A CN109142523 A CN 109142523A
- Authority
- CN
- China
- Prior art keywords
- metope
- feature
- hollowing
- data
- model
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/04—Analysing solids
- G01N29/045—Analysing solids by imparting shocks to the workpiece and detecting the vibrations or the acoustic waves caused by the shocks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/4472—Mathematical theories or simulation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/028—Material parameters
- G01N2291/0289—Internal structure, e.g. defects, grain size, texture
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- Immunology (AREA)
- General Health & Medical Sciences (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Optimization (AREA)
- Signal Processing (AREA)
- Engineering & Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Physics (AREA)
- Mathematical Analysis (AREA)
- Algebra (AREA)
- Acoustics & Sound (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging, including metope climbing robot, metope robot is by negative-pressure adsorption in metope, it is creeped by chassis driving wheel in metope, the chassis cushion different to the metope of unlike material controls frictional force, metope robot is moved according to the route of planning, is uniformly tapped after moving to the metope that need to be sampled, and following steps: 1) audio data collecting are carried out after percussion；2) preprocessed data；3) feature change of variable classification is carried out using preprocessed data；4) feature is derived；5) best model is obtained by iteration；6) best model application end is pushed to show.The invention has the following advantages: it is 1) easy to operate, avoid artificial complicated, a large amount of mathematical operation；2) quantitative digital is converted by qualitative results, hollowing nuance can also be compared to each other；3) achievement visualizes beyond the clouds, and achievement is easy-to-understand, can remotely hold a conference or consultation, decision.
Description
Technical field
The invention belongs to field of non destructive testing, and in particular to a kind of metope hollowing recognition quantitative analysis based on acoustic imaging
Method.
Background technique
According to incompletely statistics, ended for the end of the year 2016, the town-property house quantity that Zhejiang Province was completed before 2012 reaches
Building 370895, this batch of house age of dwellings that is averaged is more than 30 years, and with the growth of time, this batch of external wall of house security risk is gradually
Showing, the most common security risk is exactly that exterior wall falls off, wherein fall off with more, high-story house outdoor tile especially and influence maximum,
Ningbo in 2017, Jiaxing in 2018 all have occurred personnel death's accident, the such injures and deaths event in the whole nation also common reporter.
The omen that exterior wall falls off is exactly hollowing, can effective pre- anti-dropout generation by hollowing inspection.According to " building dress
Adorn fitting-out work inspection of quality specification " (GB50209-2002) regulation, percussion method artificial judgment can be used in metope hollowing, examines
Survey personnel tap metope, listen to echo by human ear to judge the hollowing situation at percussion position.Although human ear is listened in same position
To sound physical quantity be a constant, but treated that audition result belongs to psychologic acoustics scope again for human ear, and popular says,
One same sound, different people perceive different.Namely the same face metope, different people are different to the judgement of its hollowing
's.Easily cause erroneous judgement.
Nearly ten years, Zhejiang area multi-storey and sub high-rise residence emerges in multitude, and manually checks one to 20 to 100 meters of ceramic tile exterior walls
As all take spider-man's lanyard or hanging basket mode to be unfolded, spider-man should control rope decline, again control the left and right amplitude of oscillation, also want
Carry out metope percussion and label, risk is big, low efficiency.Although hanging basket mode reduces personnel's pendency risk, but equipment is installed, torn open
It is long, expensive to unload the period, it is difficult to promote.
In conclusion how safe and efficient acquisition taps data, how to extract the key parameter in data and establish number
Learning model and being compared, visualize is that current hollowing checks urgent problem to be solved.
Summary of the invention
Aiming at the shortcomings in the prior art, the present invention provides a kind of metope hollowing recognition quantitative analysis based on acoustic imaging
Method provides a kind of metope hollowing acquisition system and quantitative analysis, displaying side based on acoustic imaging
The invention is realized by the following technical scheme.
A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging, including metope climbing robot, metope machine
People, in metope, is creeped by chassis driving wheel in metope by negative-pressure adsorption, the chassis buffering different to the metope of unlike material
Pad control frictional force, metope robot are moved according to the route of planning, are uniformly struck after moving to the metope that need to be sampled
It hits, following steps: 1) audio data collecting is carried out after percussion；2) preprocessed data；3) characteristic quantity is carried out using preprocessed data
Transformation classification；4) feature is derived；5) best model is obtained by iteration；6) best model application end is pushed to show.
Preferably, the audio data collecting in step 1), utilizes the wheat tapped by component for being mounted on Climbing Robot
Gram elegance collection audio, is wirelessly transmitted to cloud computer.
Preferably, the preprocessed data in step 2, in collected audio, applied mathematics software MatLab is mentioned
The feature for taking loudness to count: the feature of average value, intermediate value and standard deviation and frequency domain: basic frequency, frequency spectrum entropy and mel-frequency fall
Spectral coefficient.
Preferably, carrying out feature change of variable classification using preprocessed data in step 3), before extracting feature, greatly
Most data sets require some pretreatments, including suppressing exception value and trend, fill up missing data and return to data
One changes.
Preferably, the derivation feature in step 4), using logistic regression, Linear SVM, decision tree, naive Bayesian,
One of neural network algorithm finds out best-fit algorithm, and the loudness threshold in different sections is respectively put into precipitating model and iteration
Model.
Preferably, obtaining best model by iteration in step 5), current characteristics set does not capture intrinsic in data
All variations, extract more features value, when there is over-fitting sign, by using reduction technology, be further reduced feature, borrow
The loudness eigentransformation for helping normalization etc, reduces the feature of model by the way of iteration, until can not improved model performance
Until to get arrive best model.
Preferably, best model to be pushed to application end displaying, the model generated by the above process in step 6)
Data, it will push to specified user's application end, user's application end operate in user have by oneself server or Cloud Server or
In mobile device, real-time display.
Detailed description of the invention
Fig. 1 is normal sound of the invention and hollowing sound comparison diagram.
Fig. 2 is the schematic diagram that image is tapped on the left of scanning of the invention.
Fig. 3 is the datagram of left side loudness value of the invention.
Fig. 4 is the datagram of right side loudness value of the invention.
Fig. 5 is system flow chart of the invention.
Fig. 6 is real-time display schematic diagram of the invention.
Fig. 7 creeps for the present invention and taps robot schematic diagram.
Specific embodiment
A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging, including metope climbing robot, metope machine
People, in metope, is creeped by chassis driving wheel in metope by negative-pressure adsorption, the chassis buffering different to the metope of unlike material
Pad control frictional force, metope robot are moved according to the route of planning, are uniformly struck after moving to the metope that need to be sampled
It hits, following steps: 1) audio data collecting is carried out after percussion；2) preprocessed data；3) characteristic quantity is carried out using preprocessed data
Transformation classification；4) feature is derived；5) best model is obtained by iteration；6) best model is pushed to application end to show, step
1) audio data collecting in acquires audio, wireless transmission using the microphone tapped by component for being mounted on Climbing Robot
To cloud computer, preprocessed data in step 2, in collected audio, applied mathematics software MatLab is extracted and is rung
Spend the feature of statistics: the feature of average value, intermediate value and standard deviation and frequency domain: basic frequency, frequency spectrum entropy and mel-frequency cepstrum system
Number, which is characterized in that carry out feature change of variable classification using preprocessed data in step 3), before extracting feature, mostly
Number data sets require some pretreatments, including suppressing exception value and trend, fill up missing data and carry out normalizing to data
Change, the derivation feature in step 4), using logistic regression, Linear SVM, decision tree, naive Bayesian, neural network algorithm it
One, best-fit algorithm is found out, the loudness threshold in different sections is respectively put into precipitating model and iterative model, in step 5)
Best model is obtained by iteration, current characteristics set does not capture all variations intrinsic in data, more features value is extracted, when
Occur over-fitting sign, by use reduction technology, be further reduced feature, by normalization etc loudness eigentransformation,
The feature that model is reduced by the way of iteration, until can not improved model performance until to get arriving best model, in step 6)
Best model is pushed to application end displaying, by the above process generate model data, it will push to specified user
Application end, user's application end operate in user and have by oneself in server or Cloud Server or mobile device, real-time display, creeper
During device people creeps on metope, the loudness value of each position can calculate the hollowing degree of the position.
In order to more intuitively show metope hollowing distribution situation, metope is divided into the detection zone of several homalographics,
It surveys in area and is repeatedly beaten, form one group of loudness numerical value X1、X2、X3……Xn, arithmetic mean of instantaneous value μ passes through above data
Calculate the meansquaredeviationσ of the lot number value.
System includes metope climbing robot, knocking device and adopts acoustic transmission system, acoustic wave analysis, identifies software.Machine
People, in metope, is creeped by chassis driving wheel in metope by negative-pressure adsorption, the chassis buffering different to the metope of unlike material
Pad control frictional force.Machine is moved according to the scheme of planning, is uniformly tapped after moving to the metope that need to be sampled, and is tapped
Sound, to PC collection terminal, is extracted its audio frequency characteristics through software, uploads to MATLAB Parallel in real time by sensor transmissions
Cloud, using the powerful concurrent operation ability in cloud, in conjunction with the machine learning function of MATLAB software, by the various spies of sound
Sign amount is filtered, feature selecting and transformation carry out model analysis, obtains metope hollowing data model, and show cloud
User.
Loudness, tone and the tone color of sound are called the three elements of auditory perceptual sound, these three features all with the vibration of sound source
Move it is related, reflection be sound different characteristic physical quantity.This is because hollowing metope rear portion is there are air buffer, by
After equal energy impact, frequency, the amplitude all difference of the vibration of hollowing position are exactly that loudness is poor in reflection to human auditory system
It is different.As long as we obtain the loudness parameter of beating position, that is, human auditory system judgement has been carried out digitized description, pass through ratio
Just reach compared with loudness size and has tapped determination requirement as defined in specification.
Pass through the experiment to different materials metope, it is seen that hollowing position loudness value is also big.If adopting AUDITION CC software point
One file comprising knock twice of analysis, as shown in Figure 1.
Left side is normal sound, and right side is hollowing sound, chooses a period of time on a timeline, which must envelope single
Frequency field figure is beaten, it is specific such as Fig. 2.
First scanning left side taps image, and acoustic segment is then moved to left side percussion image and is scanned, left and right two is obtained
As a result the loudness value of a sound is shown in such as Fig. 3 and such as Fig. 4.
Loudness value range may determine that the position -13.01LUFS is hollowing position in -26.31LUFS < -13.01LUFS
It sets.
During climbing robot is creeped on metope, the loudness value of each position can calculate the hollowing journey of the position, be
Metope, is divided into the detection zone of several homalographics, controlled in the present embodiment by more intuitive display metope hollowing distribution situation
It is made as 1 meter * 1 meter of test section, is repeatedly beaten surveying in area, one group of loudness numerical value X is formed1、X2、X3……Xn, arithmetic
Average value is μ, and the meansquaredeviationσ of the lot number value is calculated by above data.
Meansquaredeviationσ, definition are the square roots of overall constituent parts standard value and the arithmetic average of its average deviation square.
Dispersion degree in its reflection group between individual.Formula is as follows:
Meansquaredeviationσ is a kind of measurement of loudness average value degree of scatter.One biggish mean square deviation represents most of numerical value and its
It differs greatly between average value；One lesser mean square deviation represents these numerical value and is closer to average value.
By the sequence of the two data, following two information is provided for an area Ge Ce, one, average value represent the region
Hollowing average level, also reflection survey the current hollowing rate in area；Two, mean square deviation represents the region hollowing dispersion degree, and also reflection is surveyed
Area's hollowing dispersion.
Computer arranges long-range real-time analyzer beyond the clouds, accomplishes to acquire in real time, analyze in real time, real-time exhibition, process
It is as follows:
Audio data collecting；
Audio is acquired using the microphone tapped by component for being mounted on Climbing Robot, is wirelessly transmitted to cloud computer.
Preprocessed data；
In collected audio, using the business mathematics software MatLab that MathWorks company of the U.S. produces, extract with lower class
The feature of type:
Loudness statistics: average value, intermediate value and standard deviation；
Frequency domain: basic frequency, frequency spectrum entropy and mel-frequency cepstrum coefficient (MFCC)；
3, feature change of variable classification is carried out using preprocessed data；
Before extracting feature, most of data sets require some pretreatments, typical mission include suppressing exception value and trend,
It fills up missing data and data is normalized.
Extract character extraction be one of most important part of machine learning because it initial data is transformed into it is suitable
The information of machine learning algorithm.Feature extraction eliminates the redundancy phenomena in all kinds of measurement data, facilitates the general of study stage
Change.Extensive avoided to specific sample model over-fitting.
Derive feature；
Deriving feature (also referred to as Feature Engineering or feature extraction) is one of part mostly important in machine learning.This process can
Convert raw data into the information that machine learning algorithm can be used.Generally use logistic regression (and Linear SVM), decision
Tree, naive Bayesian, neural network scheduling algorithm, find out best-fit algorithm, and it is heavy that the loudness threshold in different sections is respectively put into
Shallow lake model and iterative model.
Best model is obtained by iteration；
If current characteristics set does not capture all variations intrinsic in data, that just needs to extract the feature that more comes in handy
Value.It, can be by using reduction technology, such as principal component analysis (PCA), linear discriminant analysis if seeing over-fitting sign
(LDA) or singular value decomposition (SVD), it is further reduced feature.Loudness feature changes very extensively on range scale, then can borrow
Help the eigentransformation of normalization etc.The feature that model can be reduced by the way of iteration, until can not improved model performance
Until, it is best model.
Model is pushed to application end to show；
The model data generated by the above process, it will pushing to specified user's application end (it is own to may operate in user
Server, on Cloud Server or in mobile device), real-time display is specific as follows.
Validity by test, this method needs to pay attention at 3 points, firstly, perpend wall needs to carry out independent mathematics point
Analysis, because of the musical instrument that different metopes can regard different personnel as, be formed using different materials, a same loudness value can not
Represent different musical instruments.But perpend wall can regard same a group of people as, same gimmick be taken with same material, in the close time
The musical instrument of interior construction, it is relatively scientific by verification unit of single side.A kind of its pounder of secondary use simultaneously controls impact strength and angle
Degree, just can guarantee that impact surface receives the constancy of energy in this way.Finally adopting acoustic device answers relative sound source to fix, and is avoided that road in this way
The error that diameter difference generates.
Protection scope of the present invention includes but is not limited to embodiment of above, and protection scope of the present invention is with claims
Subject to, replacement, deformation, the improvement that those skilled in the art that any pair of this technology is made is readily apparent that each fall within of the invention
Protection scope.
Claims (7)
1. a kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging, which is characterized in that including metope climbing robot,
Metope robot, in metope, is creeped by chassis driving wheel in metope by negative-pressure adsorption, different to the metope of unlike material
Chassis cushion controls frictional force, and metope robot is moved according to the route of planning, and it is laggard to move to the metope that need to be sampled
Row uniformly taps, and carries out following steps: 1) audio data collecting after percussion；2) preprocessed data；3) using preprocessed data into
Row feature change of variable classification；4) feature is derived；5) best model is obtained by iteration；6) best model is pushed into application end exhibition
Show.
2. a kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging according to claim 1, which is characterized in that
Audio data collecting in step 1) acquires audio using the microphone tapped by component for being mounted on Climbing Robot, wirelessly
It is transmitted to cloud computer.
3. a kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging according to claim 1, which is characterized in that
Preprocessed data in step 2, in collected audio, applied mathematics software MatLab extracts the feature of loudness statistics:
The feature of average value, intermediate value and standard deviation and frequency domain: basic frequency, frequency spectrum entropy and mel-frequency cepstrum coefficient.
4. a kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging according to claim 1, which is characterized in that
Feature change of variable classification is carried out using preprocessed data in step 3), before extracting feature, most of data sets are required
Some pretreatments, including suppressing exception value and trend, fill up missing data and data are normalized.
5. a kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging according to claim 1, feature exist
In, derivation feature in step 4), using logistic regression, Linear SVM, decision tree, naive Bayesian, neural network algorithm it
One, best-fit algorithm is found out, the loudness threshold in different sections is respectively put into precipitating model and iterative model.
6. a kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging according to claim 1, which is characterized in that
Best model is obtained by iteration in step 5), all variations intrinsic in data is captured, extracts more features value, work as appearance
Over-fitting sign is further reduced feature, by the loudness eigentransformation of normalization etc, uses by using reduction technology
The mode of iteration reduces the feature of model, until can not improved model performance until to get arrive best model.
7. a kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging according to claim 1, which is characterized in that
Best model is pushed to application end displaying, the model data generated by the above process, it will push to finger in step 6)
Fixed user's application end, user's application end operate in user and have by oneself in server or Cloud Server or mobile device, aobvious in real time
Show.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810922185.XA CN109142523A (en) | 2018-08-14 | 2018-08-14 | A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810922185.XA CN109142523A (en) | 2018-08-14 | 2018-08-14 | A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109142523A true CN109142523A (en) | 2019-01-04 |
Family
ID=64793334
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810922185.XA Pending CN109142523A (en) | 2018-08-14 | 2018-08-14 | A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109142523A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110441394A (en) * | 2019-08-08 | 2019-11-12 | 上海市建筑科学研究院 | Existing external thermal insulation system detection method based on acoustic index |
| CN111683442A (en) * | 2020-06-08 | 2020-09-18 | 上海添彩灯饰照明有限公司 | External wall light illumination control method and system |
| CN111783616A (en) * | 2020-06-28 | 2020-10-16 | 北京瓦特曼科技有限公司 | Data-driven self-learning-based nondestructive testing method |
| CN112034044A (en) * | 2020-09-01 | 2020-12-04 | 南京邮电大学 | A neural network-based intelligent identification and detection device for tile hollow |
| CN113671031A (en) * | 2021-08-20 | 2021-11-19 | 北京房江湖科技有限公司 | Wall hollowing detection method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246146A (en) * | 2008-03-07 | 2008-08-20 | 厦门大学 | Portable wall empty drum automatic detector |
| CN101883970A (en) * | 2007-11-06 | 2010-11-10 | 南非科学和工业研究会 | Method and apparatus for assessing rock mass integrity |
| CN203720150U (en) * | 2013-12-30 | 2014-07-16 | 上海市房地产科学研究院 | Device for detecting bulging defects on outer wall face of building |
| CN105015640A (en) * | 2015-06-30 | 2015-11-04 | 长安大学 | Wall surface detecting rescue robot and control method thereof |
-
2018
- 2018-08-14 CN CN201810922185.XA patent/CN109142523A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101883970A (en) * | 2007-11-06 | 2010-11-10 | 南非科学和工业研究会 | Method and apparatus for assessing rock mass integrity |
| CN101246146A (en) * | 2008-03-07 | 2008-08-20 | 厦门大学 | Portable wall empty drum automatic detector |
| CN203720150U (en) * | 2013-12-30 | 2014-07-16 | 上海市房地产科学研究院 | Device for detecting bulging defects on outer wall face of building |
| CN105015640A (en) * | 2015-06-30 | 2015-11-04 | 长安大学 | Wall surface detecting rescue robot and control method thereof |
Non-Patent Citations (2)
| Title |
|---|
| MATHWORKS: ""精通机器学习：MATLAB分步实施指南"", 《MATHWORKS HTTPS://CH.MATHWORKS.COM/VIDEOS/NAVIGATING-BIG-DATA-WITH-MATLAB-1510052363443.HTML》 * |
| 周志琨 等: ""壁面粘贴质量检测系统设计"", 《上海交通大学学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110441394A (en) * | 2019-08-08 | 2019-11-12 | 上海市建筑科学研究院 | Existing external thermal insulation system detection method based on acoustic index |
| CN111683442A (en) * | 2020-06-08 | 2020-09-18 | 上海添彩灯饰照明有限公司 | External wall light illumination control method and system |
| CN111683442B (en) * | 2020-06-08 | 2022-11-04 | 上海添彩灯饰照明有限公司 | External wall light illumination control method and system |
| CN111783616A (en) * | 2020-06-28 | 2020-10-16 | 北京瓦特曼科技有限公司 | Data-driven self-learning-based nondestructive testing method |
| CN111783616B (en) * | 2020-06-28 | 2024-03-26 | 北京瓦特曼科技有限公司 | Nondestructive testing method based on data-driven self-learning |
| CN112034044A (en) * | 2020-09-01 | 2020-12-04 | 南京邮电大学 | A neural network-based intelligent identification and detection device for tile hollow |
| CN113671031A (en) * | 2021-08-20 | 2021-11-19 | 北京房江湖科技有限公司 | Wall hollowing detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109142523A (en) | A kind of metope hollowing recognition quantitative analytic approach based on acoustic imaging | |
| CN102163427B (en) | Method for detecting audio exceptional event based on environmental model | |
| EP1536414B1 (en) | Method and apparatus for multi-sensory speech enhancement | |
| Ren et al. | Prediction of rock compressive strength using machine learning algorithms based on spectrum analysis of geological hammer | |
| US7499686B2 (en) | Method and apparatus for multi-sensory speech enhancement on a mobile device | |
| CN102027338B (en) | Signal judgment method, signal judgment apparatus, and signal judgment system | |
| Hollosi et al. | Enhancing wireless sensor networks with acoustic sensing technology: use cases, applications & experiments | |
| CN112985574A (en) | High-precision classification identification method for optical fiber distributed acoustic sensing signals based on model fusion | |
| CN109034088A (en) | A kind of unmanned plane signal detection method and device | |
| CN105916090A (en) | Hearing aid system based on intelligent speech recognition technology | |
| CN110333484B (en) | Indoor area-level localization method based on ambient background sound perception and analysis | |
| CN102346948A (en) | Circumference invasion detection method and system | |
| CN115376526A (en) | A power equipment fault detection method and system based on voiceprint recognition | |
| CN111554325B (en) | Voice recognition method and system | |
| Kanisha et al. | Speech recognition with advanced feature extraction methods using adaptive particle swarm optimization | |
| CN112863515B (en) | Method for recognizing human tongue and jaw movement using mobile phone speaker and microphone | |
| Mikhailova et al. | Software for estimating of a premises acoustic security | |
| Qadri et al. | Comparative Analysis of Gender Identification using Speech Analysis and Higher Order Statistics | |
| Zhang et al. | Geological big data acquisition based on speech recognition | |
| CN105376409B (en) | A kind of piano tone tuning system based on mobile phone terminal | |
| Cheong et al. | Active acoustic scene monitoring through spectro-temporal modulation filtering for intruder detection | |
| Balamurali et al. | Automated Classification of Vowel-Gesture Parameters Using External Broadband Excitation. | |
| CN115862636B (en) | Internet man-machine verification method based on voice recognition technology | |
| Wang et al. | Scalable Machine Learning Approach to Classifying Transportation Noise at Two Urban Sites in Greater Boston, Massachusetts | |
| CN119322312B (en) | A method and system for locating persons trapped in fire |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190104 |
|
| RJ01 | Rejection of invention patent application after publication |