CN106841403A - A kind of acoustics glass defect detection method based on neutral net - Google Patents
A kind of acoustics glass defect detection method based on neutral net Download PDFInfo
- Publication number
- CN106841403A CN106841403A CN201710053519.XA CN201710053519A CN106841403A CN 106841403 A CN106841403 A CN 106841403A CN 201710053519 A CN201710053519 A CN 201710053519A CN 106841403 A CN106841403 A CN 106841403A
- Authority
- CN
- China
- Prior art keywords
- neural network
- signal
- neutral net
- knocking
- detection method
- 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
- G01N29/04—Analysing solids
- G01N29/048—Marking the faulty objects
-
- 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
- G01N29/4481—Neural networks
-
- 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
- G01N2291/023—Solids
- G01N2291/0232—Glass, ceramics, concrete or stone
-
- 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)
- Immunology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Health & Medical Sciences (AREA)
- Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Acoustics & Sound (AREA)
- Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
Abstract
A kind of acoustics glass defect detection method based on neutral net:The knocking of glass sample is gathered using sound pick-up in actual production environment;Knocking is pre-processed;Feature extraction is carried out to pure knocking;Set the initial parameter of BP neural network:Input of the feature that will be extracted as neutral net, the input layer number for setting BP neural network is 7, the node in hidden layer for setting BP neural network is 15, the output layer nodes for setting BP neural network are 2, set output result as (0,1) represent that glass sample is defective, (1,0) represents glass sample zero defect;BP neural network is trained, the learning rate for setting BP neural network is 0.1, and target square error value is 0.1, and BP neural network is trained using LeVenberg Marquardt algorithms, when the error of neutral net is less than the target square error value for setting, deconditioning.The signal characteristic that the present invention is extracted has good discrimination, can more accurately and efficiently complete glass defect detection task.
Description
Technical field
The present invention relates to a kind of glass defect detection method.More particularly to a kind of acoustics glass based on neutral net lacks
Fall into detection method.
Background technology
Glass is one of most common material, is widely used in each industrial circle, while being also most fragile, most easily damaged
One of material.In the generation and transportation of glassware, the defects such as hole, crackle, trachoma can be caused.If can not be
Before sales rejects defective product, then will reduce the qualification rate of product, brings certain economic loss.
At present, mainly there are artificial process, Computer Vision Detection method, ultrasound for the lossless detection method of glassware defect
Detection method, vibration detection method etc..Artificial process is exactly that product is checked by experienced workman.There is following lacking in the method
Point:(1) False Rate is higher.Because the labour intensity of workman is larger, observed for a long time, human eye is easy to fatigue, meeting occur
Directly affect the accuracy of classification.(2) whether qualified judgement standard disunity is be during for small defect during hand inspection
Examiner's subjective judgement, different people's even criterions of the same people in the state of difference are difficult to accomplish unification.Calculate
Machine visual detection method is the image information for using camera to obtain body surface, then detects surface by Digital Image Processing
Crackle.The method has a wide range of application, but can only detect surface defect, and is influenceed greatly by ambient lighting.Ultrasonic Detection Method is
Using ultrasonic wave defect is detected by being detected the decay of reflected signal and received wave that object is produced.Vibration detection method is analysis
The Vibrating modal parameters (vibration shape, amplitude, damping etc.) of object detect defect.
Sound is produced by mechanical oscillation.Gas molecule around vibrating body disturbance is so that air pressure produces the cycle
The change of property.This pressure change forms close or thin air wave, and is radiate to the surrounding of object, is formed sound
Ripple.Object structures are damaged research and are shown, the damage of object structures must cause the change of rigidity, and this change and damage
Type, degree have close relationship.The situation that object is damaged is achieved with by the monitoring of the change to rigidity.However, just
Degree is to be difficult to measured directly, so the knocking for passing through object analysis, can reflect the change of rigidity, and then determine to damage
Situation.After Crack Damage occurs in vial, the distribution of the time domain energy of knocking, frequency peak and peak are all sent out
Change is given birth to, it is possible to judge that vial whether there is defect by analyzing these features of knocking.
Current existing method correlation collection signal mainly using signal in passage on a timeline, but due to sound
The unstability of message number, correlation and bad, causes the effect after sparse treatment undesirable on a timeline.
The content of the invention
The technical problems to be solved by the invention are to provide one kind and can more accurately and efficiently complete glass defect detection times
The acoustics glass defect detection method based on neutral net of business.
The technical solution adopted in the present invention is:A kind of acoustics glass defect detection method based on neutral net, including
Following steps:
1) knocking of glass sample is gathered using sound pick-up in actual production environment;
2) knocking for collecting is pre-processed, including noise reduction, end-point detection and rejecting abnormalities data, obtain pure
Net knocking;
3) feature extraction is carried out to pure knocking, including temporal signatures are extracted, frequency domain character is extracted and wavelet field
Feature extraction;
4) initial parameter of BP neural network is set:Using step 3) extract feature as neutral net input, set
The input layer number of BP neural network is 7, and the node in hidden layer for setting BP neural network is 15, sets BP neural network
Output layer nodes be 2, set output result for (0,1) expression glass sample it is defective, (1,0) expression glass sample it is intact
Fall into;
5) BP neural network is trained, the learning rate for setting BP neural network is 0.1, target square error value is
0.1, BP neural network is trained using LeVenberg-Marquardt algorithms, when the error of neutral net is less than setting
Target square error value when, deconditioning.
Step 2) in, noise reduction is carried out using high-pass filter, using short-time energy and the double threshold of short-time average zero-crossing rate
Algorithm carries out end-point detection to tapping voice signal.
Step 3) described in temporal signatures extract include:
(1) average of signal is extracted, using equation below:
Wherein, N is the length of complete knocking, xiIt is i-th amplitude of signaling point,It is the signal average extracted;
(2) the root-mean-square value RMS of signal is extracted, using equation below:
(3) the peak value peak of signal is extracted, using equation below:
Step 3) described in frequency domain character extract:Fourier transformation first is carried out to signal, signal spectrum is obtained, then ask respectively
The area and signal spectrum basic frequency of signal spectrum are taken, wherein:
The area of described signal spectrum refers to the area that spectrum signal is surrounded with reference axis;Described signal spectrum dominant frequency
Rate refers to the corresponding abscissa value of signal spectrum maximum.
Step 3) described in small echo characteristic of field extraction, be that db4 wavelet package transforms are carried out to signal, using the He of node 3.1
3.3 used as characteristic node, and uses the energy of the signal after reconstruct as small echo characteristic of field.
Step 4) described in setting BP neural network node in hidden layer be 15, be according to Kolomogorov determine
Reason is chosen:
nh=2nr+1
In formula, nhIt is hidden layer node number, nrIt is input layer number.
A kind of acoustics glass defect detection method based on neutral net of the invention, the present invention is by analyzing knocking
Time domain, frequency domain and small echo characteristic of field change, and sample is learnt and is trained using neutral net, so as to judge glass
Glass product whether there is defect, be a kind of more effective spatial audio coding method, using less transmission channel Accurate Reconstruction
Go out primary signal.The signal characteristic that the present invention is extracted has good discrimination, and using the non-linear correspondence pass of neutral net
Disaggregated model is built by system, can efficiently solve that existing detection method at present is present influenceed by objective environment greatly, detection ties
Really undesirable, excessively sensitive to noise the problems such as, more accurately and efficiently complete glass defect detection task.
Brief description of the drawings
Fig. 1 is the flow chart of acoustics glass defect detection method of the present invention based on neutral net.
Specific embodiment
With reference to embodiment and accompanying drawing to a kind of acoustics glass defect detection method based on neutral net of the invention
It is described in detail.
As shown in figure 1, a kind of acoustics glass defect detection method based on neutral net of the invention, including following step
Suddenly:
1) knocking of glass sample is gathered using sound pick-up in actual production environment;
2) due to there are noise and other inevitable factors in environment, there is noise and different in the data for causing collection
Regular signal.In order to obtain pure knocking, it is necessary to the knocking to collecting is pre-processed, including noise reduction, end points
Detection and rejecting abnormalities data, obtain pure knocking., wherein, noise reduction is carried out using high-pass filter, using in short-term
The double-threshold algorithm of amount and short-time average zero-crossing rate carries out end-point detection to tapping voice signal;
3) feature extraction is carried out to pure knocking, including temporal signatures are extracted, frequency domain character is extracted and wavelet field
Feature extraction;Wherein:
Described temporal signatures are extracted to be included:
(1) average of signal is extracted, using equation below:
Wherein, N is the length of complete knocking, xiIt is i-th amplitude of signaling point,It is the signal average extracted;
(2) the root-mean-square value RMS of signal is extracted, using equation below:
(3) the peak value peak of signal is extracted, using equation below:
Described frequency domain character is extracted:Fourier transformation first is carried out to signal, signal spectrum is obtained, then asks for letter respectively
The area and signal spectrum basic frequency of number frequency spectrum, wherein:The area of described signal spectrum refers to that spectrum signal encloses with reference axis
Into area;Described signal spectrum basic frequency refers to the corresponding abscissa value of signal spectrum maximum.
The extraction of described small echo characteristic of field, is to carry out db4 wavelet package transforms to signal, using the conduct of node 3.1 and 3.3
Characteristic node, and the energy of the signal after reconstruct is used as small echo characteristic of field.
4) initial parameter of BP neural network is set:Using step 3) extract feature as neutral net input because
The dimension of eigenmatrix is 7, so the input layer number for setting BP neural network is 7, sets the implicit of BP neural network
Node layer number is 15, and system is divided into the defective and class of zero defect two for the identification of sample, therefore sets the defeated of BP neural network
It is 2 to go out node layer number, sets output result as (0,1) represents that glass sample is defective, and (1,0) represents glass sample zero defect;
The node in hidden layer of described setting BP neural network is 15, is selected according to Kolomogorov theorems
Take:
nh=2nr+1
In formula, nhIt is hidden layer node number, nrIt is input layer number.
5) BP neural network is trained, the learning rate for setting BP neural network is 0.1, target square error value is
0.1, BP neural network is trained using LeVenberg-Marquardt algorithms, when the error of neutral net is less than setting
Target square error value when, deconditioning.
In a kind of acoustics glass defect detection method based on neutral net of the invention, duration of knocking compared with
It is short, while the real-time in order to meet detecting system, so intercepting 1024 sampled points for each signal.Spy is carried out to signal
Levy after extraction, the eigenmatrix dimension of acquisition is 7.According to Kolomogorov theorems and many experiments, selected according to following formula
Take the number of hidden layer node:
nh=2nr+1
In formula, nhIt is hidden layer node number, nrIt is input layer number.So, the BP network hidden layer nodes of use
Number is 15.BP networks are trained using LeVenberg-Marquardt algorithms, learning rate is 0.1.
The training time of neutral net can be had a huge impact from different algorithms, the present invention attempted it is various not
Same convergence algorithm, in these algorithms, LeVenberg-Marquardt convergences of algorithm speed and neural network accuracy are all preferable.
Believe that also having more preferable acoustic feature and convergence algorithm can more accurately detect using method used in the present invention
Defect sample.
Claims (6)
1. a kind of acoustics glass defect detection method based on neutral net, it is characterised in that comprise the following steps:
1) knocking of glass sample is gathered using sound pick-up in actual production environment;
2) knocking for collecting is pre-processed, including noise reduction, end-point detection and rejecting abnormalities data, obtain pure
Knocking;
3) feature extraction is carried out to pure knocking, including temporal signatures are extracted, frequency domain character is extracted and small echo characteristic of field
Extract;
4) initial parameter of BP neural network is set:Using step 3) extract feature as neutral net input, set BP god
It is 7 through the input layer number of network, the node in hidden layer for setting BP neural network is 15, sets the defeated of BP neural network
It is 2 to go out node layer number, sets output result as (0,1) represents that glass sample is defective, and (1,0) represents glass sample zero defect;
5) BP neural network is trained, the learning rate for setting BP neural network is 0.1, target square error value is 0.1,
BP neural network is trained using LeVenberg-Marquardt algorithms, when the mesh of the error less than setting of neutral net
During mark square error value, deconditioning.
2. a kind of acoustics glass defect detection method based on neutral net according to claim 1, it is characterised in that step
It is rapid 2) in, noise reduction is carried out using high-pass filter, using short-time energy and short-time average zero-crossing rate double-threshold algorithm to tap
Voice signal carries out end-point detection.
3. a kind of acoustics glass defect detection method based on neutral net according to claim 1, it is characterised in that step
It is rapid 3) described in temporal signatures extract include:
(1) average of signal is extracted, using equation below:
Wherein, N is the length of complete knocking, xiIt is i-th amplitude of signaling point,It is the signal average extracted;
(2) the root-mean-square value RMS of signal is extracted, using equation below:
(3) the peak value peak of signal is extracted, using equation below:
4. a kind of acoustics glass defect detection method based on neutral net according to claim 1, it is characterised in that step
It is rapid 3) described in frequency domain character extract:Fourier transformation first is carried out to signal, signal spectrum is obtained, then asks for signal spectrum respectively
Area and signal spectrum basic frequency, wherein:
The area of described signal spectrum refers to the area that spectrum signal is surrounded with reference axis;Described signal spectrum basic frequency is
Refer to the corresponding abscissa value of signal spectrum maximum.
5. a kind of acoustics glass defect detection method based on neutral net according to claim 1, it is characterised in that step
It is rapid 3) described in small echo characteristic of field extraction, be that db4 wavelet package transforms are carried out to signal, using node 3.1 and 3.3 as feature
Node, and the energy of the signal after reconstruct is used as small echo characteristic of field.
6. a kind of acoustics glass defect detection method based on neutral net according to claim 1, it is characterised in that step
It is rapid 4) described in the node in hidden layer of setting BP neural network be 15, be to be chosen according to Kolomogorov theorems:
nh=2nr+1
In formula, nhIt is hidden layer node number, nrIt is input layer number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710053519.XA CN106841403A (en) | 2017-01-23 | 2017-01-23 | A kind of acoustics glass defect detection method based on neutral net |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710053519.XA CN106841403A (en) | 2017-01-23 | 2017-01-23 | A kind of acoustics glass defect detection method based on neutral net |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106841403A true CN106841403A (en) | 2017-06-13 |
Family
ID=59120591
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710053519.XA Pending CN106841403A (en) | 2017-01-23 | 2017-01-23 | A kind of acoustics glass defect detection method based on neutral net |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106841403A (en) |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108269249A (en) * | 2017-12-11 | 2018-07-10 | 深圳市智能机器人研究院 | A kind of bolt detecting system and its implementation |
CN108519149A (en) * | 2018-03-28 | 2018-09-11 | 长安大学 | A kind of tunnel accident monitor and alarm system and method based on sound Time-Frequency Analysis |
CN108899048A (en) * | 2018-05-10 | 2018-11-27 | 广东省智能制造研究所 | A kind of voice data classification method based on signal Time-frequency Decomposition |
CN109142547A (en) * | 2018-08-08 | 2019-01-04 | 广东省智能制造研究所 | A kind of online lossless detection method of acoustics based on convolutional neural networks |
CN110208377A (en) * | 2019-06-19 | 2019-09-06 | 南京邮电大学 | A kind of more characteristic parameters damage degree assessment method based on Lamb wave |
CN111044621A (en) * | 2018-10-11 | 2020-04-21 | 苏州奥科姆自动化科技有限公司 | Nondestructive testing system and method based on sound quality and acoustic characteristics |
CN111443131A (en) * | 2020-04-26 | 2020-07-24 | 广州市市政工程试验检测有限公司 | Method for detecting grouting compactness of steel bar sleeve |
CN111507418A (en) * | 2020-04-21 | 2020-08-07 | 中国科学技术大学 | Encaustic tile quality detection method |
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 |
CN112185419A (en) * | 2020-09-30 | 2021-01-05 | 天津大学 | Glass bottle crack detection method based on machine learning |
CN112255308A (en) * | 2020-09-09 | 2021-01-22 | 中国大唐集团科学技术研究院有限公司火力发电技术研究院 | Bolt knocking detection method based on K-means clustering algorithm |
CN112683926A (en) * | 2021-01-09 | 2021-04-20 | 杭州晶硝子玻璃科技有限公司 | Glass defect detection device |
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 |
CN113706468A (en) * | 2021-07-27 | 2021-11-26 | 河北光兴半导体技术有限公司 | Glass defect detection method based on BP neural network |
CN113671031B (en) * | 2021-08-20 | 2024-06-21 | 贝壳找房(北京)科技有限公司 | Wall hollowing detection method and device |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE4207728A1 (en) * | 1992-03-11 | 1993-09-23 | Fraunhofer Ges Forschung | Quality testing of test objects - externally stimulating object to emit sound waves which are analysed to classify objects using neural network taught using objects of known class |
EP1375415A1 (en) * | 2002-06-20 | 2004-01-02 | Gilbarco S.p.A. | Apparatus and method for controlling the filing of a tank |
CN101509604A (en) * | 2009-03-20 | 2009-08-19 | 武汉大学 | Method and device for detecting and assessing deposit in metal pipe |
CN101673340A (en) * | 2009-08-13 | 2010-03-17 | 重庆大学 | Method for identifying human ear by colligating multi-direction and multi-dimension and BP neural network |
US20140165729A1 (en) * | 2012-12-13 | 2014-06-19 | Korea Institute Of Machinery & Materials | Acoustic emission diagnosis device for gas vessel using probabilistic neural network and method of diagnosing defect of cylinder using the same |
CN104634878A (en) * | 2014-12-16 | 2015-05-20 | 北京林业大学 | Wood damage monitoring method based on acoustic emission technique |
-
2017
- 2017-01-23 CN CN201710053519.XA patent/CN106841403A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE4207728A1 (en) * | 1992-03-11 | 1993-09-23 | Fraunhofer Ges Forschung | Quality testing of test objects - externally stimulating object to emit sound waves which are analysed to classify objects using neural network taught using objects of known class |
EP1375415A1 (en) * | 2002-06-20 | 2004-01-02 | Gilbarco S.p.A. | Apparatus and method for controlling the filing of a tank |
CN101509604A (en) * | 2009-03-20 | 2009-08-19 | 武汉大学 | Method and device for detecting and assessing deposit in metal pipe |
CN101673340A (en) * | 2009-08-13 | 2010-03-17 | 重庆大学 | Method for identifying human ear by colligating multi-direction and multi-dimension and BP neural network |
US20140165729A1 (en) * | 2012-12-13 | 2014-06-19 | Korea Institute Of Machinery & Materials | Acoustic emission diagnosis device for gas vessel using probabilistic neural network and method of diagnosing defect of cylinder using the same |
CN104634878A (en) * | 2014-12-16 | 2015-05-20 | 北京林业大学 | Wood damage monitoring method based on acoustic emission technique |
Non-Patent Citations (4)
Title |
---|
俞缙 著: "《缺陷岩体纵波传播特性分析技术》", 31 May 2013, 冶金工业出版社 * |
叶孟蜍 等: "基于BP 神经网络的陶瓷产品质量检验专家系统", 《大连轻工业学院学报》 * |
时建峰: "基于时、频域-小波分析和神经网络方法的齿轮箱故障诊断研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
郑源 主编: "《水力机组状态监测与故障诊断》", 30 June 2016, 中国水利水电出版社 * |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108269249A (en) * | 2017-12-11 | 2018-07-10 | 深圳市智能机器人研究院 | A kind of bolt detecting system and its implementation |
CN108519149A (en) * | 2018-03-28 | 2018-09-11 | 长安大学 | A kind of tunnel accident monitor and alarm system and method based on sound Time-Frequency Analysis |
CN108899048A (en) * | 2018-05-10 | 2018-11-27 | 广东省智能制造研究所 | A kind of voice data classification method based on signal Time-frequency Decomposition |
CN109142547B (en) * | 2018-08-08 | 2021-02-23 | 广东省智能制造研究所 | Acoustic online nondestructive testing method based on convolutional neural network |
CN109142547A (en) * | 2018-08-08 | 2019-01-04 | 广东省智能制造研究所 | A kind of online lossless detection method of acoustics based on convolutional neural networks |
CN111044621B (en) * | 2018-10-11 | 2022-04-26 | 苏州奥科姆自动化科技有限公司 | Nondestructive testing system and method based on sound quality and acoustic characteristics |
CN111044621A (en) * | 2018-10-11 | 2020-04-21 | 苏州奥科姆自动化科技有限公司 | Nondestructive testing system and method based on sound quality and acoustic characteristics |
CN110208377A (en) * | 2019-06-19 | 2019-09-06 | 南京邮电大学 | A kind of more characteristic parameters damage degree assessment method based on Lamb wave |
CN111507418A (en) * | 2020-04-21 | 2020-08-07 | 中国科学技术大学 | Encaustic tile quality detection method |
CN111507418B (en) * | 2020-04-21 | 2022-09-06 | 中国科学技术大学 | Encaustic tile quality detection method |
CN111443131A (en) * | 2020-04-26 | 2020-07-24 | 广州市市政工程试验检测有限公司 | Method for detecting grouting compactness of steel bar sleeve |
CN111443131B (en) * | 2020-04-26 | 2022-11-18 | 广州市市政工程试验检测有限公司 | Method for detecting grouting compactness of steel bar sleeve |
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 |
CN112255308A (en) * | 2020-09-09 | 2021-01-22 | 中国大唐集团科学技术研究院有限公司火力发电技术研究院 | Bolt knocking detection method based on K-means clustering algorithm |
CN112185419A (en) * | 2020-09-30 | 2021-01-05 | 天津大学 | Glass bottle crack detection method based on machine learning |
CN112683926B (en) * | 2021-01-09 | 2023-08-22 | 临汾常兴玻璃有限公司 | Glass defect detection device |
CN112683926A (en) * | 2021-01-09 | 2021-04-20 | 杭州晶硝子玻璃科技有限公司 | Glass defect detection device |
CN113706468A (en) * | 2021-07-27 | 2021-11-26 | 河北光兴半导体技术有限公司 | Glass defect detection method based on BP neural network |
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 |
CN113671031B (en) * | 2021-08-20 | 2024-06-21 | 贝壳找房(北京)科技有限公司 | Wall hollowing detection method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106841403A (en) | A kind of acoustics glass defect detection method based on neutral net | |
CN109142547B (en) | Acoustic online nondestructive testing method based on convolutional neural network | |
CN109616140B (en) | Abnormal sound analysis system | |
CN112232400A (en) | Stainless steel weld ultrasonic defect detection method based on depth feature fusion | |
Zhang et al. | Deep convolutional neural network probability imaging for plate structural health monitoring using guided waves | |
CN111983020B (en) | System and method for detecting and identifying internal defects of metal component through knocking | |
CN103776903B (en) | A kind of wind electricity blade delamination detection method and detection system | |
Zhou et al. | A new method to classify railway vehicle axle fatigue crack AE signal | |
CN110243934A (en) | A kind of ultrasonic weld seam detection method based on wavelet convolution neural network | |
CN111767897A (en) | Support vector machine-based steel rail crack defect identification method | |
CN104634872A (en) | Online high-speed railway steel rail damage monitoring method | |
CN110487917A (en) | A kind of measure on stress pulse neural network based and analysis system | |
CN108269249A (en) | A kind of bolt detecting system and its implementation | |
CN109063762A (en) | A kind of line clogging fault recognition method based on DT-CWT and S4VM | |
CN104316277A (en) | Acoustic detection and blind signal separation-based air tightness monitoring method and apparatus | |
CN114137079A (en) | Ultrasonic guided wave nondestructive testing method based on combination of deep learning and Duffing system | |
CN106198765A (en) | A kind of acoustic signal recognition methods for Metal Crack monitoring | |
CN115754022A (en) | Steel truss welding node damage degree identification method based on acoustic emission | |
CN108630228A (en) | Sound quality recognition methods, device, system and vehicle | |
CN2809640Y (en) | Ultrasonic online intelligent recognition system for structural defect | |
Dushyanth et al. | Detection and localization of damage using empirical mode decomposition and multilevel support vector machine | |
CN104374576A (en) | Method for extracting fault stress waves of low-speed bearing | |
JP3885297B2 (en) | Abnormal sound determination device and abnormal sound determination method | |
CN205844271U (en) | A kind of detecting system based on difference frequency non-linear ultrasonic detection sheet metal micro-crack | |
Liu et al. | Fuzzy pattern recognition of impact acoustic signals for nondestructive evaluation |
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: 20170613 |
|
RJ01 | Rejection of invention patent application after publication |