CN106714067B - Automatic detection method and device on production line - Google Patents
Automatic detection method and device on production line Download PDFInfo
- Publication number
- CN106714067B CN106714067B CN201510791066.1A CN201510791066A CN106714067B CN 106714067 B CN106714067 B CN 106714067B CN 201510791066 A CN201510791066 A CN 201510791066A CN 106714067 B CN106714067 B CN 106714067B
- Authority
- CN
- China
- Prior art keywords
- audio data
- neural network
- network model
- sample
- vibration
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention discloses an automatic detection method and device on a production line. When the fact that the equipment to be tested on the production line plays the test audio is detected, audio data of the equipment to be tested under the preset frequency and vibration offset corresponding to the audio data are collected; acquiring a neural network model obtained by vibration sample data learning training; and inputting the audio data and the vibration offset into a neural network model, comparing the audio data and the vibration offset with a sample characteristic space in the neural network model, outputting a comparison result, and determining the detection result of the equipment to be detected according to the comparison result. In the embodiment of the invention, an automatic detection method on a production line is adopted, the condition of low manual detection accuracy is avoided, and whether the equipment to be detected has the vibro-acoustic defect or not is determined by acquiring the audio data of the equipment to be detected under the preset frequency and the vibration offset corresponding to the audio data for analysis and comparison, so that the precision of vibro-acoustic defect detection is greatly improved.
Description
Technical Field
The invention relates to the field of testing, in particular to an automatic detection method and device on a production line.
Background
The quality detection of the produced equipment on the production line, for example, the quality detection of the loudspeaker of the equipment, including the existence of silence, the existence of large-amplitude attenuation flaw, the existence of vibration sound flaw and the like, wherein the vibration sound refers to the phenomenon that a loud banging sound is emitted to generate resonance of a machine shell and the loudspeaker, and the phenomenon is promoted by a plurality of factors, such as the material and thickness of the machine shell of the equipment, the designed structure, the installation position and the like.
The audio can be played through a loudspeaker and identified by human ears, but the manual detection is inefficient. In the automatic detection of the prior art, the audio played by the loudspeaker is collected by the microphone, whether defects exist is detected by methods such as frequency analysis on the collected audio, but noise exists on the site of a production line, interference is easy to cause, the difficulty of analyzing and detecting the collected audio is increased, and the accuracy of detection on the vibration defect is greatly reduced.
Disclosure of Invention
The invention aims to provide an automatic detection method and device on a production line, which avoid the condition of low manual detection accuracy and greatly improve the detection precision of vibro-acoustic defects by adopting the automatic detection method on the production line.
The invention provides an automatic detection method on a production line, which comprises the following steps:
when detecting that the equipment to be tested on the production line plays a test audio, acquiring audio data of the equipment to be tested under a preset frequency and a vibration offset corresponding to the audio data;
acquiring a neural network model obtained by vibration sample data learning training;
and inputting the audio data and the vibration offset into the neural network model, comparing the audio data and the vibration offset with a sample feature space in the neural network model, outputting a comparison result, and determining the detection result of the equipment to be detected according to the comparison result.
Preferably, when it is detected that the device to be tested on the production line plays the test audio, before acquiring the audio data of the device to be tested at the preset frequency and the vibration offset corresponding to the audio data, the method further includes:
collecting the sample audio data of the equipment to be tested under the preset frequency; collecting the sample vibration offset acquired by the vibration sensor; the sample vibration offset corresponds to the preset frequency;
and forming the sample characteristic space by the sample audio data and the sample vibration offset, and performing learning training on the sample characteristic space to obtain the neural network model through training.
Preferably, inputting the audio data and the vibration offset into the neural network model, comparing the audio data and the vibration offset with a sample feature space in the neural network model, outputting a comparison result, and determining a detection result of the device under test according to the comparison result, including:
if the vibration offset corresponding to the audio data meets a preset threshold value in the neural network model, determining the category of the comparison result, wherein the category of the comparison result comprises: vibronic defective devices or sound intact devices.
Preferably, the vibration sensor includes: an optical vibration sensor, the vibration offset comprising: the light emitted by the optical vibration sensor refracts the light intensity received behind the device under test.
Preferably, after determining the detection result of the device under test according to the comparison result, the method further includes:
inputting the detection result of the equipment to be detected into the neural network model, and calculating the detection accuracy;
and if the accuracy obtained by calculation is lower than a reference threshold value, updating the neural network model according to the detection result of the equipment to be detected.
The second aspect of the present invention provides an automatic detection device on a production line, comprising:
the device comprises an acquisition module, a processing module and a control module, wherein the acquisition module is used for acquiring audio data of the device to be tested under a preset frequency and vibration offset corresponding to the audio data when the device to be tested on the production line is detected to play a test audio;
the acquisition module is used for acquiring a neural network model obtained by vibration sample data learning training;
and the detection module is used for inputting the audio data and the vibration offset into the neural network model, comparing the audio data and the vibration offset with a sample feature space in the neural network model, outputting a comparison result, and determining the detection result of the equipment to be detected according to the comparison result.
Preferably, the apparatus further comprises: a training module;
the acquisition module is further configured to acquire the sample audio data of the device under test at the preset frequency; collecting the sample vibration offset acquired by the vibration sensor; the sample vibration offset corresponds to the preset frequency;
and the training module is used for forming the sample characteristic space by the sample audio data and the sample vibration offset, performing learning training on the sample characteristic space, and training to obtain the neural network model.
Preferably, the detection module is specifically configured to:
if the vibration offset corresponding to the audio data meets a preset threshold value in the neural network model, determining the category of the comparison result, wherein the category of the comparison result comprises: vibronic defective devices or sound intact devices.
Preferably, the vibration sensor includes: an optical vibration sensor, the vibration offset comprising: the light emitted by the optical vibration sensor refracts the light intensity received behind the device under test.
Preferably, the apparatus further comprises:
the calculation module is used for inputting the detection result of the equipment to be detected into the neural network model and calculating the detection accuracy;
and the updating module is used for updating the neural network model according to the detection result of the equipment to be detected when the calculated accuracy is lower than a reference threshold value.
The implementation of the invention has the following beneficial effects:
in the embodiment of the invention, when the device to be tested on the production line is detected to play the test audio, audio data of the device to be tested under the preset frequency and the vibration offset corresponding to the audio data are collected; acquiring a neural network model obtained by vibration sample data learning training; and inputting the audio data and the vibration offset into a neural network model, comparing the audio data and the vibration offset with a sample characteristic space in the neural network model, outputting a comparison result, and determining the detection result of the equipment to be detected according to the comparison result. According to the embodiment of the invention, the situation of low manual detection accuracy is avoided by adopting an automatic detection method on the production line, whether the equipment to be detected has the vibro-acoustic defect is determined by acquiring the audio data of the equipment to be detected under the preset frequency and analyzing and comparing the vibration offset corresponding to the audio data, the interference of sound or light waves on the production line on the vibro-acoustic detection is avoided by combining the audio acquisition and the vibration test, and the vibro-acoustic defect detection precision is greatly improved.
Drawings
FIG. 1 is a flow chart of an automated inspection method for an assembly line according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of another automated inspection method for an assembly line according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an automatic detection device on a production line according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The embodiment of the invention provides an automatic detection method and device on a production line, wherein the automatic detection device on the production line can be detection equipment on the production line and used for collecting data and processing the data, such as a machine vision robot and the like used for detecting products such as televisions, displays, sound boxes and the like on the production line. The production line provided by the embodiment of the invention comprises a production line, equipment to be detected, an automatic detection device and a sensor, wherein the equipment to be detected is arranged on the production line, the automatic detection device is connected with the sensor, and the automatic detection device can control the equipment to be detected. By the embodiment of the invention, automatic vibro-acoustic detection on a production line can be realized, and vibro-acoustic detection precision of equipment to be detected is improved. The following is detailed by way of specific examples.
Referring to fig. 1, a schematic diagram of an automatic detection method on a production line according to an embodiment of the present invention is provided. As shown in the figure, the automatic detection method on the production line provided by the embodiment of the invention can comprise steps S101-S103.
S101, when it is detected that the equipment to be tested on the production line plays the test audio, acquiring audio data of the equipment to be tested under the preset frequency and vibration offset corresponding to the audio data.
In the specific implementation, when the equipment to be tested on the production line plays the test audio, the test audio played by the equipment to be tested can be collected through the microphone to obtain audio data, and further, the audio data under the preset frequency is collected. For example, collecting the device under test with a microphoneRespectively recording the arrays A of the corresponding volume data obtained by the actual measurement of the microphone when the equipment to be tested is debugged to the frequencies of 25,50,75 and 100sound＝{a1,a2,a3,a4}。
Specifically, a plane where a horn of the equipment to be tested is located serves as a system plane, an infrared laser transmitter is installed at one end, laser emitted by a laser device irradiates the plane of the horn of the equipment to be tested, an infrared laser receiver is installed along a light emitting path, and the two infrared laser transmitters are fixedly installed. Under the condition that the precision of the infrared receiver is enough, according to the law of light wave reflection, if the plane of a horn of the equipment to be tested is stable and does not move, the energy emitted by the infrared laser and the energy received by the receiver are conserved.
Further, when the infrared laser transmitter and the receiver are normally operated, the receiver is connected to the quantized light intensity read data d1. When the equipment to be tested plays audio, if the defect of vibro-acoustic exists, the casing of the equipment to be tested can irregularly vibrate along with the loudspeaker, so that the diffuse reflection phenomenon occurs, the actual reflection path of the light wave of the infrared laser is changed, the receiving path of the infrared laser is unchanged, part of the emitted light wave is positioned outside the receiving area, so that the brightness of the collected light is reduced, and the data d is read out after the brightness is quantized at the moment1±Δx。
The amount of vibration offset corresponding to the audio data may be collected by a vibration sensor, such as an optical vibration sensor. The vibration offset includes: the light emitted by the optical vibration sensor refracts the intensity of light received behind the device under test. Different frequencies may be set corresponding to different intensities of infrared laser light emitted by the system. For example, the test audio is played for 10 seconds or more, collected every 200 milliseconds at regular intervals, and 50 pieces of audio data are collected quantitatively and recorded as Δ X ═ { Δ X ═1,Δx2,...Δx50Recovering the acquired discrete light intensity data into continuous light intensity data, and recording corresponding light intensity data I with frequencies of 50Hz, 75Hz, 100Hz, 125Hz and 150HzHz＝{i1,i2,i3,i4,i5}。
And S102, acquiring a neural network model obtained by vibration sample data learning training.
In specific implementation, the neural network model is obtained by learning and training vibration sample data in advance, such as a bp (back propagation) neural network. Further, the method steps before step S101 are implemented as follows:
collecting sample audio data of equipment to be tested under a preset frequency; collecting sample vibration offset acquired by a vibration sensor; the sample vibration offset corresponds to a preset frequency;
and forming a sample characteristic space by the sample audio data and the sample vibration offset, and performing learning training on the sample characteristic space to obtain a neural network model through training.
Specifically, when the equipment to be tested on the production line plays the test audio, the test audio played by the equipment to be tested can be collected through the microphone to obtain sample audio data, and further, the sample audio data under the preset frequency is collected. Furthermore, for enriching the sample set, the device to be tested is set to play the same sound audio file each time, and the sound file covers continuous sound of multiple frequencies as much as possible. For example, the played audio is set to be a continuous periodic sinusoidal signal, and the laser emitted by the infrared laser is a continuous periodic sinusoidal signal synchronized with the audio. Periodically collecting by an infrared receiver, for example, the test audio is played for more than 10 seconds, periodically collecting every 200 milliseconds, quantitatively collecting 50 audio data, recording as delta X ', recovering continuous light intensity data from the collected discrete light intensity data, and recording corresponding light intensity data I ' with frequencies of 50Hz, 75Hz, 100Hz, 125Hz and 150Hz 'Hz. Collecting a plurality of normal devices to be tested, wherein the sample is Y { (A)sound,IHz)|AsoundE (25,50,75,100) }, marked Result ═ 1; and collecting multiple TV sets with vibration, if there are not multiple devices to be tested with vibration defect, then collecting multiple times with one defective device to be tested, the sample is Y { (A)sound,IHz)|AsoundE (25,50,75,100) }, marked Result ═ 0.
With Δ X 'and I'HzForming a sample feature space, and performing learning training through a BP (Back propagation) neural network. For example, 40 samples are provided in the sample library, each sample has 50 parameters Δ x as the input of the first layer cells, the learning training result is output, and the experienced neural network is obtained after learning.
S103, inputting the audio data and the vibration offset into the neural network model, comparing the audio data and the vibration offset with a sample feature space in the neural network model, outputting a comparison result, and determining a detection result of the equipment to be detected according to the comparison result.
In specific implementation, the collected audio data and the vibration offset are input into a neural network model, and compared with a sample feature space in the neural network model, classification is performed, a comparison result is output, and the category of the comparison result is determined, wherein the category of the comparison result comprises: vibronic defective devices or sound intact devices.
Further optionally, step S103 may specifically include:
if the vibration offset corresponding to the audio data meets a preset threshold value in the neural network model, determining the category of the comparison result, wherein the category of the comparison result comprises: vibronic defective devices or sound intact devices.
In a specific implementation, in order to eliminate interference of other co-frequency sound/light waves on a production line, the following convention may be made: during testing, the system judges that the light intensity received by the infrared receiver can be changed into the light intensity with any 3 frequencies within the preset threshold value of the BP neural network under the condition of determining the frequency of the equipment to be tested, namely, under the 5 frequencies of 50Hz, 75Hz, 100Hz, 125Hz and 150Hz, the corresponding neural network training result can be output.
In the interference of sound/light waves on the production line, the condition that any 3 frequencies of 50Hz, 75Hz, 100Hz, 125Hz and 150Hz exist and interfere the test result rarely occurs at the same time, so that the misjudgment condition can be greatly reduced, and the method can be realized on a common production line without a very harsh quiet environment.
Further, after step S103, the method provided in the embodiment of the present invention may further include the steps of:
inputting the detection result of the equipment to be detected into the neural network model, and calculating the detection accuracy;
and if the accuracy obtained by calculation is lower than the reference threshold value, updating the neural network model according to the detection result of the equipment to be detected.
In specific implementation, the neural network algorithm of the embodiment of the invention further implements a relearning function, returns the detection result of the device to be detected to the input neural network model, calculates the accuracy of the detection, updates the neural network model when the accuracy is lower than a reference threshold value, and corrects the neural network model according to the device to be detected.
As shown in fig. 2, which is another schematic diagram of the automated inspection method on the production line according to the embodiment of the present invention, the device to be inspected is disposed on the production line, and the automated inspection apparatus may be disposed on the production line or outside the production line, and is configured to perform wireless communication. The automatic detection device is connected with the vibration sensor and provided with a microphone for collecting audio data. The method comprises the steps of obtaining a neural network model through sample vibration offset and sample audio data learning training, inputting the vibration offset and the audio data into the trained neural network model when detecting equipment to be detected, carrying out classification detection, finally outputting a result to obtain a detection result, and returning the detection result to the neural network model for relearning and correction.
The embodiment of the invention provides an automatic detection method on a production line, which comprises the steps of collecting audio data of equipment to be detected under a preset frequency and vibration offset corresponding to the audio data when the equipment to be detected on the production line is detected to play a test audio; acquiring a neural network model obtained by vibration sample data learning training; and inputting the audio data and the vibration offset into a neural network model, comparing the audio data and the vibration offset with a sample characteristic space in the neural network model, outputting a comparison result, and determining the detection result of the equipment to be detected according to the comparison result. According to the embodiment of the invention, the situation of low manual detection accuracy is avoided by adopting an automatic detection method on a production line, whether the equipment to be detected has the vibro-acoustic defect is determined by acquiring the audio data of the equipment to be detected under the preset frequency and the vibration offset corresponding to the audio data for analysis and comparison, the interference of sound or light waves on vibro-acoustic detection on the production line is avoided by combining the audio acquisition and the vibration test, the obtained detection result is returned to the neural network model for learning and correction, and the vibro-acoustic defect detection precision is greatly improved.
In the following, an automatic detection device on a production line according to an embodiment of the present invention will be described in detail with reference to fig. 3, it should be noted that the device shown in fig. 3 is an implementation body of the method shown in fig. 1, for convenience of description, only a part related to the embodiment of the present invention is shown, and details of the technology are not disclosed, please refer to the embodiment shown in fig. 1 and fig. 2 according to the embodiment of the present invention.
As shown in fig. 3, an automatic detection apparatus on a production line according to an embodiment of the present invention includes: an acquisition module 301, an acquisition module 302, and a detection module 303.
The acquisition module 301 is configured to acquire audio data of the device to be tested at a preset frequency and a vibration offset corresponding to the audio data when it is detected that the device to be tested on the production line plays a test audio.
In the specific implementation, when the equipment to be tested on the production line plays the test audio, the test audio played by the equipment to be tested can be collected through the microphone to obtain audio data, and further, the audio data under the preset frequency is collected. For example, a microphone is used to collect the actually played audio of the device under test, and array a of the corresponding volume data actually measured by the microphone when the device under test is tuned to frequencies 25,50,75 and 100 is recordedsound＝{a1,a2,a3,a4}。
The amount of vibration offset corresponding to the audio data may be collected by a vibration sensor, such as an optical vibration sensor. The vibration offset includes: the light emitted by the optical vibration sensor refracts the intensity of light received behind the device under test. Different frequencies may be set corresponding to different intensities of infrared laser light emitted by the system. For example, the test audio is played for more than 10 seconds and is periodically collected every 200 millisecondsQuantitative acquisition of 50 audio data recorded as Δ X ═ Δ X1,Δx2,...Δx50Recovering the acquired discrete light intensity data into continuous light intensity data, and recording corresponding light intensity data I with frequencies of 50Hz, 75Hz, 100Hz, 125Hz and 150HzHz＝{i1,i2,i3,i4,i5}。
An obtaining module 302, configured to obtain a neural network model obtained through vibration sample data learning training.
Further optionally, the apparatus provided in the embodiment of the present invention further includes: a training module 304;
the acquisition module 301 is further configured to acquire sample audio data of the device under test at a preset frequency; collecting sample vibration offset acquired by a vibration sensor; the sample vibration offset corresponds to a preset frequency;
and the training module 304 is configured to form a sample feature space by the sample audio data and the sample vibration offset, perform learning training on the sample feature space, and train to obtain a neural network model.
How to train the neural network model is shown in fig. 1, and is not repeated here.
The detection module 303 is configured to input the audio data and the vibration offset into the neural network model, compare the audio data and the vibration offset with a sample feature space in the neural network model, output a comparison result, and determine a detection result of the device to be detected according to the comparison result.
In specific implementation, the detecting module 303 inputs the collected audio data and the vibration offset into the neural network model, compares the audio data and the vibration offset with a sample feature space in the neural network model, classifies the audio data and the vibration offset, outputs a comparison result, and determines a category of the comparison result, where the category of the comparison result includes: vibronic defective devices or sound intact devices.
Further, the detecting module 303 is specifically configured to:
if the vibration offset corresponding to the audio data meets a preset threshold value in the neural network model, determining the type of a comparison result, wherein the type of the comparison result comprises: vibronic defective devices or sound intact devices.
In a specific implementation, in order to eliminate interference of other co-frequency sound/light waves on a production line, the following convention may be made: during testing, the system judges that the light intensity received by the infrared receiver can be changed into the light intensity with any 3 frequencies within the preset threshold value of the BP neural network under the condition of determining the frequency of the equipment to be tested, namely, under the 5 frequencies of 50Hz, 75Hz, 100Hz, 125Hz and 150Hz, the corresponding neural network training result can be output. The method can greatly reduce the misjudgment condition, so that the method can be realized on a common production line without a very harsh quiet environment.
Further optionally, the apparatus provided in the embodiment of the present invention further includes: a calculation module 305 and an update module 306.
A calculation module 305, configured to input a detection result of the device under test into the neural network model, and calculate a detection accuracy;
and the updating module 306 is configured to update the neural network model according to the detection result of the device to be detected when the calculated accuracy is lower than the reference threshold.
In specific implementation, the neural network algorithm of the embodiment of the present invention further implements a relearning function, the detection result of the device to be tested is returned to the input neural network model, the calculation module 305 calculates the accuracy of the current detection, and when the accuracy is lower than the reference threshold, the update module 306 updates the neural network model and modifies the neural network model according to the device to be tested.
The embodiment of the invention provides an automatic detection device on a production line, wherein when detecting that equipment to be detected on the production line plays a test audio, an acquisition module is used for acquiring audio data of the equipment to be detected under a preset frequency and a vibration offset corresponding to the audio data; the acquisition module is used for acquiring a neural network model obtained by vibration sample data learning training; the detection module is used for inputting the audio data and the vibration offset into the neural network model, comparing the audio data and the vibration offset with a sample characteristic space in the neural network model, outputting a comparison result, and determining the detection result of the equipment to be detected according to the comparison result. According to the embodiment of the invention, the situation of low manual detection accuracy is avoided by adopting an automatic detection method on a production line, whether the equipment to be detected has the vibro-acoustic defect is determined by acquiring the audio data of the equipment to be detected under the preset frequency and the vibration offset corresponding to the audio data for analysis and comparison, the interference of sound or light waves on vibro-acoustic detection on the production line is avoided by combining the audio acquisition and the vibration test, the obtained detection result is returned to the neural network model for learning and correction, and the vibro-acoustic defect detection precision is greatly improved.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (8)
1. An automatic detection method on a production line is characterized by comprising the following steps:
when detecting that the equipment to be tested on the production line plays a test audio, acquiring audio data of the equipment to be tested under a preset frequency and a vibration offset corresponding to the audio data;
acquiring a neural network model obtained by vibration sample data learning training;
inputting the audio data and the vibration offset into the neural network model, comparing the audio data and the vibration offset with a sample feature space in the neural network model, outputting a comparison result, and determining a detection result of the equipment to be detected according to the comparison result;
wherein the vibration offset includes: the light emitted by the optical vibration sensor refracts the intensity of light received behind the device under test.
2. The method as claimed in claim 1, wherein before acquiring audio data of the device under test at a preset frequency and a vibration offset corresponding to the audio data when it is detected that the device under test on the production line plays the test audio, the method further includes:
collecting the sample audio data of the equipment to be tested under the preset frequency; collecting the sample vibration offset acquired by the vibration sensor; the sample vibration offset corresponds to the preset frequency;
and forming the sample characteristic space by the sample audio data and the sample vibration offset, and performing learning training on the sample characteristic space to obtain the neural network model through training.
3. The method of claim 1, wherein inputting the audio data and the vibration offset into the neural network model, comparing the audio data and the vibration offset with a sample feature space in the neural network model, and outputting a comparison result, and determining a detection result of the device under test according to the comparison result comprises:
if the vibration offset corresponding to the audio data meets a preset threshold value in the neural network model, determining the category of the comparison result, wherein the category of the comparison result comprises: vibronic defective devices or sound intact devices.
4. The method of claim 1, wherein after determining the detection result of the device under test from the comparison result, the method further comprises:
inputting the detection result of the equipment to be detected into the neural network model, and calculating the detection accuracy;
and if the accuracy obtained by calculation is lower than a reference threshold value, updating the neural network model according to the detection result of the equipment to be detected.
5. An automatic detection device on a production line is characterized by comprising:
the device comprises an acquisition module, a processing module and a control module, wherein the acquisition module is used for acquiring audio data of the device to be tested under a preset frequency and vibration offset corresponding to the audio data when the device to be tested on the production line is detected to play a test audio;
the acquisition module is used for acquiring a neural network model obtained by vibration sample data learning training;
the detection module is used for inputting the audio data and the vibration offset into the neural network model, comparing the audio data and the vibration offset with a sample feature space in the neural network model, outputting a comparison result, and determining a detection result of the equipment to be detected according to the comparison result;
wherein the vibration offset includes: the light emitted by the optical vibration sensor refracts the intensity of light received behind the device under test.
6. The apparatus of claim 5, wherein the apparatus further comprises: a training module;
the acquisition module is further configured to acquire the sample audio data of the device under test at the preset frequency; collecting the sample vibration offset acquired by the vibration sensor; the sample vibration offset corresponds to the preset frequency;
and the training module is used for forming the sample characteristic space by the sample audio data and the sample vibration offset, performing learning training on the sample characteristic space, and training to obtain the neural network model.
7. The apparatus of claim 5, wherein the detection module is specifically configured to:
if the vibration offset corresponding to the audio data meets a preset threshold value in the neural network model, determining the category of the comparison result, wherein the category of the comparison result comprises: vibronic defective devices or sound intact devices.
8. The apparatus of claim 5, wherein the apparatus further comprises:
the calculation module is used for inputting the detection result of the equipment to be detected into the neural network model and calculating the detection accuracy;
and the updating module is used for updating the neural network model according to the detection result of the equipment to be detected when the calculated accuracy is lower than a reference threshold value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510791066.1A CN106714067B (en) | 2015-11-17 | 2015-11-17 | Automatic detection method and device on production line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510791066.1A CN106714067B (en) | 2015-11-17 | 2015-11-17 | Automatic detection method and device on production line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106714067A CN106714067A (en) | 2017-05-24 |
| CN106714067B true CN106714067B (en) | 2020-12-15 |
Family
ID=58932407
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510791066.1A Active CN106714067B (en) | 2015-11-17 | 2015-11-17 | Automatic detection method and device on production line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106714067B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107948640B (en) * | 2017-12-19 | 2020-09-25 | 百度在线网络技术（北京）有限公司 | Video playing test method and device, electronic equipment and storage medium |
| CN109908578B (en) * | 2019-01-28 | 2022-07-05 | 努比亚技术有限公司 | Game vibration sense control method, terminal and computer readable storage medium |
| CN115512688A (en) * | 2022-09-02 | 2022-12-23 | 广东美云智数科技有限公司 | Abnormal sound detection method and device |
| CN117825036A (en) * | 2024-03-06 | 2024-04-05 | 梁山通亚重工机械有限公司 | Automatic quality inspection system for trailer production line |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1302524A (en) * | 1997-05-08 | 2001-07-04 | 艾利森公司 | Sound Transducer and method having light detector for detecting displacement of transducer diaphragm |
| CN101782743A (en) * | 2010-02-11 | 2010-07-21 | 浙江大学 | Neural network modeling method and system |
| CN202168211U (en) * | 2011-04-02 | 2012-03-14 | 嘉兴中科声学科技有限公司 | Device for measuring loudspeaker parameters by laser sensor |
| CN103245524A (en) * | 2013-05-24 | 2013-08-14 | 南京大学 | Acoustic fault diagnosis method based on neural network |
| CN104780489A (en) * | 2015-03-31 | 2015-07-15 | 歌尔声学股份有限公司 | Structure for detecting vibration displacement of loudspeaker and double-effect device capable of realizing acoustic-electric inter-conversion |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104365119B (en) * | 2012-06-07 | 2018-07-06 | 思睿逻辑国际半导体有限公司 | The nonlinear Control of loud speaker |
-
2015
- 2015-11-17 CN CN201510791066.1A patent/CN106714067B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1302524A (en) * | 1997-05-08 | 2001-07-04 | 艾利森公司 | Sound Transducer and method having light detector for detecting displacement of transducer diaphragm |
| CN101782743A (en) * | 2010-02-11 | 2010-07-21 | 浙江大学 | Neural network modeling method and system |
| CN202168211U (en) * | 2011-04-02 | 2012-03-14 | 嘉兴中科声学科技有限公司 | Device for measuring loudspeaker parameters by laser sensor |
| CN103245524A (en) * | 2013-05-24 | 2013-08-14 | 南京大学 | Acoustic fault diagnosis method based on neural network |
| CN104780489A (en) * | 2015-03-31 | 2015-07-15 | 歌尔声学股份有限公司 | Structure for detecting vibration displacement of loudspeaker and double-effect device capable of realizing acoustic-electric inter-conversion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106714067A (en) | 2017-05-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106714067B (en) | Automatic detection method and device on production line | |
| CN110044472B (en) | Intelligent detection system for abnormal sound and abnormal sound of online product | |
| RU2605522C2 (en) | Device containing plurality of audio sensors and operation method thereof | |
| EP3166239B1 (en) | Method and system for scoring human sound voice quality | |
| JP2008151538A (en) | Device for inspecting inside | |
| CN104807540B (en) | Noise check method and system | |
| US8917878B2 (en) | Microphone inspection method | |
| CN105979462A (en) | Test processing method and device based on microphone | |
| CN109443525A (en) | A kind of equipment abnormal sound detection system and detection method | |
| CN111276160A (en) | Audio detection method and device | |
| CN112382313A (en) | Audio communication quality evaluation system and method | |
| CN110784815B (en) | Device and method for testing acoustic performance of product | |
| US20150143912A1 (en) | Apparatus for nondestructive crack inspection | |
| KR20210070586A (en) | Method for treating sound using cnn | |
| JP5539437B2 (en) | Pure tone inspection apparatus and control method thereof | |
| CN114974301A (en) | Abnormal sound detection method, computer readable storage medium and electronic device | |
| US11313837B2 (en) | Device, system and method for imaging defects in a structure by transmitting and receiving mechanical waves in this structure | |
| CN102650618A (en) | Part assembling quality inspection method | |
| JP5958362B2 (en) | Abnormal sound inspection apparatus and abnormal sound inspection method | |
| Skoglind et al. | Localized Acoustic-Event Measurement Probe: Connector Confirmation Utilizing Acoustic Signatures | |
| JP4247310B2 (en) | Hardness measuring device | |
| KR20150061906A (en) | Apparatus for nondestructive crack inspection | |
| CN111711915B (en) | Audio part selection method and audio part selection device thereof | |
| US20240090489A1 (en) | Method of acoustically detecting early termite infestation | |
| TWI595791B (en) | Method of detecting audio signal |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230712 Address after: Yanxiang Technology Building, No. 31 Gaoxin Middle Fourth Road, Maling Community, Yuehai Street, Nanshan District, Shenzhen City, Guangdong Province, 518057 Patentee after: Shenzhen Yanxiang Smart Technology Co.,Ltd. Address before: 518107 5th floor, No.1, Yanxiang Zhigu chuangxiangdi, No.11, Gaoxin Road, Guangming New District, Shenzhen City, Guangdong Province Patentee before: EVOC INTELLIGENT TECHNOLOGY Co.,Ltd. |