CN110006664A - Automobile brake noise expert's detection method neural network based - Google Patents
Automobile brake noise expert's detection method neural network based Download PDFInfo
- Publication number
- CN110006664A CN110006664A CN201910268162.6A CN201910268162A CN110006664A CN 110006664 A CN110006664 A CN 110006664A CN 201910268162 A CN201910268162 A CN 201910268162A CN 110006664 A CN110006664 A CN 110006664A
- Authority
- CN
- China
- Prior art keywords
- neural network
- noise
- expert
- detection method
- training
- 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
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
A kind of automobile brake noise expert's detection method neural network based of automobile technical field, comprising the following steps: first, the training of neural network is carried out to collected brake block vibration and noise information；Second, the noise measuring of neural network is carried out to new collected measurement data；Third, the improvement of neural network, the test result based on second step, EDS extended data set, to carry out the repetitive exercise of neural network next time.The present invention to greatly reduce the workload of inspector, while still maintaining the assay reproducibility and reliability of height using the method judgement detection brake oil of neural network.
Description
Technical field
The present invention relates to a kind of brake oil detection methods of automobile technical field, especially a kind of to be based on nerve net
Automobile brake noise expert's detection method of network.
Background technique
In the test process to motor vehicle product, a ring is important to the brake detection of brake(-holder) block.In motor vehicle system
It is different depending on the quality of brake block when dynamic, there is probability that can generate brake oil.Other than influencing user experience, brake oil
It is also highly relevant with the quality of brake block to generate probability itself.Therefore, the detection of brake oil also can quality control to brake block
It is formed with referenced.Existing noise detecting method is: 1. have collected brake block vibration and microphone data 2. by human ear to noise
It is checked, the spectrum analysis of Fast Fourier Transform (FFT) is cooperated to judge whether brake block generates noise in braking process, and
The type of noise.This detection process requirement inspector has the industry experience of height, sensitive auditory system, and training cost
It is high；In addition, inspector will often be chronically exposed under noise circumstance, there is great negative shadow to auditory system and physical and mental health
It rings.Therefore a kind of computer assisted noise detection system is needed, inspector's workload and pressure are mitigated, it is reliable to improve detection
Property and consistency.
Summary of the invention
In view of the deficiencies of the prior art, the present invention proposes a kind of automobile brake noise experts neural network based to detect
Method, using the method judgement detection brake oil of neural network, to greatly reduce the workload of inspector, while still
Keep the assay reproducibility and reliability of height.
The present invention is achieved through the following technical solutions, and the present invention is the following steps are included: first, to collected brake
The training of vehicle piece vibration and noise information progress neural network；Second, neural network is carried out to new collected measurement data
Noise measuring；Third, the improvement of neural network, the test result based on second step, EDS extended data set are next time neural to carry out
The repetitive exercise of network.
Further, in the present invention, first step neural network training the following steps are included:
First, from acceleration transducer is arranged in automobile brake test from the brake block of four wheels, in driver's cabin
Cloth microphone obtains the brake block vibration and noise information in braking process；
Second, by professional inspection person to whether thering is noise to be marked in each brake record, by mark information with electricity
Sublist case form, the information for including in electrical form have: audio file pathname, the period for having noiseless, noise to occur.
Third writes script based on Python, is split, classifies to the data set in second step, and training, which is concentrated with, makes an uproar
Sound and muting audio quantitative proportion are maintained at 1:1；
4th, the data set after regular is randomized, and data set is divided into three parts: training set, verifying collection
With test set, training set, to the effect of new data, is avoided for optimizing network parameter, verifying collection for the parameter after verifying optimization
The overfitting problem of model, test set are used for the effect of lateral comparison difference model；
5th, the open source machine learning mould group based on Python, PyTorch constructs the neural network mould judged for noise
Type is optimized with the data the set pair analysis model of preparation in the 4th, is compared, chooses the model of optimal utility.
Further, in the present invention, second step neural network noise measuring the following steps are included:
First, after neural network is chosen, new measurement data is detected；
Second, manually inspect new measurement data by random samples；
Neural network is compared with artificial sampling observation detection, judges the validity of neural network by third.
Further, in the present invention, the spreadsheet in the training second step of neural network is * .csv format
Or * .xlsx format.
Further, in the present invention, training set, verifying collection and test set data in the training third step of neural network
Ratio is 6:2:2.
Compared with prior art, the present invention has the advantages that are as follows: utilizes the method judgement detection system of neural network
Moving noise to greatly reduce the workload of inspector, while still maintaining the assay reproducibility and reliability of height.
Detailed description of the invention
Fig. 1 is flow chart of the invention.
Specific embodiment
It elaborates with reference to the accompanying drawing to the embodiment of the present invention, before the present embodiment is with technical solution of the present invention
It mentions, the detailed implementation method and specific operation process are given, but protection scope of the present invention is not limited to following embodiments.
Embodiment
Shown in specific embodiment Fig. 1.
The first step, the training of neural network:
A, from acceleration transducer is arranged in automobile brake test from the brake block of four wheels, in layont of Interior Cab
Microphone obtains the brake block vibration and noise information in braking process.
B, by professional inspection person to whether thering is noise to be marked in each brake record.By mark information with electronic watch
Case form, as * .csv format or * .xlsx format save.The information for including in electrical form has: audio file pathname has
Noiseless, the period that noise occurs.
C, script is write based on Python, the data set in step b is split, classified.Training be concentrated with noise with
Muting audio quantitative proportion remains at roughly 1:1.
D, the data set after regular is randomized, and data set is divided by three parts with the ratio of 6:2:2: training
Collection, verifying collection and test set.Training set is for optimizing network parameter, and verifying collection is for the parameter after verifying optimization to new data
Effect, avoids the overfitting problem of model, and test set is used for the effect of lateral comparison difference model.
E, the open source machine learning mould group based on Python, PyTorch construct the neural network model judged for noise.
It is optimized with the data the set pair analysis model of preparation in step d, compares, choose the model of optimal utility.Second step, neural network
Noise measuring:
A, after neural network is chosen, new measurement data is detected.
B, new measurement data is manually inspected by random samples
C, neural network is compared with artificial sampling observation detection, judges the validity of neural network.
Third step, the improvement of neural network:
Test result based on second step, EDS extended data set, to carry out the repetitive exercise of neural network next time.
Claims (5)
1. a kind of automobile brake noise expert's detection method neural network based, it is characterised in that the following steps are included: first,
The training of neural network is carried out to collected brake block vibration and noise information；Second, to new collected measurement data into
The noise measuring of row neural network；Third, the improvement of neural network, the test result based on second step, EDS extended data set, with into
The repetitive exercise of capable neural network next time.
2. automobile brake noise expert's detection method neural network based according to claim 1, it is characterised in that institute
State the training of first step neural network the following steps are included:
First, from acceleration transducer is arranged in automobile brake test from the brake block of four wheels, in layont of Interior Cab
Microphone obtains the brake block vibration and noise information in braking process；
Second, by professional inspection person to whether thering is noise to be marked in each brake record, by mark information with electronic watch
Case form, the information for including in electrical form have: audio file pathname, the period for having noiseless, noise to occur.
Third writes script based on Python, is split, classifies to the data set in second step, training be concentrated with noise with
Muting audio quantitative proportion is maintained at 1:1；
4th, the data set after regular is randomized, and data set is divided into three parts: training set, verifying collection and survey
Examination collection, training set, to the effect of new data, avoid model for the parameter after verifying optimization for optimizing network parameter, verifying collection
Overfitting problem, test set be used for lateral comparison difference model effect；
5th, the open source machine learning mould group based on Python, PyTorch constructs the neural network model judged for noise,
It optimized, compared with the data the set pair analysis model of preparation in the 4th, choose the model of optimal utility.
3. automobile brake noise expert's detection method neural network based according to claim 1, it is characterised in that institute
State the noise measuring of second step neural network the following steps are included:
First, after neural network is chosen, new measurement data is detected；
Second, manually inspect new measurement data by random samples；
Neural network is compared with artificial sampling observation detection, judges the validity of neural network by third.
4. automobile brake noise expert's detection method neural network based according to claim 2, it is characterised in that institute
Stating the spreadsheet in second step is * .csv format or * .xlsx format.
5. automobile brake noise expert's detection method neural network based according to claim 2, it is characterised in that institute
State training set in third step, verifying integrates with test set ratio data as 6:2:2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910268162.6A CN110006664A (en) | 2019-04-03 | 2019-04-03 | Automobile brake noise expert's detection method neural network based |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910268162.6A CN110006664A (en) | 2019-04-03 | 2019-04-03 | Automobile brake noise expert's detection method neural network based |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110006664A true CN110006664A (en) | 2019-07-12 |
Family
ID=67169830
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910268162.6A Pending CN110006664A (en) | 2019-04-03 | 2019-04-03 | Automobile brake noise expert's detection method neural network based |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110006664A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111238825A (en) * | 2020-01-10 | 2020-06-05 | 东南大学 | Intelligent driving automatic emergency braking performance testing method for combined test pavement |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1571982A (en) * | 2002-03-26 | 2005-01-26 | 科学与工业研究会 | Improved performance of artificial neural network model in the presence of instrumental noise and measurement error |
| CN202711022U (en) * | 2012-06-26 | 2013-01-30 | 上海好耐电子科技有限公司 | Remote vehicle-mounted data acquisition system |
| CN105825241A (en) * | 2016-04-15 | 2016-08-03 | 长春工业大学 | Driver braking intention identification method based on fuzzy neural network |
| CN108256689A (en) * | 2018-02-06 | 2018-07-06 | 华中科技大学 | A kind of neural network prediction method of non-crystaline amorphous metal thermoplastic forming performance |
| CN109062177A (en) * | 2018-06-29 | 2018-12-21 | 无锡易通精密机械股份有限公司 | A kind of Trouble Diagnostic Method of Machinery Equipment neural network based and system |
| CN109387565A (en) * | 2018-10-12 | 2019-02-26 | 山东理工大学 | A method of brake block internal flaw is detected by analysis voice signal |
| CN109543749A (en) * | 2018-11-22 | 2019-03-29 | 云南大学 | Drawing sentiment analysis method based on deep learning |
-
2019
- 2019-04-03 CN CN201910268162.6A patent/CN110006664A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1571982A (en) * | 2002-03-26 | 2005-01-26 | 科学与工业研究会 | Improved performance of artificial neural network model in the presence of instrumental noise and measurement error |
| CN202711022U (en) * | 2012-06-26 | 2013-01-30 | 上海好耐电子科技有限公司 | Remote vehicle-mounted data acquisition system |
| CN105825241A (en) * | 2016-04-15 | 2016-08-03 | 长春工业大学 | Driver braking intention identification method based on fuzzy neural network |
| CN108256689A (en) * | 2018-02-06 | 2018-07-06 | 华中科技大学 | A kind of neural network prediction method of non-crystaline amorphous metal thermoplastic forming performance |
| CN109062177A (en) * | 2018-06-29 | 2018-12-21 | 无锡易通精密机械股份有限公司 | A kind of Trouble Diagnostic Method of Machinery Equipment neural network based and system |
| CN109387565A (en) * | 2018-10-12 | 2019-02-26 | 山东理工大学 | A method of brake block internal flaw is detected by analysis voice signal |
| CN109543749A (en) * | 2018-11-22 | 2019-03-29 | 云南大学 | Drawing sentiment analysis method based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| 张建南 等: ""基于神经网络方法的车内噪声自适应主动控制"", 《设计 计算 研究》 * |
| 徐鹏: "《监狱智能化安全防范关键技术研究》", 30 December 2017 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111238825A (en) * | 2020-01-10 | 2020-06-05 | 东南大学 | Intelligent driving automatic emergency braking performance testing method for combined test pavement |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9109971B2 (en) | Respiratory condition analysis apparatus, respiratory condition display apparatus, processing method therein, and program | |
| US7892189B2 (en) | Movement analysis display apparatus and movement analyzing method | |
| CN107209512B (en) | Method, system and the computer program in the study stage of the acoustics or vibration analysis for machine | |
| JP6882397B2 (en) | Abnormal noise detection system, equipment, method and program | |
| CN109211546B (en) | Rotary machine fault diagnosis method based on noise reduction automatic encoder and increment learning | |
| CN105841797A (en) | Window motor abnormal noise detection method and apparatus based on MFCC and SVM | |
| CN108305680B (en) | Intelligent Parkinson's disease auxiliary diagnosis method and device based on multivariate biological characteristics | |
| CN110987434A (en) | Rolling bearing early fault diagnosis method based on denoising technology | |
| CN111513730B (en) | Psychological stress prediction method and system based on multichannel physiological data | |
| CN111964909A (en) | Rolling bearing operation state detection method, fault diagnosis method and system | |
| JP2012018066A (en) | Device for inspecting abnormality | |
| CN110632484A (en) | ELM-based GIS partial discharge defect diagnosis and classification system and method | |
| KR20220080815A (en) | Bio-signal-based degenerative brain disease prediction and training system | |
| CN110196098B (en) | Heart rate change-based vehicle sound quality evaluation method | |
| CN110006664A (en) | Automobile brake noise expert's detection method neural network based | |
| JP3966500B2 (en) | Defect inspection system due to abnormal noise | |
| KR102123149B1 (en) | A device for diagnosis of peripheral neuropathy using wavelet transform of needle electromyography signal | |
| CN113796873B (en) | Wearable dynamic electrocardiosignal classification method and system | |
| CN109409216A (en) | Speed adaptive indoor human body detection method based on subcarrier dynamic select | |
| JP3885297B2 (en) | Abnormal sound determination device and abnormal sound determination method | |
| Al-Mulla et al. | Statistical class separation using sEMG features towards automated muscle fatigue detection and prediction | |
| CN107271184B (en) | A kind of the kernel regression decomposition method and system of rolling bearing fault diagnosis | |
| US8326557B2 (en) | Detection of an abnormal signal in a compound sampled | |
| CN105489228A (en) | Rhonchus identification method based on frequency domain image processing | |
| KR20240074822A (en) | How to identify and characterize noise generated by vehicle braking systems using artificial intelligence |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190712 |