CN109318056A - A kind of Tool Wear Monitoring method based on multiple types sensor composite signal - Google Patents
A kind of Tool Wear Monitoring method based on multiple types sensor composite signal Download PDFInfo
- Publication number
- CN109318056A CN109318056A CN201811220952.9A CN201811220952A CN109318056A CN 109318056 A CN109318056 A CN 109318056A CN 201811220952 A CN201811220952 A CN 201811220952A CN 109318056 A CN109318056 A CN 109318056A
- Authority
- CN
- China
- Prior art keywords
- model
- cloud
- tool
- data
- signal
- 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
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23Q—DETAILS, COMPONENTS, OR ACCESSORIES FOR MACHINE TOOLS, e.g. ARRANGEMENTS FOR COPYING OR CONTROLLING; MACHINE TOOLS IN GENERAL CHARACTERISED BY THE CONSTRUCTION OF PARTICULAR DETAILS OR COMPONENTS; COMBINATIONS OR ASSOCIATIONS OF METAL-WORKING MACHINES, NOT DIRECTED TO A PARTICULAR RESULT
- B23Q17/00—Arrangements for observing, indicating or measuring on machine tools
- B23Q17/09—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool
- B23Q17/0904—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool before or after machining
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23Q—DETAILS, COMPONENTS, OR ACCESSORIES FOR MACHINE TOOLS, e.g. ARRANGEMENTS FOR COPYING OR CONTROLLING; MACHINE TOOLS IN GENERAL CHARACTERISED BY THE CONSTRUCTION OF PARTICULAR DETAILS OR COMPONENTS; COMBINATIONS OR ASSOCIATIONS OF METAL-WORKING MACHINES, NOT DIRECTED TO A PARTICULAR RESULT
- B23Q17/00—Arrangements for observing, indicating or measuring on machine tools
- B23Q17/24—Arrangements for observing, indicating or measuring on machine tools using optics or electromagnetic waves
- B23Q17/2452—Arrangements for observing, indicating or measuring on machine tools using optics or electromagnetic waves for measuring features or for detecting a condition of machine parts, tools or workpieces
- B23Q17/2457—Arrangements for observing, indicating or measuring on machine tools using optics or electromagnetic waves for measuring features or for detecting a condition of machine parts, tools or workpieces of tools
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Physics & Mathematics (AREA)
- Optics & Photonics (AREA)
- Machine Tool Sensing Apparatuses (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The present invention relates to a kind of Tool Wear Monitoring methods based on multiple types sensor composite signal.The present invention, can be to avoid the defect that single signal itself is had by oneself by the method for two kinds of signal acquisitions using acoustic emission sensor and the relevant signal message of power sensor acquisition lathe tool wear.Using two kinds of information of coupling of cloud model algorithm science, and the characteristic factor for reflecting tool abrasion in signal can be extracted, model is established using sparse Bayesian method and then predicts tool abrasion, data are modeled using the recognition methods based on SBL, it is optimized using width parameter of the Bayesian matching tracing algorithm to SBL model kernel function, the Accurate Prediction for realizing tool abrasion, improves the efficiency and accuracy of Tool Wear Monitoring.
Description
Technical field
The present invention relates to a kind of Tool Wear Monitoring methods based on multiple types sensor composite signal, belong to tool wear
Detection field.
Background technique
Core one of of the smart machine as wisdom factory, to the Urine scent of operating status, self-teaching and self dimension
Shield ability is its important feature.According to statistics, tool changing and the 20% of operation hours is accounted for about to knife in process.In addition, knife
The abrasion of tool and the damaged personal safety to processing quality, processing efficiency, lathe service life even operator have a major impact.
Therefore, accurately and efficiently cutter operating status Urine scent and automatic early-warning are significant to the level of intelligence for improving lathe,
Can effectively save the cost, improve efficiency
Since the Tool Wear Process in high-speed milling is complicated, model parameter is excessive and it is difficult to predict tool wears, how to pass through
More efficient approach prediction tool wear becomes the focus of researcher.The prediction of domestic and foreign scholars' Cutter wear has been done big
Quantifier elimination, and achieve many progress.
Sound emission (Acoustic emission, AE) technology as advanced detection means to various metal materials and its
Internal microcrack is very sensitive, can find the early changes of material, therefore is widely used in various mechanical equipment detections.But due to
Fault message contained by each sound emission scatterplot is different, therefore also different to assessment equipment status information contribution degree；And each sound emission
The characteristic parameter physical meaning that scatterplot is extracted is different, also different to Fault-Sensitive degree.Certain characteristic parameters can be in failure early stage
It mutates, and certain characteristic parameter variation tendencies are relatively gentle, can not provide early warning for equipment failure state.
Another key problem of tool condition monitoring is to construct effective algorithm on the basis of analyzing signal characteristic and carry out
The prediction of tool abrasion.Algorithms most in use includes artificial neural network and support vector machines etc..The pre- measuring and calculating of artificial neural network
Method, model is excessively complicated, needs a large amount of experiment sample, and calculates convergence difficulties support vector machines and may be implemented in small sample
The prediction of bottom tool abrasion loss, but easily there is overfitting phenomenon, the sparsity of model is limited, and can not provide the general of prediction result
Rate information.
Summary of the invention
In order to solve above-mentioned technical problem, the present invention provides a kind of answering based on multiple types sensor for efficiently and accurately
Close the Tool Wear Monitoring method of signal, the purpose of the present invention is what is be achieved through the following technical solutions: one kind being based on multiple types
The Tool Wear Monitoring method of sensor composite signal, which comprises the steps of:
(1) data acquire: using the acoustic emission signal of acoustic emission sensor acquisition lathe, being added using power sensor acquisition lathe
Work power signal, while being taken pictures using microscope to cutter after the processing period every time, and measure cutter flank
Attrition value obtains tool wear data for comparing；
(2) feature extraction: denoising is carried out using signal of the cloud algorithm to acquisition, interference band is filtered out and extraction and cutter is ground
Then the influence of damage state relevant feature parameters carries out feature extraction to data, it is related to tool abrasion to analyze each feature
Property simultaneously chooses the strong feature of correlation；
(3) classify after constructing model and optimization and prediction: data characteristics after post treatment and being measured microscopically tool abrasion
Data form sample group, are modeled using the recognition methods based on SBL to data, using Bayesian matching tracing algorithm pair
The width parameter of SBL model kernel function optimizes, and realizes the Accurate Prediction of tool abrasion.
Cloud algorithm includes: that Normal Cloud Generator and backward cloud generator calculate when described step (2) feature extraction, respectively
For generating enough water dusts and calculating cloud numerical characteristic (Ex, En, He), wherein Ex is that different abrasion loss signals press the selected phase
Prestige value, En are entropy, and He is super entropy；
The Normal Cloud Generator mainly has following steps:
It is expectation that (2.1.1), which is generated with En, using He^2 as the normal random number En ' of variance；
It is expectation that (2.1.2), which is generated with Ex, using En ' ^2 as the normal random number x of variance；
It is u=exp(-(x-Ex that (2.1.3), which calculates degree of membership namely determination) ^2/2*En ' ^2), then (x, u) is opposite
In a water dust of domain U；Here common " bell " function u=exp(-(x-a is selected) ^2/2*b^2) for subordinating degree function；
The backward cloud generator main process is as follows:
(2.2.1) calculates sample average X and variance S^2
(2.2.2) .Ex=X
(2.2.3) .En=S^2
(2.2.4) .He=sqrt(S^2-En^2)；
Different abrasion loss signals are subjected to cloud model building by selected desired value Ex, entropy En and super entropy He, obtain different mills
The cloud atlas in damage stage.Cloud model is filtered using the uncertainty of impulsive noise, the cloud model after being filtered；Pass through
Backward cloud generator extracts cloud numerical characteristic as the characteristic value that can reflect cutting-tool wear state from the cloud model after filtering,
It is combined into the feature vector that can reflect cutting-tool wear state, feature vector and the cutter wear of the tool flank value of measurement are normalized
Processing, that is, all data are converted to the number between [0,1], data normalization processing generallys use following formula:
。
Classify and predict after step (3) the building model and optimization, detailed process is as follows:
The data of acquisition are divided into two groups, first group is 60 groups of data, for the training to management loading model, in addition
36 groups for testing the accuracy of training pattern；
Based on non-negative least square management loading model, least square method is that unknown ginseng is estimated under linear regression model (LRM)
A kind of several methods, since cloud model is Normal Distribution, least square method is a kind of maximum likelihood estimator module, linear
Model is writeable are as follows:
It is worth vector；Assuming that there are noises for surrounding, according to the content of rarefaction representation, signal model is writeable are as follows:
Wherein e is the noise of ambient enviroment；
The model is exactly to pass through bayes method to learn independent regular parameter, acquires the sparse solution for keeping this formula optimal, i.e. realization knife
Has the Accurate Prediction of abrasion loss.
Beneficial effects of the present invention: the method for the present invention is using acoustic emission sensor and power sensor acquisition machine tool mill
Relevant signal message is damaged, it can be to avoid the defect that single signal itself is had by oneself using the method for two kinds of signal acquisitions.Second, adopting
With two kinds of information of coupling of cloud model algorithm science, and the characteristic factor for reflecting tool abrasion in signal can be extracted, used
Sparse Bayesian method establishes model and then predicts tool abrasion, realizes the monitoring of Cutter wear, improves tool wear
The efficiency and accuracy of monitoring.
Detailed description of the invention
Fig. 1 is the general construction block diagram of the invention.
Fig. 2 is the flow chart of Test Data Collecting of the present invention.
Specific embodiment
A kind of Tool Wear Monitoring method based on multiple types sensor composite signal, includes the following steps:
(1) data acquire: using the acoustic emission signal of acoustic emission sensor acquisition lathe, being added using power sensor acquisition lathe
Work power signal, while being taken pictures using microscope to cutter after the processing period every time, and measure cutter flank
Attrition value obtains tool wear data for comparing；
Specifically: under a certain operating condition, the titanium alloy bar that diameter is 110mm is processed, altogether using 8 same model
Cutter (cutter model: QNMG 090408-NF), acoustic emission sensor is fixed on testing stand cutter hub, more with U.S. PAC
Sound emission data collection system in channel carries out data acquisition acoustic emission signal and power signal.Add to preferably study in difference
The forecasting problem of cutting-tool wear state and abrasion magnitude relation under the conditions of work, using by 3 kinds of cutting parameters (cutting speeds, the amount of feeding
And back engagement of the cutting edge) global combinatorial orthogonal experiment method, under least experiment number arrange multiple groups cutting parameter tested, this
The planning of experiments table of secondary experiment is as follows:
| Cutting speed (m/, min) | The amount of feeding (mm/r) | Back engagement of the cutting edge (mm) | |
| 1 | 120 | 0.2 | 0.3 |
| 2 | 120 | 0.2 | 0.4 |
| 3 | 120 | 0.25 | 0.3 |
| 4 | 120 | 0.25 | 0.4 |
| 5 | 122 | 0.2 | 0.3 |
| 6 | 122 | 0.2 | 0.4 |
| 7 | 122 | 0.25 | 0.3 |
| 8 | 122 | 0.25 | 0.4 |
It is every that a kind of cutting parameter of tool selection is carried out machining to bar, during acquiring cutter from running-in wear to blunt
Acoustic emission signal and power signal.During the cutting process, stop processing every 10s, measure the rear knife of cutter after processing every time
Standard of the surface wear amount as cutting-tool wear state extracts wherein 96 groups of data and its corresponding cutter tool flank wear conduct
Construct the input of Tool Wear Monitoring model；
(2) feature extraction: denoising is carried out using signal of the cloud algorithm to acquisition, interference band is filtered out and extraction and cutter is ground
Then the influence of damage state relevant feature parameters carries out feature extraction to data, it is related to tool abrasion to analyze each feature
Property simultaneously chooses the strong feature of correlation；
Its medium cloud algorithm includes: that Normal Cloud Generator and backward cloud generator calculate, and is respectively intended to generate enough water dusts and meter
It calculates cloud numerical characteristic (Ex, En, He), wherein Ex is that different abrasion loss signals press selected desired value, and En is entropy, and He is super entropy
Value；
Wherein " cloud " or " water dust " is the basic unit of cloud model, and so-called cloud refers to that its one on domain is distributed, can be with
It is indicated with the form (x, u) of joint probability.
The Normal Cloud Generator mainly has following steps:
It is expectation that (2.1.1), which is generated with En, using He^2 as the normal random number En ' of variance；
It is expectation that (2.1.2), which is generated with Ex, using En ' ^2 as the normal random number x of variance；
It is u=exp(-(x-Ex that (2.1.3), which calculates degree of membership namely determination) ^2/2*En ' ^2), then (x, u) is opposite
In a water dust of domain U.Here common " bell " function u=exp(-(x-a is selected) ^2/2*b^2) for subordinating degree function；
The backward cloud generator main process is as follows:
(2.2.1) calculates sample average X and variance S^2
(2.2.2) .Ex=X
(2.2.3) .En=S^2
(2.2.4) .He=sqrt(S^2-En^2)；
The purpose of backward cloud generator is the cloud numerical characteristic for calculating cloud model；
Different abrasion loss signals are subjected to cloud model building by selected desired value Ex, entropy En and super entropy He, obtain different mills
The cloud atlas in damage stage.Cloud model is filtered using the uncertainty that pulse is early demonstrate,proved, the cloud model after being filtered；Pass through
Backward cloud generator extracts cloud numerical characteristic as the characteristic value that can reflect cutting-tool wear state from the cloud model after filtering,
It is combined into the feature vector that can reflect cutting-tool wear state, feature vector and the cutter wear of the tool flank value of measurement are normalized
Processing, that is, all data are converted to the number between [0,1], it is differential the purpose is to cancel the quantity between each dimension data
Not, it avoids causing neural network forecast error larger because inputoutput data order of magnitude difference is excessive, data normalization processing is logical
Frequently with following formula:
。
Entropy reflects the degree of uncertainty of the corresponding qualitativing concept of different wear stages, shows as signal in the abrasion
Stage corresponds to the tolerance interval size of cloud concept.Based on entropy En of the reverse cloud algorithm of no degree of certainty to each cloud concept
It calculates.
It randomly selects difference state of wear reconstruction signal s ' (t) under 4 groups of identical machining conditions and calculates discovery, increase with abrasion loss
Add entropy En in downward trend after first increasing.The entropy at abrasion initial stage, clock signal is smaller, illustrates the cloud concept of clock signal
The range covered is smaller, this is because cutter is very fast in abrasion early period, enters mid-term wear stage quickly；Hereafter, cutter is ground
Damage progresses into mid-term stage, and the entropy of clock signal is gradually increased, and illustrates that the range that concept is covered becomes wide, this is because should
Stage tool wear is slower, and tool wear will enter one and smoothly normally cut the stage；Abrasion continues to aggravate, clock signal
Entropy becomes smaller again, and cloud concept institute's coverage area reduces, this is because crash rate significantly increases, knife when into later period wear stage
Tool abrasion is accelerated.
It is a kind of corresponding relationship that the degree of tool wear is cashed with Ex desired value, can be shown in a sense
The state of wear of cutter flank, tool abrasion increase corresponding desired value and show a increasing trend.When abrasion loss increases in experiment
When, cutter and surface to be machined friction aggravation, power signal increase, and acoustic emission phenomenon enhancing, the uncertainty of signal becomes larger, leads
It causes Ring-down count in the unit time to reduce, desired value Ex is caused to increase.Therefore Ex can reflect tool wear phenomenon-vibration indirectly
Bell counts size.In addition, it is expected that value can most represent qualitativing concept, desired value increase shows the corresponding abrasion loss of clock signal not
It is disconnected to increase.
The uncertainty of qualitative cloud concept entropy is described in super entropy, and numerical value reflects acoustic emission signal sampled data
The randomness of dispersion degree, sample data is also associated by it with ambiguity.
Different abrasion loss signals are subjected to cloud model building by selected desired value Ex, entropy En and super entropy He, no
With wear stage cloud atlas, the characteristics of core water dust is gradually discrete, general normal distribution is integrally presented in cloud atlas is shown, has cloud special
Property.The cloud atlas of the signal completely envelope histogram of voltage magnitude statistical distribution, and ensure that core water dust quantity it is excellent
Gesture ensure that the sample size of conceptual core to the maximum extent, and the unification to different statistical samples may be implemented using cloud characteristic
Modeling.It can be seen that the good results are evident for the tool wear cloud characteristic parameter reaction cutting-tool wear state extracted.
(3) classify after constructing model and optimization and prediction: data characteristics after post treatment and being measured microscopically cutter mill
Damage amount data form sample group, are modeled using the recognition methods based on SBL to data, are tracked and are calculated using Bayesian matching
Method optimizes the width parameter of SBL model kernel function, realizes the Accurate Prediction of tool abrasion；
Management loading is the machine learning for returning and classifying that Tipping is proposed on the basis of support vector machines
Method, SBL use Bayesian inference method, and model has good sparsity, can avoid overfitting phenomenon, while having probability
Predictive ability.
Specifically: the data of acquisition are divided into two groups, first group is 60 groups of data, for pair
The training of management loading model, in addition 36 groups for testing the accuracy of training pattern.The present invention is based on non-
Negative least square management loading model, least square method are that a kind of side of unknown parameter is estimated under linear regression model (LRM)
Method, since cloud model is Normal Distribution, least square method is a kind of maximum likelihood estimator module.Its linear model is writeable
Are as follows:
It is worth vector；Assuming that there are noises for surrounding, according to the content of rarefaction representation, signal model is writeable are as follows:
Wherein e is the noise of ambient enviroment；
The model is exactly to pass through bayes method to learn independent regular parameter, acquires the sparse solution for keeping this formula optimal, i.e. realization knife
Has the Accurate Prediction of abrasion loss.
It is assumed that the use of the length of the signal of acquisition being respectively 128, degree of rarefication K=10 are that the coefficient that signal has a position is not
Zero, the dimension of observing matrix is 64*128, observation noise variance 0.05, initiation parameter 0.1 after feature is submitted.By extraction
Feature vector is trained management loading model as output, obtains as input, the attrition value of cutter flank
Trained management loading model.
Remaining 36 groups in 96 groups of cutting signals of acquisition are used for Tool Wear Monitoring experiment, using the number in step 2
Feature relevant to tool wear is extracted according to extracting method, the input after normalized as sparse Bayesian monitoring model
End, the output end of this monitoring model is the cutter wear of the tool flank value of model prediction.The result shows that this Tool Wear Monitoring mould
The response time of type and accuracy of identification are all satisfied the requirement of on-line monitoring.
The concrete application example of the method for the present invention:
Step 1: in order to test accuracy of this model in actual production process, this experimental monitoring lathe is Shanghai weight
The CK61100 type numerically controlled lathe of type machine tool plant production, cutter for same are the 16ER type lathe tool of Israel Carmex company production, vehicle
Cutting material is aero-engine common used material high temperature alloy, and diameter of work 526mm is axial to cut 0.3mm, straight-cut 0.4mm,
It is processed with constant cut parameter, acquires sound emission when turning and power signal, record a data every 10s, and survey
Measure the corresponding attrition value in the face lathe tool Hou Dao.Acquire 46 groups of valid data altogether, will wherein 30 groups be used as training, remaining 16 groups for surveying
Examination.
Step 2: feature extraction is established cloud model and is filtered to the tool wear signal of acquisition, and passes through positive cloud hair
Raw device establishes cloud atlas, and backward cloud generator extracts cloud feature vector relevant to cutting-tool wear state.And the cutter that will have been measured
Wear of the tool flank value and the feature vector of extraction are normalized.
Step 3: using the feature vector for containing cutting-tool wear state as inputting, using cutter wear of the tool flank value as defeated
Out, sparse Bayesian model is trained, other 16 groups of data as test sample is done into feature extraction, and by extraction
Feature vector imports in the management loading model trained, and the attrition value of prediction and the attrition value of measurement are carried out pair
Than as shown in the table:
| Data number | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
| Attrition value actual measurement | 0.162 | 0.226 | 0.243 | 0.248 | 0.249 | 0.276 | 0.286 | 0.292 |
| Predicted value/mm | 0.160 | 0.225 | 0.243 | 0.248 | 0.250 | 0.268 | 0.288 | 0.291 |
| Data number | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 |
| Attrition value actual measurement | 0.190 | 0.255 | 0.256 | 0.274 | 0.292 | 0.293 | 0.315 | 0.343 |
| Predicted value/mm | 0.190 | 0.254 | 0.256 | 0.275 | 0.287 | 0.289 | 0.311 | 0.343 |
Tool wear predicted value is compared with actual measured value, after practical operation, the prediction of cutter wear of the tool flank value
Value accuracy rate reaches 92%, meets the needs of actual production on-line monitoring.
Claims (3)
1. a kind of Tool Wear Monitoring method based on multiple types sensor composite signal, which comprises the steps of:
(1) data acquire: using the acoustic emission signal of acoustic emission sensor acquisition lathe, being added using power sensor acquisition lathe
Work power signal, while being taken pictures using microscope to cutter after the processing period every time, and measure cutter flank
Attrition value obtains tool wear data for comparing；
(2) feature extraction: denoising is carried out using signal of the cloud algorithm to acquisition, interference band is filtered out and extraction and cutter is ground
Then the influence of damage state relevant feature parameters carries out feature extraction to data, it is related to tool abrasion to analyze each feature
Property simultaneously chooses the strong feature of correlation；
(3) classify after constructing model and optimization and prediction: data characteristics after post treatment and being measured microscopically tool abrasion
Data form sample group, are modeled using the recognition methods based on SBL to data, using Bayesian matching tracing algorithm pair
The width parameter of SBL model kernel function optimizes, and realizes the Accurate Prediction of tool abrasion.
2. a kind of Tool Wear Monitoring method based on multiple types sensor composite signal as described in claim 1, feature
It is, cloud algorithm includes: that Normal Cloud Generator and backward cloud generator calculate when the step (2) feature extraction, is used respectively
To generate enough water dusts and calculate cloud numerical characteristic (Ex, En, He), wherein Ex is that different abrasion loss signals press selected expectation
Value, En are entropy, and He is super entropy；
The Normal Cloud Generator mainly has following steps:
It is expectation that (2.1.1), which is generated with En, using He^2 as the normal random number En ' of variance；
It is expectation that (2.1.2), which is generated with Ex, using En ' ^2 as the normal random number x of variance；
(2.1.3) calculate degree of membership namely determination be u=exp(-(x-Ex) ^2/2*En ' ^2), then (x, u) be relative to
A water dust of domain U；Here common " bell " function u=exp(-(x-a is selected) ^2/2*b^2) for subordinating degree function；
The backward cloud generator main process is as follows:
(2.2.1) calculates sample average X and variance S^2
(2.2.2) .Ex=X
(2.2.3) .En=S^2
(2.2.4) .He=sqrt(S^2-En^2)；
Different abrasion loss signals are subjected to cloud model building by selected desired value Ex, entropy En and super entropy He, obtain different mills
The cloud atlas in damage stage；Cloud model is filtered using the uncertainty that pulse is early demonstrate,proved, the cloud model after being filtered；Pass through
Backward cloud generator extracts cloud numerical characteristic as the characteristic value that can reflect cutting-tool wear state from the cloud model after filtering,
It is combined into the feature vector that can reflect cutting-tool wear state, feature vector and the cutter wear of the tool flank value of measurement are normalized
Processing, that is, all data are converted to the number between [0,1], data normalization processing generallys use following formula:
。
3. a kind of Tool Wear Monitoring method based on multiple types sensor composite signal as claimed in claim 1 or 2, special
Sign is that the step (3) is classified after constructing model and optimization and prediction, and detailed process is as follows:
The data of acquisition are divided into two groups, first group is 60 groups of data, for the training to management loading model, in addition
36 groups for testing the accuracy of training pattern；
Based on non-negative least square management loading model, least square method is that unknown ginseng is estimated under linear regression model (LRM)
A kind of several methods, since cloud model is Normal Distribution, least square method is a kind of maximum likelihood estimator module, linear
Model is writeable are as follows:
It is worth vector；Assuming that there are noises for surrounding, according to the content of rarefaction representation, signal model is writeable are as follows:
Wherein e is the noise of ambient enviroment；
The model is exactly to pass through bayes method to learn independent regular parameter, acquires the sparse solution for keeping this formula optimal, i.e. realization knife
Has the Accurate Prediction of abrasion loss.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710990870.1A CN107553219A (en) | 2017-10-23 | 2017-10-23 | A kind of Tool Wear Monitoring method based on multiple types sensor composite signal |
| CN2017109908701 | 2017-10-23 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109318056A true CN109318056A (en) | 2019-02-12 |
Family
ID=60985862
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710990870.1A Pending CN107553219A (en) | 2017-10-23 | 2017-10-23 | A kind of Tool Wear Monitoring method based on multiple types sensor composite signal |
| CN201811220952.9A Pending CN109318056A (en) | 2017-10-23 | 2018-10-19 | A kind of Tool Wear Monitoring method based on multiple types sensor composite signal |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710990870.1A Pending CN107553219A (en) | 2017-10-23 | 2017-10-23 | A kind of Tool Wear Monitoring method based on multiple types sensor composite signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN107553219A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110046427A (en) * | 2019-04-15 | 2019-07-23 | 中南大学 | Based on orthogonal design and normal cloud model Machine-made Sand T beam concrete match ratio ratio method |
| CN110286649A (en) * | 2019-06-28 | 2019-09-27 | 江苏南高智能装备创新中心有限公司 | For the modeling and its method of sensor acquisition data on numerically-controlled machine tool |
| CN110472635A (en) * | 2019-07-19 | 2019-11-19 | 西北工业大学 | Tool characteristics recognition methods based on deep learning |
| CN112192319A (en) * | 2020-09-28 | 2021-01-08 | 上海交通大学 | Tool wear monitoring method and system of unsupervised model |
| CN112241145A (en) * | 2019-07-16 | 2021-01-19 | 神通资讯科技股份有限公司 | Abnormality detection system and method for automatic processing machine |
| CN111230159B (en) * | 2020-03-02 | 2021-04-16 | 西南交通大学 | Multi-sensor fusion turning tool state monitoring method and system |
| CN113941901A (en) * | 2020-07-17 | 2022-01-18 | 智能云科信息科技有限公司 | Machine tool cutter monitoring method and device and electronic equipment |
| CN114800039A (en) * | 2022-04-08 | 2022-07-29 | 山东大学 | Characteristic strengthening method and system for on-line monitoring of state of milling cutter of thin-wall part |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108427375B (en) * | 2018-04-11 | 2020-10-27 | 温州大学 | Method for monitoring cutter state based on band-pass filtering processing multi-sensor |
| CN108536095B (en) * | 2018-04-24 | 2020-01-17 | 湖北文理学院 | Real-time prediction method for lead screw abrasion loss |
| CN110032981B (en) * | 2019-04-19 | 2022-07-26 | 电子科技大学 | Rotary machine fault identification method based on improved support vector machine |
| CN111890125B (en) * | 2020-06-30 | 2021-06-22 | 厦门嵘拓物联科技有限公司 | Cutter state online monitoring method and management system |
| CN113084237B (en) * | 2021-04-16 | 2022-07-29 | 贵州大学 | Method and device for predicting wear value of milling cutter, electronic device and storage medium |
| CN114880814B (en) * | 2022-07-08 | 2022-09-20 | 南通恒强轧辊有限公司 | Roller remanufacturing auxiliary optimization method based on big data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130179624A1 (en) * | 2012-01-09 | 2013-07-11 | Timothy M. Lambert | Systems and methods for tracking and managing non-volatile memory wear |
| CN103295068A (en) * | 2013-06-09 | 2013-09-11 | 常州信息职业技术学院 | Network course satisfaction evaluation system and method based on cloud model technology |
| CN103335637A (en) * | 2013-06-03 | 2013-10-02 | 南京大学 | Hydrological sequence extension method based on wavelet-cloud model |
| CN104361529A (en) * | 2014-11-04 | 2015-02-18 | 海南电网有限责任公司 | Reliability detecting and evaluating method of power distribution system on basis of cloud model |
-
2017
- 2017-10-23 CN CN201710990870.1A patent/CN107553219A/en active Pending
-
2018
- 2018-10-19 CN CN201811220952.9A patent/CN109318056A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130179624A1 (en) * | 2012-01-09 | 2013-07-11 | Timothy M. Lambert | Systems and methods for tracking and managing non-volatile memory wear |
| CN103335637A (en) * | 2013-06-03 | 2013-10-02 | 南京大学 | Hydrological sequence extension method based on wavelet-cloud model |
| CN103295068A (en) * | 2013-06-09 | 2013-09-11 | 常州信息职业技术学院 | Network course satisfaction evaluation system and method based on cloud model technology |
| CN104361529A (en) * | 2014-11-04 | 2015-02-18 | 海南电网有限责任公司 | Reliability detecting and evaluating method of power distribution system on basis of cloud model |
Non-Patent Citations (3)
| Title |
|---|
| 关山等: "车削刀具磨损声发射信号的云特征分析", 《农业工程学报》 * |
| 方平: "数控机床刀具磨损的多传感器在线实时监测的研究", 《工程科技Ⅰ辑》 * |
| 谢楠等: "基于功率传感器的刀具磨损量预测方法", 《同济大学学报（自然科学版）》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110046427A (en) * | 2019-04-15 | 2019-07-23 | 中南大学 | Based on orthogonal design and normal cloud model Machine-made Sand T beam concrete match ratio ratio method |
| CN110046427B (en) * | 2019-04-15 | 2021-11-19 | 中南大学 | Orthogonal design and normal cloud model based sand T-beam concrete mixing proportion method |
| CN110286649A (en) * | 2019-06-28 | 2019-09-27 | 江苏南高智能装备创新中心有限公司 | For the modeling and its method of sensor acquisition data on numerically-controlled machine tool |
| CN112241145A (en) * | 2019-07-16 | 2021-01-19 | 神通资讯科技股份有限公司 | Abnormality detection system and method for automatic processing machine |
| CN110472635A (en) * | 2019-07-19 | 2019-11-19 | 西北工业大学 | Tool characteristics recognition methods based on deep learning |
| CN111230159B (en) * | 2020-03-02 | 2021-04-16 | 西南交通大学 | Multi-sensor fusion turning tool state monitoring method and system |
| CN113941901A (en) * | 2020-07-17 | 2022-01-18 | 智能云科信息科技有限公司 | Machine tool cutter monitoring method and device and electronic equipment |
| CN113941901B (en) * | 2020-07-17 | 2024-04-23 | 智能云科信息科技有限公司 | Machine tool cutter monitoring method, machine tool cutter monitoring device and electronic equipment |
| CN112192319A (en) * | 2020-09-28 | 2021-01-08 | 上海交通大学 | Tool wear monitoring method and system of unsupervised model |
| CN114800039A (en) * | 2022-04-08 | 2022-07-29 | 山东大学 | Characteristic strengthening method and system for on-line monitoring of state of milling cutter of thin-wall part |
| CN114800039B (en) * | 2022-04-08 | 2023-03-14 | 山东大学 | Characteristic strengthening method and system for on-line monitoring of state of milling cutter of thin-wall part |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107553219A (en) | 2018-01-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109318056A (en) | A kind of Tool Wear Monitoring method based on multiple types sensor composite signal | |
| Madhusudana et al. | Condition monitoring of face milling tool using K-star algorithm and histogram features of vibration signal | |
| CN109514349B (en) | Tool wear state monitoring method based on vibration signal and Stacking integrated model | |
| CN107378641B (en) | A kind of Monitoring Tool Wear States in Turning based on characteristics of image and LLTSA algorithm | |
| KozłowsKi et al. | Assessment model of cutting tool condition for real-time supervision system | |
| Li et al. | Similarity-measured isolation forest: Anomaly detection method for machine monitoring data | |
| CN111037365B (en) | Cutter state monitoring data set enhancing method based on generative countermeasure network | |
| CN110059442B (en) | Turning tool changing method based on part surface roughness and power information | |
| Ali et al. | Acoustic emission signal analysis and artificial intelligence techniques in machine condition monitoring and fault diagnosis: a review | |
| CN113255848B (en) | Water turbine cavitation sound signal identification method based on big data learning | |
| WO2016004749A1 (en) | Method for recognizing tool abrasion degree of large numerical control milling machine | |
| CN111366123A (en) | Part surface roughness and cutter wear prediction method based on multi-task learning | |
| Wang et al. | Force sensor based online tool wear monitoring using distributed Gaussian ARTMAP network | |
| CN111782624A (en) | Machine tool cutter vibration monitoring and analyzing method based on artificial intelligence and big data | |
| CN112766182A (en) | Method for identifying wear state of broaching tool | |
| CN112207631B (en) | Method for generating tool detection model, method, system, device and medium for detecting tool detection model | |
| Madhusudana et al. | Use of discrete wavelet features and support vector machine for fault diagnosis of face milling tool | |
| Wei et al. | Tool wear state recognition based on feature selection method with whitening variational mode decomposition | |
| Du et al. | Intelligent turning tool monitoring with neural network adaptive learning | |
| CN114429152A (en) | Rolling bearing fault diagnosis method based on dynamic index antagonism self-adaption | |
| Kostrzewska et al. | The classical and Bayesian logistic regression in the research on the financial standing of enterprises after bankruptcy in Poland | |
| CN110930012B (en) | Energy consumption abnormity positioning method based on sensitivity analysis and improved negative selection method | |
| Zhao et al. | Performance prediction methodology based on pattern recognition | |
| Kothuru et al. | Application of audible sound signals for tool wear monitoring and workpiece hardness identification in gear milling using machine learning techniques | |
| Martins et al. | Neural networks models for wear patterns recognition of single-point dresser |
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: 20190212 |