CN112733289A - Novel machine learning method for diagnosing motor bearing fault based on multi-scale permutation entropy - Google Patents
Novel machine learning method for diagnosing motor bearing fault based on multi-scale permutation entropy Download PDFInfo
- Publication number
- CN112733289A CN112733289A CN202011617413.6A CN202011617413A CN112733289A CN 112733289 A CN112733289 A CN 112733289A CN 202011617413 A CN202011617413 A CN 202011617413A CN 112733289 A CN112733289 A CN 112733289A
- Authority
- CN
- China
- Prior art keywords
- data
- bearing
- permutation entropy
- machine learning
- motor
- 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
- 238000010801 machine learning Methods 0.000 title claims abstract description 14
- 238000000034 method Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 12
- 238000003745 diagnosis Methods 0.000 claims description 11
- 238000013145 classification model Methods 0.000 claims description 9
- 238000012795 verification Methods 0.000 claims description 8
- 238000010200 validation analysis Methods 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 230000006835 compression Effects 0.000 claims description 4
- 238000007906 compression Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 239000013598 vector Substances 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 abstract description 3
- 238000005070 sampling Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000005096 rolling process Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Geometry (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to the technical field of bearing predictive maintenance products, and discloses a novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy, which solves the problem that multi-scale parameters of permutation entropy in the current market are difficult to select.
Description
Technical Field
The invention belongs to the technical field of bearing predictive maintenance, and particularly relates to a novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy.
Background
The bearing is one of the most widely used mechanical parts in rotating equipment such as a motor and the like, and is also the most easily-failed key part, statistics is made that in the rotating equipment using a rolling bearing, about 30% of mechanical failures are caused by the rolling bearing, and various potential failures generated by the rolling bearing comprise inner ring failures, outer ring failures, ball failures, retainer failures and the like, so that the failure diagnosis of the bearing plays a very important role in motor predictive maintenance, the main problem currently encountered in the bearing failure diagnosis is that abnormal data are difficult to obtain, so that the generally-calibrated failure data quantity is small, the classical method is based on spectrum analysis and diagnosis is carried out through the inherent failure frequency of the bearing, but the method needs additional parameters of the bearing and has larger diagnosis errors and is difficult to popularize, and the deep learning method is based on the limitation of the failure data quantity and is difficult to train an accurate model, therefore, machine learning methods are more preferred, but accuracy is more improved.
Disclosure of Invention
Aiming at the situation, in order to overcome the defects of the prior art, the invention provides a novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy, and the problem that multi-scale parameters of the permutation entropy in the current market are difficult to select is effectively solved.
In order to achieve the purpose, the invention provides the following technical scheme: a novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy comprises the following steps:
s1, preparing motor bearings and motor bearing vibration data, wherein the motor bearings and the motor bearing vibration data comprise bearing outer ring faults, inner ring faults and normal bearing vibration data, and the calibration 0 is normal bearings, 1 is outer ring faults and 2 is inner ring faults;
s2, selecting parameters of the multi-scale permutation entropy, performing feature extraction on data training data, and performing feature dimension compression by using PCA, wherein the parameters of the multi-scale permutation entropy include several important parameters:
firstly, data points N are the number of data points used for calculating a feature vector each time;
embedding dimension m, wherein the embedding dimension m is a dimension for calculating the arrangement entropy of a certain time point;
time delay t, which is the interval between the data points used;
fourthly, performing s times of downsampling on the original data, and calculating the characteristic of one permutation entropy after each downsampling;
s3, carrying out classification model training by Bayes, and determining a lower classification model parameter;
s4, carrying out Bayesian model verification by using the test set;
s5, collecting multiple segments of data in actual use, carrying out Bayesian model prediction on single data, and carrying out model result fusion by a majority voting method;
and S6, outputting the final diagnosis result and recording the numerical value.
Preferably, the model verification is trained and verified by using data of the bearing race MFPT, and comprises fault data and normal data of different loads under the condition of 25 Hz rotating speed of the motor.
Preferably, the model validation uses half of the data for training and half of the data for model validation.
Preferably, the motor vibration data acquisition can be divided into n segments of data of 1 second, and then the single data of 1 second is subjected to the fault classification judgment.
Compared with the prior art, the invention has the beneficial effects that:
1) the invention simplifies the selection of key scale parameters of the multi-scale permutation entropy, compresses the characteristic dimensions through PCA, simplifies the dimensions and enables the classifier to be easier to learn, combines the Bayes classifier and the majority voting method to combine to enable the bearing fault classification to be more accurate, has stronger practicability, and is very suitable for the embedded application of the edge end in view of the size of the Bayes model, can enable hardware to enable the abnormal detection of the bearing edge, and avoids a large amount of data to be transmitted back to a server.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a diagram illustrating the results of model verification according to the present invention;
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments; all other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
A novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy comprises the following steps:
s1, preparing vibration data of a motor bearing and a motor bearing, wherein the vibration data comprises bearing outer ring faults, inner ring faults and normal bearing vibration data, the sampling rate of a used vibration sensor is generally higher than 12000 Hz, the early fault frequency of the bearing is generally between 30K and 40K Hz, and if the vibration data is used for early bearing fault diagnosis, the sampling rate of the vibration data needing to be prepared is generally higher than 100K Hz, wherein the standard 0 is the normal bearing, 1 is the outer ring fault, and 2 is the inner ring fault;
s2, selecting parameters of the multi-scale permutation entropy, performing feature extraction on data training data, and performing feature dimension compression by using PCA, wherein the parameters of the multi-scale permutation entropy include several important parameters:
a data point N, which is the number of data points used to calculate one eigenvector at a time, may be a sampling rate based on the vibration data, and if the sampling rate is 12000 hz, a 1 second time window, for example, may be used, and then N equals 12000.
And secondly, embedding dimension m, wherein the embedding dimension m is a dimension for calculating the permutation entropy at a certain time point, and all permutation combinations of the dimension m are calculated, so that the relation between the calculated amount and the dimension is exponential, the general selection range is 4-8, and the accuracy is not improved too much when the dimension is higher than 8.
And thirdly, time delay t is the interval between the data points, the time delay t is not sensitive to the bearing diagnosis result, and can be selected to be 1, namely continuous data are used, and no interval exists in the middle.
And fourthly, performing s times of downsampling on original data, calculating the characteristic of an arrangement entropy after each time of downsampling, wherein the variable plays a key role in expressing the characteristic of the multi-scale arrangement entropy, and no good selection rule exists at present. The method has the advantages that redundant dimensions can be eliminated, the features are compressed and principal component feature vectors are mutually orthogonal, the features enable a machine learning classifier to be easier to train, s is 16 in the method, and then the first 8 principal component variables are used after principal component analysis.
S3, Bayes is used for training a classification model, the parameters of the lower classification model are determined, 0 is used for representing a normal bearing in the data calibration method, 1 represents an outer ring fault, 2 represents an inner ring fault, the model is trained until Bayes convergence, a multi-classification model is output, after the model parameters are determined through training, vibration data with the length of N is input, and the classification result and the confidence coefficient are output.
And S4, carrying out Bayesian model verification by using the test set.
And S5, collecting multiple segments of data in actual use, carrying out Bayesian model prediction on single data, and carrying out model result fusion by using a majority voting method.
And S6, outputting the final diagnosis result and recording the numerical value.
Further, the model verification is trained and verified by using data of the bearing race MFPT, and comprises fault data and normal data of different loads under the condition that the motor rotates at 25 Hz.
Further, model validation uses half of the data for training and half of the data for model validation, see fig. 1 below for the validation results.
Further, the motor vibration data acquisition can be divided into n sections of data of 1 second, then the fault classification judgment is carried out on single data of 1 second, the success rate of certain fault classification is assumed to be p, and then the worst condition of 3 types of faults in the n sections, wherein the fault is n/3. At least half of n/3 predicts the correct rate and is 1-binocdf (n/6, n/3, p), binocdf is binomial cumulative distribution function, the bearing diagnoses in the test data of MFPT as normal with the accuracy rate p being 0.82, when n being 30, it is assumed that the diagnosis is normal and 11 times, the correct rate of 6 times of 11 times is 96.66%, so the prediction result fusion reaches a high accuracy rate by collecting multiple data, summarizing based on a relatively high p and a relatively large n, the accuracy rate can be obviously improved by the majority voting method.
In summary, the following steps: the invention provides a novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy, which comprises the steps of preparing motor bearing and motor bearing vibration data comprising bearing outer ring faults, inner ring faults and normal bearing vibration data, calibrating 0 to be a normal bearing, 1 to be an outer ring fault and 2 to be an inner ring fault, then selecting parameters of the multi-scale permutation entropy, carrying out feature extraction on data training data, carrying out feature dimension compression by PCA, carrying out classification model training by Bayes, determining lower classification model parameters, carrying out Bayes model verification by a test set, then collecting multiple segments of data in actual use, carrying out Bayes model prediction on single data, carrying out model result fusion by a majority voting method, finally outputting a final diagnosis result and recording the result, and in conclusion, simplifying the selection of key scale parameters of the multi-scale permutation entropy, the characteristic dimensionality is compressed through PCA, the dimensionality is simplified, the classifier is easier to learn, bearing fault classification is more accurate and the practicability is higher by combining the Bayes classifier and a majority voting method, and in view of the size of the Bayes model, the invention is very suitable for embedded application of an edge end, can enable hardware to enable abnormal detection of the edge of a bearing, and avoids a large amount of data from being transmitted back to a server.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments or portions thereof without departing from the spirit and scope of the invention.
Claims (4)
1. A novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy is characterized by comprising the following steps of: the method comprises the following steps:
s1, preparing motor bearings and motor bearing vibration data, wherein the motor bearings and the motor bearing vibration data comprise bearing outer ring faults, inner ring faults and normal bearing vibration data, and the calibration 0 is normal bearings, 1 is outer ring faults and 2 is inner ring faults;
s2, selecting parameters of the multi-scale permutation entropy, performing feature extraction on data training data, and performing feature dimension compression by using PCA, wherein the parameters of the multi-scale permutation entropy include several important parameters:
firstly, data points N are the number of data points used for calculating a feature vector each time;
embedding dimension m, wherein the embedding dimension m is a dimension for calculating the arrangement entropy of a certain time point;
time delay t, which is the interval between the data points used;
fourthly, performing s times of downsampling on the original data, and calculating the characteristic of one permutation entropy after each downsampling;
s3, carrying out classification model training by Bayes, and determining a lower classification model parameter;
s4, carrying out Bayesian model verification by using the test set;
s5, collecting multiple segments of data in actual use, carrying out Bayesian model prediction on single data, and carrying out model result fusion by a majority voting method;
and S6, outputting the final diagnosis result and recording the numerical value.
2. The novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy as claimed in claim 1, wherein: the model verification is trained and verified by using data of the bearing race MFPT, and comprises fault data and normal data of different loads under the condition of 25 Hz rotating speed of the motor.
3. The novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy as claimed in claim 1, wherein: model validation half of the data was used for training and half for model validation.
4. The novel machine learning method for diagnosing motor bearing faults based on multi-scale permutation entropy as claimed in claim 1, wherein: the motor vibration data acquisition can be divided into n segments of data of 1 second, then the fault classification judgment is carried out on the single data of 1 second, and the accuracy is improved by fusing through a majority voting method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011617413.6A CN112733289A (en) | 2020-12-31 | 2020-12-31 | Novel machine learning method for diagnosing motor bearing fault based on multi-scale permutation entropy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011617413.6A CN112733289A (en) | 2020-12-31 | 2020-12-31 | Novel machine learning method for diagnosing motor bearing fault based on multi-scale permutation entropy |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112733289A true CN112733289A (en) | 2021-04-30 |
Family
ID=75611186
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011617413.6A Pending CN112733289A (en) | 2020-12-31 | 2020-12-31 | Novel machine learning method for diagnosing motor bearing fault based on multi-scale permutation entropy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112733289A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104849050A (en) * | 2015-06-02 | 2015-08-19 | 安徽工业大学 | Rolling bearing fault diagnosis method based on composite multi-scale permutation entropies |
| CN107101813A (en) * | 2017-04-26 | 2017-08-29 | 河北工业大学 | A kind of frame-type circuit breaker mechanical breakdown degree assessment method based on vibration signal |
| CN111738309A (en) * | 2020-06-03 | 2020-10-02 | 哈尔滨工业大学 | Gas sensor fault mode identification method based on multi-scale analysis and integrated learning |
-
2020
- 2020-12-31 CN CN202011617413.6A patent/CN112733289A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104849050A (en) * | 2015-06-02 | 2015-08-19 | 安徽工业大学 | Rolling bearing fault diagnosis method based on composite multi-scale permutation entropies |
| CN107101813A (en) * | 2017-04-26 | 2017-08-29 | 河北工业大学 | A kind of frame-type circuit breaker mechanical breakdown degree assessment method based on vibration signal |
| CN111738309A (en) * | 2020-06-03 | 2020-10-02 | 哈尔滨工业大学 | Gas sensor fault mode identification method based on multi-scale analysis and integrated learning |
Non-Patent Citations (2)
| Title |
|---|
| 郑近德等: "多尺度排列熵及其在滚动轴承故障诊断中的应用", 《中国机械工程》 * |
| 陈鹏等: "基于多尺度排列熵和改进多分类相关向量机的滚动轴承故障诊断方法", 《电子测量与仪器学报》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111914883B (en) | Spindle bearing state evaluation method and device based on deep fusion network | |
| CN107657250B (en) | Bearing fault detection and positioning method and detection and positioning model implementation system and method | |
| CN110376522B (en) | Motor fault diagnosis method of data fusion deep learning network | |
| CN110807257A (en) | Method for predicting residual life of aircraft engine | |
| CN112184654A (en) | High-voltage line insulator defect detection method based on generation countermeasure network | |
| CN110610035A (en) | Rolling bearing residual life prediction method based on GRU neural network | |
| CN112631240A (en) | Spacecraft fault active detection method and device | |
| CN110163075A (en) | A kind of multi-information fusion method for diagnosing faults based on Weight Training | |
| US20220334573A1 (en) | Sensor-agnostic mechanical machine fault identification | |
| CN113538353B (en) | Five-phase asynchronous motor rolling bearing fault diagnosis method based on single-channel diagram data enhancement and migration training residual error network | |
| CN113569990B (en) | Strong noise interference environment-oriented performance equipment fault diagnosis model construction method | |
| CN117076935B (en) | Digital twin-assisted mechanical fault data lightweight generation method and system | |
| Sadoughi et al. | A deep learning approach for failure prognostics of rolling element bearings | |
| Bao et al. | Remaining useful life assessment of slewing bearing based on spatial-temporal sequence | |
| CN112819029A (en) | Novel machine learning method for diagnosing gearbox fault based on multi-scale permutation entropy | |
| CN110956112B (en) | Novel high-reliability slewing bearing service life assessment method | |
| CN112733289A (en) | Novel machine learning method for diagnosing motor bearing fault based on multi-scale permutation entropy | |
| CN116662800A (en) | Rolling bearing fault diagnosis method based on self-adaptive attention mechanism | |
| CN114564877B (en) | Rolling bearing life prediction method, system, equipment and readable storage medium | |
| CN113537288B (en) | Method and system for training target model based on modification of sample signal | |
| CN113487141B (en) | Bearing state evaluation method based on multisource information clustering fusion | |
| CN112798279B (en) | Novel detection method for diagnosing motor bearing fault | |
| Shao et al. | Multi-signal Joint HVCB Fault Diagnosis Research Based on Adaptive Framing MFCC Feature Extraction Method | |
| CN113051809A (en) | Virtual health factor construction method based on improved restricted Boltzmann machine | |
| CN112949035A (en) | Fault diagnosis method and device for sensor |
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: 20210430 |