CN112417954B - Bearing fault mode diagnosis method and system for small sample data set - Google Patents
Bearing fault mode diagnosis method and system for small sample data set Download PDFInfo
- Publication number
- CN112417954B CN112417954B CN202011091094.XA CN202011091094A CN112417954B CN 112417954 B CN112417954 B CN 112417954B CN 202011091094 A CN202011091094 A CN 202011091094A CN 112417954 B CN112417954 B CN 112417954B
- Authority
- CN
- China
- Prior art keywords
- bearing
- dimensional
- matching
- module
- small sample
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Library & Information Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Acoustics & Sound (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Signal Processing (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to a bearing fault mode diagnosis method and system for a small sample data set, wherein the method comprises the following steps: 1) Collecting vibration signal data of a bearing under different equipment operating conditions through an acceleration sensor; 2) Preprocessing the signals, converting the original one-dimensional signals into two-dimensional signals through a continuous wavelet transform algorithm, and forming image data; 3) Constructing a bearing fault diagnosis model frame based on a convolutional neural network, wherein the frame comprises a coding module and a matching module, randomly sampling from image data, and constructing a plurality of learning tasks of small sample sets so as to train the model; 4) And acquiring a vibration signal of a target bearing, and diagnosing a bearing fault mode according to the preprocessing method and the bearing fault diagnosis model. Compared with the prior art, the method combines the deep learning algorithm and the meta learning algorithm, and can improve the diagnosis precision under the condition of insufficient data quantity.
Description
Technical Field
The invention relates to the technical field of high-end equipment structure fault diagnosis, in particular to a bearing fault mode diagnosis method and system for a small sample data set.
Background
The rolling bearing is used as a key part in modern high-end equipment, has weak impact bearing capacity and is extremely easy to fatigue and damage. Once a fault occurs, the whole production process is greatly and negatively affected, so that not only is serious economic loss caused, but also the life safety of related personnel is even endangered. Therefore, it is necessary to conduct a technical study of fault diagnosis for the rolling bearing, and the technical study is of great significance for predictive maintenance of high-end equipment.
At present, many fault diagnosis technologies based on machine learning and even deep learning exist, such as a support vector machine, a random forest, a gradient lifting tree, a cyclic neural network, a boltzmann machine and the like, but all of them need to be supported by enough samples, and a good effect can be shown only when the training set and the test set samples are distributed in a consistent manner. In an actual production environment, however, it is difficult to satisfy the above two requirements. On one hand, in the initial stage of equipment operation, the mechanical fault condition is less, and the number of samples is difficult to support model training; on the other hand, the operation condition of the equipment is complex and variable, and the collected data are often in different working conditions, which also brings great challenges to the prediction accuracy of the model.
Unlike the conventional machine learning and deep learning methods, meta learning focuses more on how to quickly adapt to learning of a new task by using known knowledge, and thus can effectively solve the two problems. Meta-learning starts to rise in recent years and plays a remarkable role in solving the learning problem of a small amount of labeled samples and even unlabeled sample data. But the application of the method in the field of high-end equipment fault diagnosis is quite lacking at present.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide a bearing fault mode diagnosis method and system for a small sample data set.
The purpose of the invention can be realized by the following technical scheme:
a bearing fault mode diagnosis method for a small sample data set comprises the following steps:
step 1: collecting vibration signal data of the bearing under different equipment operating conditions through an acceleration sensor, and storing the vibration signal data into a server;
step 2: preprocessing the signals in a server, converting original one-dimensional signals into two-dimensional signals through a continuous wavelet transform algorithm, and storing the two-dimensional signals in a database in the form of images;
and step 3: constructing a bearing fault diagnosis model frame based on a convolutional neural network, wherein the bearing fault diagnosis model frame comprises a coding module and a matching module, randomly sampling from image data in a database, and constructing a plurality of learning tasks of small sample data sets so as to train a bearing fault diagnosis model;
and 4, step 4: and (3) acquiring a vibration signal of the target bearing, diagnosing the vibration signal according to the preprocessing method in the step (2) and the bearing fault diagnosis model frame in the step (3) and obtaining a bearing fault mode.
Further, the step 2 comprises the following sub-steps:
step 201: vibration data collected by the acceleration sensor are one-dimensional continuous time sequence signals, and the signals are preprocessed through a continuous wavelet transform algorithm to obtain two-dimensional signals;
step 202: the two-dimensional signal is subjected to image graying to be converted into an image form and stored in a database.
Further, the two-dimensional signal in step 201 is described by the formula:
in the formula, CWT f (a, b) is a two-dimensional signal, f (t) is a one-dimensional vibration signal,
being the complex conjugate of the wavelet function ψ (t), a and b represent the stretch and translation factors, respectively;
in the step 202, the two-dimensional signal is subjected to image graying, and the description formula is as follows:
wherein, image (CWT) f (a, b)) is image data obtained by converting a two-dimensional signal into an image gradation value, max (·) is a maximum function, and min (·) is a minimum function.
Further, the step 3 comprises the following sub-steps:
step 301: randomly selecting l types of samples from all images as a small sample set for training to construct a learning task;
step 302: the sampling set and the query set in each learning task pass through the coding module to obtain high-dimensional coding expression, and after the coding values of the obtained sampling sets of the same class are averaged, the coding expression is spliced with the query set in a characteristic dimension to form a matching pair;
step 303: passing each matching pair through the matching module to obtain a matching score;
step 304: using the average variance quantization model to predict the error between the matching score and the actual matching score;
step 305: and optimizing the model parameters by using a back propagation algorithm in deep learning until the final training is finished.
Further, the learning task in step 301 describes the formula as:
in the formula, task i In order to perform the task of learning,
in order to be a set of samples,
the method comprises the following steps of (1) setting a query set, wherein m is the number of samples of a sampling set, n is the number of samples of the query set, and k and j are natural numbers;
the matching pair in step 302 is described by the formula:
in the formula, pair (l, k) is a matching Pair, cat (·) is a splicing function on a characteristic dimension,
for function mapping of coding modules, M i,l The average value of the coded values of the l class sampling samples in the ith task is obtained;
the matching score in step 303 is described by the formula:
r l,k ＝g φ (Pair(l,k))
in the formula, r l,k ∈[0,1]Matching score for kth query set and l class, g φ () is a function mapping of the matching module;
the error between the predicted matching score and the actual matching score of the model in step 304 is described by the following formula:
in the formula, loss is the error between the matching score predicted by the model and the actual matching score;
the model parameters are optimized in the step 305 by using a back propagation algorithm in deep learning, and the description formula is as follows:
further, the step 4 comprises the following sub-steps:
step 401: passing all samples of known classes in the database through a coding module, and storing the output high-dimensional coding expression in the database;
step 402: after passing through a preprocessing and coding module, the vibration signal of the target bearing and the high-dimensional code of the known class in the database are simultaneously used as the input of a matching module, so that a matching score of each class in the known class is obtained;
step 403: and taking the maximum value of the matching scores of each of all the known classes, wherein the corresponding class is the fault mode of the target bearing.
Further, in step 403, the maximum value of the matching scores of all the known classes is obtained, which is described by the following formula:
in the formula, class is a category corresponding to the maximum matching score, namely a fault mode of the target bearing.
Further, the convolutional neural network adopts a deep neural network with sparse connection and parameter sharing characteristics.
The invention also provides a system for the bearing fault mode diagnosis method facing the small sample data set, which comprises the following steps:
the preprocessing module is used for converting the one-dimensional vibration signal of the bearing into a two-dimensional signal through continuous wavelet transformation and carrying out image gray value quantization on the two-dimensional signal;
the task generation module is used for randomly sampling from the preprocessed image data so as to construct a learning task of a plurality of small sample sets, wherein each small sample set comprises a sampling set and a query set;
the coding module is used for carrying out function mapping on the samples of the sampling set and the query set to obtain coding expression in a higher-dimensional space;
the matching module is used for matching the codes of the query set samples with the codes of all classes in the sampling set so as to obtain the corresponding classes of the query set samples;
and the diagnosis module is used for acquiring a vibration signal of the target bearing and diagnosing the bearing fault mode according to the vibration signal of the target bearing and the bearing fault diagnosis model.
Further, the basic architectures of the coding module and the matching module are convolutional neural networks, and the convolutional neural networks adopt deep neural networks with sparse connections and parameter sharing characteristics.
Compared with the prior art, the invention has the following advantages:
(1) There is still a higher diagnostic accuracy in the small sample set: under the condition that the number of samples is insufficient, a plurality of learning tasks are generated through random sampling, so that the model can learn migratable deep knowledge, the new task can be quickly adapted, and more accurate fault diagnosis precision can be obtained.
(2) The problem of performance reduction caused by the inconsistency of sample distribution of the training set and the test set can be solved: the diagnosis method of the invention does not directly map the sample characteristics to the corresponding failure modes, but indirectly diagnoses the failure modes by matching the scores, thereby avoiding the problems caused by inconsistent distribution to a certain extent.
(3) The coding module and the matching module in the invention both use the convolutional neural network as a basic framework, have two characteristics of sparse connection and parameter sharing, and are extremely suitable for deep feature expression mining of image data.
(4) The preprocessing module in the invention can process non-stationary and non-linear signals by a continuous wavelet transform method to obtain more robust feature expression.
(5) The task generation module in the invention can construct a plurality of learning tasks of small sample sets through image data, so that the model learns transferable knowledge among different tasks, and the model is assisted to adapt to a new task quickly.
Drawings
FIG. 1 is a partial bearing signal pre-processing image of the present invention.
FIG. 2 is a schematic diagram of a bearing failure mode diagnostic framework of the present invention.
FIG. 3 is a schematic diagram of the model structure of the present invention.
FIG. 4 is a flow chart of the method of the present invention.
Fig. 5 is a schematic diagram of a fault diagnosis system of 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 some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, shall fall within the protection scope of the present invention.
Examples
The invention provides a bearing fault mode diagnosis method for a small sample data set, which comprises the following steps as shown in figure 4:
1) The bearing vibration time sequence signal that this example was collected totally four kinds of different operating mode conditions, includes 10 bearing failure modes under each operating mode, except normal, also has 9 fault types, including three kinds of different trouble positions: inner ring failure, ball failure, and outer ring failure, with each failure location having three different failure sizes.
2) After various bearing signals are subjected to continuous wavelet transformation and image graying, a bearing fault characteristic diagram is formed, and a part of the bearing fault characteristic diagram is shown in figure 1 and is finally stored in a database of a server.
The step 2) specifically comprises the following steps:
step 201: vibration data collected by the acceleration sensor are one-dimensional continuous time sequence signals, and the signals are preprocessed through a continuous wavelet transform algorithm to obtain two-dimensional signals;
step 202: the two-dimensional signal is subjected to image graying to be converted into an image form and stored in a database.
In step 201, the two-dimensional signal is described by the formula:
in the formula, CWT f (a, b) is a two-dimensional signal, f (t) is a one-dimensional vibration signal,
being the complex conjugate of the wavelet function ψ (t), a and b represent the stretch and translation factors, respectively;
in step 202, the two-dimensional signal is subjected to image graying, and the description formula is as follows:
wherein, image (CWT) f (a, b)) is image data obtained by subjecting a two-dimensional signal to image gradation, max (·) is a maximum function, and min (·) is a minimum function.
3) Randomly extracting 4 types of samples from 40 types of images to serve as a small training sample set, extracting 5 samples from each type to serve as a sampling set, and extracting 20 samples to serve as a query set to form a learning task;
the above steps are repeatedly executed 2000 times, and 2000 learning task training models are generated.
In the formula, task i In order to perform the task of learning,
is a set of samples to be taken,
the method comprises the following steps that (1) a query set is obtained, m is the number of samples of the sampling set, n is the number of samples of the query set, k and j are natural numbers, and each sample in the sampling set and the query set is composed of a pair of fault characteristics and a fault mode;
and the sampling sample and the query sample of each task pass through a coding module to obtain a high-dimensional coding expression. After sampling sample code values of the same class are averaged, splicing processing is carried out on the sampling sample and the query sample on the characteristic dimension to form matching pairs:
in the formula, pair (l, k) is a matching Pair, cat (-) is a splicing function on a characteristic dimension,
for function mapping of coding modules, M i,l The average value of the code values of the class i sample samples in the ith task,
for coding modulesA parameter;
each matching pair passes through a matching module to obtain a matching score:
r l,k ＝g φ (Pair(l,k))
in the formula, r l,k ∈[0,1]Matching score for kth query set and l class, g φ (. H) is a function mapping of the matching module, phi is a parameter of the matching module;
match score predicted using mean variance quantization model and actual match score error:
in the formula, loss is the error between the matching score predicted by the model and the actual matching score;
model parameters were optimized using the back propagation algorithm in deep learning:
4) And passing the samples of all known classes in the database through an encoding module, and storing the high-dimensional encoding expression in the database. After passing through the preprocessing and coding module, the vibration signal of the target bearing is used as an input of the matching module together with the codes of the known classes in the database, so as to obtain a matching score with each class, as shown in fig. 2 and 3.
The category corresponding to the maximum matching score is the failure mode of the target bearing:
in the formula, class is a category corresponding to the maximum matching score, namely a fault mode of the target bearing.
Next, a bearing failure mode diagnosis system for a small sample data set proposed according to an embodiment of the present invention is described with reference to the drawings.
FIG. 5 is a schematic structural diagram of a bearing fault mode diagnosis system for a small sample data set according to the invention.
As shown in fig. 5, the bearing failure mode diagnosis system 10 for a small sample data set includes: the system comprises a preprocessing module 100, a task generating module 200, an encoding module 300, a matching module 400 and a diagnosis module 500.
The preprocessing module 100 converts the one-dimensional vibration signal of the bearing into a two-dimensional signal through continuous wavelet transform, and performs image graying on the two-dimensional signal. The task generation module 200 randomly samples the preprocessed image data to construct a learning task of a plurality of small sample sets, wherein each small sample set comprises a sampling set and a query set. The encoding module 300 performs function mapping on the samples of the sampling set and the query set to obtain an encoding expression in a higher dimensional space. The matching module 400 matches the codes of the query set samples with the codes of the categories in the sample set, so as to obtain the corresponding categories of the query set samples. The diagnosing module 500 is configured to collect a vibration signal of a target bearing, and diagnose a fault mode of the bearing according to the vibration signal of the bearing and the bearing fault diagnosis model. The system 10 of the embodiment of the present invention combines the deep learning and meta learning algorithms to improve the diagnostic accuracy in the small sample data set.
Further, in an embodiment of the present invention, the preprocessing module 100 can process non-stationary and non-linear signals by a continuous wavelet transform method, so as to obtain a more robust feature expression.
Further, in an embodiment of the present invention, the task generation module 200 can construct a plurality of learning tasks of small sample sets through the image data, so that the model learns migratable knowledge between different tasks, and the model is assisted to adapt to a new task quickly.
Further, in an embodiment of the present invention, the encoding module 300 and the matching module 400 both use a convolutional neural network as a basic architecture, have two characteristics of sparse connection and parameter sharing, and are very suitable for deep feature expression mining of image data.
While the invention has been described with reference to specific embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (7)
1. A bearing fault mode diagnosis method for small sample data sets is characterized by comprising the following steps:
step 1: collecting vibration signal data of the bearing under different equipment operating conditions through an acceleration sensor, and storing the vibration signal data into a server;
and 2, step: preprocessing signals in a server, converting original one-dimensional signals into two-dimensional signals through a continuous wavelet transform algorithm, and storing the two-dimensional signals in a database in the form of images;
and 3, step 3: constructing a bearing fault diagnosis model framework based on a convolutional neural network, wherein the bearing fault diagnosis model framework comprises a coding module and a matching module which both take the convolutional neural network as a basic framework, randomly sampling from image data in a database, and constructing a plurality of learning tasks of small sample data sets so as to train a bearing fault diagnosis model;
and 4, step 4: collecting vibration signals of a target bearing, diagnosing the vibration signals according to the preprocessing method in the step 2 and the bearing fault diagnosis model frame in the step 3, and obtaining a bearing fault mode;
the step 3 comprises the following sub-steps:
step 301: randomly selecting l types of samples from all images as a small sample set for training to construct a learning task;
step 302: the sampling set and the query set in each learning task pass through the coding module to obtain high-dimensional coding expression, and after the coding values of the obtained sampling sets of the same class are averaged, the coding expression is spliced with the query set in a characteristic dimension to form a matching pair;
step 303: passing each matching pair through the matching module to obtain a matching score;
step 304: using the average variance quantization model to predict the error between the matching score and the actual matching score;
step 305: optimizing model parameters by using a back propagation algorithm in deep learning until the final training is finished;
the learning task in step 301 is described as follows:
in the formula, task i In order to perform the task of learning,
is a set of samples to be taken,
the method comprises the following steps of (1) setting a query set, wherein m is the number of samples of a sampling set, n is the number of samples of the query set, and k and j are natural numbers;
the matching pair in step 302 is described by the following formula:
in the formula, pair (l, k) is a matching Pair, cat (·) is a splicing function on a characteristic dimension,
for functional mapping of coding modules, M i,l The average value of the coding values of the l-th type sampling samples in the ith task is obtained;
the matching score in step 303 is described by the formula:
r l,k ＝g φ (Pair(l,k))
in the formula, r l,k ∈[0,1]For the matching score of the kth set of queries to the l-th class, g φ () is a function mapping of the matching module;
the error between the predicted matching score and the actual matching score of the model in step 304 is described by the following formula:
in the formula, loss is the error between the matching score predicted by the model and the actual matching score;
in said 305, the model parameters are optimized by using a back propagation algorithm in deep learning, and the description formula is as follows:
the step 4 comprises the following sub-steps:
step 401: passing all samples of known classes in the database through a coding module, and storing the output high-dimensional coding expression in the database;
step 402: after passing through a preprocessing and coding module, a vibration signal of a target bearing and a high-dimensional code of a known category in a database are simultaneously used as input of a matching module, so that a matching score of each category in the known category is obtained;
step 403: and taking the maximum value of the matching scores of each of all the known classes, wherein the corresponding class is the fault mode of the target bearing.
2. A small sample dataset oriented bearing fault mode diagnostic method according to claim 1, wherein said step 2 comprises the sub-steps of:
step 201: vibration data collected by the acceleration sensor are one-dimensional continuous time sequence signals, and the signals are preprocessed through a continuous wavelet transform algorithm to obtain two-dimensional signals;
step 202: the two-dimensional signal is subjected to image graying to be converted into an image form and stored in a database.
3. A method for diagnosing a failure mode of a bearing based on a small sample data set as claimed in claim 2, wherein the two-dimensional signal in step 201 is described by the formula:
in the formula, CWT f (a, b) are two-dimensional signals, f (t) is a one-dimensional vibration signal,
being the complex conjugate of the wavelet function ψ (t), a and b represent the stretch and translation factors, respectively;
in the step 202, the two-dimensional signal is subjected to image graying, and the description formula is as follows:
wherein, image (CWT) f (a, b)) is image data obtained by subjecting a two-dimensional signal to image gradation, max (·) is a maximum function, and min (·) is a minimum function.
4. A method for diagnosing a failure mode of a bearing facing a small sample data set as claimed in claim 1, wherein the step 403 takes the maximum value of the matching scores of each of all and known classes, and the description formula is as follows:
in the formula, class is a category corresponding to the maximum matching score, namely a fault mode of the target bearing.
5. The method as claimed in claim 1, wherein the convolutional neural network is a deep neural network with sparse connection and parameter sharing characteristics.
6. A system for a small sample dataset oriented bearing failure mode diagnostic method as claimed in any one of claims 1 to 5, the system comprising:
the preprocessing module is used for converting the one-dimensional vibration signal of the bearing into a two-dimensional signal through continuous wavelet transformation and carrying out image gray-scale quantization on the two-dimensional signal;
the task generation module is used for randomly sampling from the preprocessed image data so as to construct a learning task of a plurality of small sample sets, wherein each small sample set comprises a sampling set and a query set;
the coding module is used for carrying out function mapping on the samples of the sampling set and the query set to obtain a coding expression in a higher dimensional space;
the matching module is used for matching the codes of the query set samples with the codes of all classes in the sampling set so as to obtain the corresponding classes of the query set samples;
and the diagnosis module is used for acquiring a vibration signal of the target bearing and diagnosing the bearing fault mode according to the vibration signal of the target bearing and the bearing fault diagnosis model.
7. The system for the small sample data set-oriented bearing fault mode diagnosis method according to claim 6, characterized in that the infrastructure of the coding module and the matching module are both convolutional neural networks, and the convolutional neural networks adopt deep neural networks with sparse connections and parameter sharing characteristics.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011091094.XA CN112417954B (en) | 2020-10-13 | 2020-10-13 | Bearing fault mode diagnosis method and system for small sample data set |
| PCT/CN2021/093451 WO2022077901A1 (en) | 2020-10-13 | 2021-05-12 | Bearing failure mode diagnosis method using small sample data sets, and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011091094.XA CN112417954B (en) | 2020-10-13 | 2020-10-13 | Bearing fault mode diagnosis method and system for small sample data set |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112417954A CN112417954A (en) | 2021-02-26 |
| CN112417954B true CN112417954B (en) | 2022-12-16 |
Family
ID=74854429
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011091094.XA Active CN112417954B (en) | 2020-10-13 | 2020-10-13 | Bearing fault mode diagnosis method and system for small sample data set |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112417954B (en) |
| WO (1) | WO2022077901A1 (en) |
Families Citing this family (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112417954B (en) * | 2020-10-13 | 2022-12-16 | 同济大学 | Bearing fault mode diagnosis method and system for small sample data set |
| CN113128561A (en) * | 2021-03-22 | 2021-07-16 | 南京航空航天大学 | Machine tool bearing fault diagnosis method |
| CN113758709A (en) * | 2021-09-30 | 2021-12-07 | 河南科技大学 | Rolling bearing fault diagnosis method and system combining edge calculation and deep learning |
| CN114818811B (en) * | 2022-05-02 | 2024-08-13 | 大连理工大学 | Aeroengine rolling bearing fault diagnosis method based on twin network metric learning |
| CN114925725B (en) * | 2022-05-09 | 2024-08-06 | 频率探索智能科技江苏有限公司 | Rolling bearing fault diagnosis method and system based on multi-scale LBP-CNN network |
| CN114993677B (en) * | 2022-05-11 | 2023-05-02 | 山东大学 | Rolling bearing fault diagnosis method and system for unbalanced small sample data |
| CN114818826B (en) * | 2022-05-19 | 2024-08-23 | 石家庄铁道大学 | Fault diagnosis method based on lightweight Vision Transformer module |
| CN114896733B (en) * | 2022-05-20 | 2024-02-13 | 合肥工业大学 | Rolling bearing fault type identification method based on deep reinforcement learning |
| CN115424053B (en) * | 2022-07-25 | 2023-05-02 | 北京邮电大学 | Small sample image recognition method, device, equipment and storage medium |
| CN115753103A (en) * | 2022-08-30 | 2023-03-07 | 山东建筑大学 | Fault diagnosis method and system based on standard self-learning data enhancement |
| CN115952408B (en) * | 2023-01-05 | 2024-08-20 | 东北大学 | Stamping production line bearing fault diagnosis method for multi-channel cross-domain few samples |
| CN116127279B (en) * | 2023-02-24 | 2024-04-19 | 上海师范大学 | Cavitation diagnosis method and device for small sample hydraulic pump |
| CN116701912B (en) * | 2023-06-14 | 2023-11-14 | 盐城工学院 | Bearing fault diagnosis method and system based on one-dimensional convolutional neural network |
| CN116558828B (en) * | 2023-07-10 | 2023-09-15 | 昆明理工大学 | Rolling bearing health state assessment method based on autocorrelation coefficient sparsity characteristic |
| CN116728291B (en) * | 2023-08-16 | 2023-10-31 | 湖南大学 | Robot polishing system state monitoring method and device based on edge calculation |
| CN116821697B (en) * | 2023-08-30 | 2024-05-28 | 聊城莱柯智能机器人有限公司 | Mechanical equipment fault diagnosis method based on small sample learning |
| CN116894190B (en) * | 2023-09-11 | 2023-11-28 | 江西南昌济生制药有限责任公司 | Bearing fault diagnosis method, device, electronic equipment and storage medium |
| CN116994076B (en) * | 2023-09-28 | 2024-01-19 | 中国海洋大学 | Small sample image recognition method based on double-branch mutual learning feature generation |
| CN117407796B (en) * | 2023-12-15 | 2024-03-01 | 合肥工业大学 | Cross-component small sample fault diagnosis method, system and storage medium |
| CN117892194A (en) * | 2024-01-10 | 2024-04-16 | 兰州理工大学 | Small sample fault diagnosis method and system based on Meta-GraphSage network |
| CN118013289B (en) * | 2024-04-09 | 2024-07-16 | 北京理工大学 | Variable working condition small sample fault diagnosis method, device, medium and product based on information fusion element transfer learning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101126929A (en) * | 2007-09-05 | 2008-02-20 | 东北大学 | Continuous miner remote real-time failure forecast and diagnosis method and device |
| CN109633369A (en) * | 2018-12-08 | 2019-04-16 | 国网山东省电力公司德州供电公司 | One kind being based on the matched electric network failure diagnosis method of multidimensional data similitude |
| CN109682596A (en) * | 2018-12-20 | 2019-04-26 | 南京航空航天大学 | High-speed overload input stage method for diagnosing faults under lack of balance sample |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018128927A1 (en) * | 2017-01-05 | 2018-07-12 | The Trustees Of Princeton University | Hierarchical health decision support system and method |
| US10831997B2 (en) * | 2018-08-09 | 2020-11-10 | CloudMinds Technology, Inc. | Intent classification method and system |
| CN109271936B (en) * | 2018-09-18 | 2021-09-24 | 哈尔滨工程大学 | Airplane vibration fault database construction and retrieval method based on perceptual hash algorithm |
| CN110516305B (en) * | 2019-07-26 | 2021-02-12 | 西安交通大学 | Intelligent fault diagnosis method under small sample based on attention mechanism meta-learning model |
| CN110823574B (en) * | 2019-09-30 | 2021-04-06 | 安徽富煌科技股份有限公司 | Fault diagnosis method based on semi-supervised learning deep countermeasure network |
| CN111597948A (en) * | 2020-05-11 | 2020-08-28 | 苏州求臻智能科技有限公司 | Fault detection and classification method for bearing vibration signals |
| CN111721535B (en) * | 2020-06-23 | 2021-11-30 | 中国人民解放军战略支援部队航天工程大学 | Bearing fault detection method based on convolution multi-head self-attention mechanism |
| CN112417954B (en) * | 2020-10-13 | 2022-12-16 | 同济大学 | Bearing fault mode diagnosis method and system for small sample data set |
-
2020
- 2020-10-13 CN CN202011091094.XA patent/CN112417954B/en active Active
-
2021
- 2021-05-12 WO PCT/CN2021/093451 patent/WO2022077901A1/en active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101126929A (en) * | 2007-09-05 | 2008-02-20 | 东北大学 | Continuous miner remote real-time failure forecast and diagnosis method and device |
| CN109633369A (en) * | 2018-12-08 | 2019-04-16 | 国网山东省电力公司德州供电公司 | One kind being based on the matched electric network failure diagnosis method of multidimensional data similitude |
| CN109682596A (en) * | 2018-12-20 | 2019-04-26 | 南京航空航天大学 | High-speed overload input stage method for diagnosing faults under lack of balance sample |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112417954A (en) | 2021-02-26 |
| WO2022077901A1 (en) | 2022-04-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112417954B (en) | Bearing fault mode diagnosis method and system for small sample data set | |
| CN109635677B (en) | Compound fault diagnosis method and device based on multi-label classification convolutional neural network | |
| CN113203566B (en) | Motor bearing fault diagnosis method based on one-dimensional data enhancement and CNN | |
| CN111460728B (en) | Method and device for predicting residual life of industrial equipment, storage medium and equipment | |
| CN111523081B (en) | Aeroengine fault diagnosis method based on enhanced gate control circulating neural network | |
| CN115290326A (en) | Rolling bearing fault intelligent diagnosis method | |
| CN114861740B (en) | Self-adaptive mechanical fault diagnosis method and system based on multi-head attention mechanism | |
| CN113128338A (en) | Intelligent diagnosis method for printing machine roller fault under small sample | |
| Zhang et al. | Intelligent machine fault diagnosis using convolutional neural networks and transfer learning | |
| CN113361559A (en) | Multi-mode data knowledge information extraction method based on deep width joint neural network | |
| CN114818826B (en) | Fault diagnosis method based on lightweight Vision Transformer module | |
| CN114595730A (en) | Rolling bearing fault diagnosis method based on GAF-DRSN | |
| CN112766200A (en) | Planetary gear box small sample fault diagnosis method based on one-dimensional convolutional neural network | |
| CN118094342A (en) | Self-adaptive fault diagnosis method and system for unsupervised domain of rotary machine | |
| Zheng et al. | An unsupervised transfer learning method based on SOCNN and FBNN and its application on bearing fault diagnosis | |
| CN117556317A (en) | Rotary bearing fault diagnosis method based on time-frequency image | |
| CN113758709A (en) | Rolling bearing fault diagnosis method and system combining edge calculation and deep learning | |
| CN113469013A (en) | Motor fault prediction method and system based on transfer learning and time sequence | |
| CN111382792B (en) | Rolling bearing fault diagnosis method based on double-sparse dictionary sparse representation | |
| CN116796224A (en) | Gear box fault diagnosis method based on time-frequency domain image and convolutional neural network | |
| CN117171907A (en) | Rolling bearing residual life prediction method and system | |
| CN114235409B (en) | Rolling bearing multi-user cooperative intelligent fault diagnosis method for light weight communication | |
| CN115640531A (en) | Fault diagnosis method based on residual error learning and attention mechanism fusion | |
| CN116007937A (en) | Intelligent fault diagnosis method and device for mechanical equipment transmission part | |
| CN113723592A (en) | Fault diagnosis method based on wind power gear box monitoring system |
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 |