[go: up one dir, main page]

CN120086803B - Rolling bearing fault diagnosis method, system, device, medium and program product - Google Patents

Rolling bearing fault diagnosis method, system, device, medium and program product

Info

Publication number
CN120086803B
CN120086803B CN202510549457.6A CN202510549457A CN120086803B CN 120086803 B CN120086803 B CN 120086803B CN 202510549457 A CN202510549457 A CN 202510549457A CN 120086803 B CN120086803 B CN 120086803B
Authority
CN
China
Prior art keywords
domain
features
target domain
source domain
dual
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
Application number
CN202510549457.6A
Other languages
Chinese (zh)
Other versions
CN120086803A (en
Inventor
马赛
朱鹏
徐晟博
王政
朱洪涛
邹斌
姚鹏
刘玥
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong University
Original Assignee
Shandong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shandong University filed Critical Shandong University
Priority to CN202510549457.6A priority Critical patent/CN120086803B/en
Publication of CN120086803A publication Critical patent/CN120086803A/en
Application granted granted Critical
Publication of CN120086803B publication Critical patent/CN120086803B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M13/00—Testing of machine parts
    • G01M13/04—Bearings
    • G01M13/045—Acoustic or vibration analysis
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0475—Generative networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06N3/094—Adversarial learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06N3/096—Transfer learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Acoustics & Sound (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

The invention discloses a rolling bearing fault diagnosis method, a system, equipment, a medium and a program product, which relate to the technical field of rotary machinery fault diagnosis and comprise the steps of training a dual-mode anti-deep migration learning network based on source domain features and target domain features, training a dual-classifier based on the source domain features and the target domain features to obtain source domain classification errors and dual-classifier deterministic errors, training a domain discriminator based on multi-linear mapping formed by prediction results of the source domain features, the target domain features and the dual-classifier to obtain domain classification errors, combining the maximum mean difference errors of the source domain features and the target domain features as a loss function, and classifying target domain samples by adopting the dual-mode anti-deep migration learning network to obtain fault diagnosis results. The dual-mode anti-deep migration learning network is designed, category information is reserved by utilizing multi-linear mapping and dual-classifier deterministic difference, and maximum mean value difference is introduced, so that fault diagnosis performance is improved.

Description

Rolling bearing fault diagnosis method, system, device, medium and program product
Technical Field
The present invention relates to the field of fault diagnosis technology for rotating machinery, and in particular, to a method, a system, an apparatus, a medium, and a program product for diagnosing faults of a rolling bearing.
Background
In the fault diagnosis of the rolling bearing, a model trained by a large number of marked bearing vibration data under a certain working condition can be migrated to the working condition of load change or rotation speed change through anti-migration learning, so that the bearing fault under the new condition can be accurately identified.
However, the conventional migration countermeasure method has at least the following problems:
(1) The traditional domain countermeasure method mostly adopts edge distribution, aligns source domain and target domain samples, and ignores the influence of categories.
(2) The conventional domain countermeasure method does not directly align the source domain and the target domain features, so that the similarity of the source domain and the target domain features is reduced.
Disclosure of Invention
In order to solve the problems, the invention provides a rolling bearing fault diagnosis method, a system, equipment, a medium and a program product, wherein a dual-mode anti-deep migration learning network is designed, a constructed sparse wavelet convolution module is utilized to extract characteristics containing specific fault frequencies, multi-linear mapping and dual-classifier deterministic difference are utilized to retain category information, and the maximum mean difference is introduced to better distribute the characteristics of Ji Yuanyu and a target domain, so that fault diagnosis performance is improved.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
In a first aspect, the present invention provides a rolling bearing fault diagnosis method, including:
taking an obtained vibration signal to be measured of the rolling bearing as a target domain sample, taking a historical vibration signal as a source domain sample, and extracting features of the target domain sample and the source domain sample to obtain source domain features and target domain features;
training a built dual-mode countermeasure deep migration learning network based on source domain features and target domain features, training a dual classifier based on the source domain features and the target domain features to obtain source domain classification errors and dual classifier deterministic errors, training a domain arbiter based on multi-linear mapping formed by prediction results of the source domain features, the target domain features and the dual classifier to obtain domain classification errors, and combining the maximum mean difference errors of the source domain features and the target domain features to serve as a loss function in the training process;
and classifying the target domain samples by adopting the trained dual-mode anti-deep migration learning network to obtain a fault diagnosis result.
The method comprises the steps of selecting a target domain sample and a source domain sample, respectively extracting features by adopting a sparse wavelet convolution module, specifically comprising the steps of introducing a spectral kurtosis constraint term into the wavelet convolution layer and introducing the sparse constraint term into the sparse constraint module, initializing the wavelet convolution layer by adopting a wavelet convolution kernel with translation parameters and scale parameters, respectively extracting features by adopting the initialized wavelet convolution layer and the initialized wavelet convolution kernel to the target domain sample and the initialized wavelet convolution kernel, and specifically comprising the following steps of:
;;
wherein u and s are translation parameters and scale parameters, respectively, t is the index of the convolution kernel, For the selected wavelet basis function,For the input samples, h is the output of the wavelet convolution layer,Is convolution operation;
In the sparse constraint module, a random vector is transformed into half of the original length through a full-connection layer, then the random vector is converted by a ReLU activation function, then the original length is restored through another full-connection layer, finally the random vector is normalized to be in the range of (0, 1) through a sigmoid activation function, a final sparse vector V is obtained, and h is multiplied by the sparse vector V, so that the output characteristic is obtained.
As an alternative embodiment, the spectral kurtosis constraint termThe method comprises the following steps:
;;
Sparse constraint term The method comprises the following steps:;
wherein I represents The number of the features in (a) is,Representation ofThe ith feature of (2), L representsIs provided for the length of (a),Representation ofIs a first element of the (c) a (c),An envelope curve representing h; Is a kernel hilbert space; Is the L1 norm.
As an alternative embodiment, the error is classified based on the source domainDeterministic error for dual classifierSum domain classification errorThe loss function of the training process is obtained as follows:
;
Wherein, the A first classifier, a second classifier and a domain discriminator respectivelyNetwork parameters of (a); Weights of the deterministic error and the domain classification error of the dual classifier respectively; The number of the source domain samples is; The number of the target domain samples is; Is cross entropy loss; the prediction result of the ith source domain sample under the jth classifier is obtained; An actual tag that is the ith source domain sample; operators for calculating deterministic errors of the dual classifiers; Is a multi-linear mapping; The method comprises the steps of predicting a result of a first classifier on an ith target domain sample; the prediction result of the ith target domain sample for the second classifier; to satisfy source domain distribution Mathematical expectations calculated for source domain samples of (2); To meet the target domain distribution Mathematical expectations calculated for the target domain samples of (2); respectively an ith source domain sample and a jth target domain sample; The combined variable is the characteristic g of the ith source domain sample and the predicted result p; Is the joint variable of the feature g of the jth target domain sample and the predicted result p.
As an alternative embodiment, the inputs to the domain arbiter are:
;
The method comprises the steps of extracting source domain features or target domain features from a target domain, wherein g is an extracted source domain feature or target domain feature, p is a double classifier prediction result, and taking an average value after two classifiers are respectively predicted; And Are all multi-linear mappings; Is tensor product, while the term as add is element product, Is the dimension of g and p.
As an alternative embodiment, the maximum mean difference error is used to align the extracted features;
;
Wherein, the Network parameters that are feature extractors; deterministic errors for the dual classifiers, respectively Domain classification errorsAnd maximum mean difference errorWeights of (2); The number of the source domain samples is; The number of the target domain samples is; operators for calculating deterministic errors of the dual classifiers; The method comprises the steps of predicting a result of a first classifier on an ith target domain sample; the prediction result of the ith target domain sample for the second classifier; to satisfy source domain distribution Mathematical expectations calculated for source domain samples of (2); To meet the target domain distribution Mathematical expectations calculated for the target domain samples of (2); respectively an ith source domain sample and a jth target domain sample; The combined variable is the characteristic g of the ith source domain sample and the predicted result p; the joint variable of the feature g of the jth target domain sample and the prediction result p; Is a multi-linear mapping; is a domain arbiter, H is a regenerated kernel Hilbert space, Is a feature mapping function mapped to the regenerated kernel hilbert space; features extracted for the ith source domain sample; features extracted for the ith target domain sample.
In a second aspect, the present invention provides a rolling bearing failure diagnosis system including:
The characteristic extraction module is configured to take an acquired vibration signal to be detected of the rolling bearing as a target domain sample, take a historical vibration signal as a source domain sample, and extract characteristics of the target domain sample and the source domain sample to obtain source domain characteristics and target domain characteristics;
The training module is configured to train the built dual-mode anti-deep migration learning network based on the source domain features and the target domain features, train the dual-classifier based on the source domain features and the target domain features to obtain a source domain classification error and a dual-classifier deterministic error, train the domain discriminator based on the multi-linear mapping formed by the prediction results of the source domain features, the target domain features and the dual-classifier to obtain a domain classification error, and then combine the maximum mean difference error of the source domain features and the target domain features to serve as a loss function in the training process;
the classification module is configured to classify the target domain samples by adopting the trained dual-mode anti-deep migration learning network to obtain fault diagnosis results.
In a third aspect, the invention provides an electronic device comprising a memory and a processor and computer instructions stored on the memory and running on the processor, which when executed by the processor, perform the method of the first aspect.
In a fourth aspect, the present invention provides a computer readable storage medium storing computer instructions which, when executed by a processor, perform the method of the first aspect.
In a fifth aspect, the invention provides a computer program product comprising a computer program which, when executed by a processor, implements the method of the first aspect.
Compared with the prior art, the invention has the beneficial effects that:
The invention provides a rolling bearing fault diagnosis method based on a dual-mode countermeasure deep migration learning network, which utilizes a constructed sparse wavelet convolution module to extract characteristics containing specific fault frequency, utilizes multi-linear mapping and dual-classifier deterministic difference to reserve category information, introduces maximum mean value difference to better distribute the characteristics of Ji Yuanyu and a target domain, and has better fault diagnosis performance and knowledge migration capability.
The sparse wavelet convolution module constructed by the invention adopts the wavelet convolution kernel with trainable translation and scale parameters to carry out convolution initialization, and limits the extracted features by utilizing spectral kurtosis constraint and sparse constraint so as to effectively extract the features containing fault features, improve the sensitivity of the model to transient impact extraction, allow the most suitable wavelet kernel to be selected for fault diagnosis, improve the fault diagnosis performance and enhance the interpretability.
The invention designs a dual-mode countermeasure deep migration learning network, introduces the predicted variables of the dual classifier as condition information into countermeasure learning, can better distribute the features of Ji Yuanyu and a target domain, simultaneously retains category information, takes the multi-linear mapping of the features and category prediction as the input of a discriminator, and can capture the complex interaction relationship of the features and the category prediction.
The method and the device generate more discriminant feature representation by maximizing the deterministic difference of the two classifiers, enhance classification and generalization capability, effectively cope with the uncertainty of the target domain, and simultaneously introduce the maximum mean value difference to better help the feature extractor extract the domain invariant feature.
Additional aspects of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for diagnosing a rolling bearing failure according to embodiment 1 of the present invention;
fig. 2 is a schematic diagram of a fault diagnosis method for a rolling bearing according to embodiment 1 of the present invention.
Detailed Description
The invention is further described below with reference to the drawings and examples.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the invention. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms also are intended to include the plural forms, and furthermore, it is to be understood that the terms "comprises" and "comprising" and any variations thereof are intended to cover a non-exclusive inclusion, e.g., a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Embodiments of the invention and features of the embodiments may be combined with each other without conflict.
Example 1
As shown in fig. 1-2, the present embodiment provides a rolling bearing fault diagnosis method, which specifically includes the following steps:
taking an obtained vibration signal to be measured of the rolling bearing as a target domain sample, taking a historical vibration signal as a source domain sample, and extracting features of the target domain sample and the source domain sample to obtain source domain features and target domain features;
training a built dual-mode countermeasure deep migration learning network based on source domain features and target domain features, training a dual classifier based on the source domain features and the target domain features to obtain source domain classification errors and dual classifier deterministic errors, training a domain arbiter based on multi-linear mapping formed by prediction results of the source domain features, the target domain features and the dual classifier to obtain domain classification errors, and combining the maximum mean difference errors of the source domain features and the target domain features to serve as a loss function in the training process;
and classifying the target domain samples by adopting the trained dual-mode anti-deep migration learning network to obtain a fault diagnosis result.
In this embodiment, an obtained rolling bearing vibration signal of an unknown fault type is taken as a target domain sample, a historical vibration signal of the rolling bearing in each fault state and a fault type label corresponding to the historical vibration signal are taken as a source domain sample, the target domain sample and the source domain sample are input into a dual-mode anti-deep migration learning network, and fault diagnosis of the rolling bearing is realized by training the network.
The sensor can be used for collecting vibration signals of the rolling bearing, and vibration signals of the rolling bearing uploaded by the collecting terminal can also be obtained.
In the embodiment, a sparse wavelet convolution module is utilized to conduct feature extraction on a target domain sample and a source domain sample to obtain a source domain feature and a target domain feature, the sparse wavelet convolution module comprises two parts, including a wavelet convolution layer containing a wavelet convolution kernel and a sparse constraint module composed of two full-connection layers and two activation functions, a spectral kurtosis constraint term is introduced into the wavelet convolution layer, and a sparse constraint term is introduced into the sparse constraint module. The wavelet convolution layer is initialized by utilizing wavelet convolution check with trainable translation parameters and scale parameters, extracted features are limited by utilizing spectral kurtosis constraint and sparse constraint, and therefore, the extracted source domain features and the extracted target domain features are obtained after target domain samples and source domain samples are input into a sparse wavelet convolution module. The design of the sparse wavelet convolution module improves the sensitivity of the network to transient impact extraction, allows the most suitable wavelet kernel to be selected for fault diagnosis, improves the model performance and enhances the interpretability.
Specific:
The kernels of the wavelet convolution layers in the sparse wavelet convolution module are initialized by wavelet convolution kernels with trainable panning and scale parameters. The initialization formula of the wavelet convolution kernel is:
;
Wherein u and s represent a translation parameter and a scale parameter, respectively, t represents an index of a specific value of the convolution kernel, Representing the selected wavelet basis functions.
The operation of the wavelet convolution layer is expressed as:
;
Wherein, the For the time domain samples of the input source and target domains, h is the output of the wavelet convolution layer,Is a convolution operation.
In the embodiment, three different wavelet convolution kernels of Laplace wavelet, mexh wavelet and Morlet wavelet are adopted, and the scale parameters on each channel of the convolution kernels are uniformly distributed in the ranges of (0.1,2), (0.1,3) and (0.1, 4.5) respectively.
In the embodiment, the spectral kurtosis is used as a typical sparsity measure for better selecting the frequency band of fault feature extraction, and the spectral kurtosis constraint termExpressed as:
;;
Wherein I represents The number of the features in (a) is,Representation ofThe ith feature of (2), L representsIs provided for the length of (a),Representation ofIs a first element of the (c) a (c),An envelope curve representing h; Is the kernel hilbert space.
While initializing with wavelet convolution kernels may yield more interpretable and representative features, diversification of translation parameters, scale parameters, and wavelet basis function choices may prevent improvement in fault diagnosis accuracy. A sparse constraint module is therefore introduced to screen the extracted features. Firstly, converting a random vector into half of the original length through a full connection layer, then converting the random vector into a ReLU activation function, then recovering the original length through another full connection layer, finally normalizing the values into a (0, 1) range through a sigmoid activation function to obtain a final sparse vector V, and multiplying h by the sparse vector V to obtain an output feature vector。
To remove channels with little influence on the result, reserving relatively critical channels with fault characteristics, introducing L1 norm sparse constraint terms:;Is the L1 norm.
In the embodiment, the sparse wavelet convolution module enriches the physical meaning of the model by using the sparse wavelet convolution kernel and the spectral kurtosis constraint term, improves the interpretability of the model, and simultaneously effectively extracts the characteristics containing the fault characteristics, thereby improving the performance of fault diagnosis.
In this embodiment, training the built dual-mode anti-deep migration learning network based on the source domain features and the target domain features specifically includes:
firstly, the source domain features are used for carrying out basic training on the double classifiers, so that the first classifier and the second classifier can accurately classify the source domain samples, and the source domain prediction results are based on Determining source domain classification errorsThe method specifically comprises the following steps:
;
Wherein, the Network parameters of the feature extractor, the first classifier and the second classifier respectively; The number of the source domain samples is; Is cross entropy loss; the prediction result of the ith source domain sample under the jth classifier is obtained; is the actual label of the ith source domain sample.
Then, training the dual classifier by adopting the target domain features on the premise of ensuring the correct classification result of the source domain sample so as to learn the decision boundary of the discriminant on the target domain sample, thereby predicting the result based on the target domainDetermining a dual classifier deterministic errorAnd simultaneously training the domain arbiter to enhance the ability of the domain arbiter to distinguish whether the extracted features are derived from source domain samples or target domain samples.
In summary, the training process of the dual-mode countered deep migration learning network is expressed as:
;
;
;
Wherein, the Classifying errors for a domain; Weights of the deterministic error and the domain classification error of the dual classifier respectively; Is a domain arbiter; Network parameters for a domain arbiter; The number of the target domain samples is; operators for calculating deterministic errors of the dual classifiers; The method comprises the steps of predicting a result of a first classifier on an ith target domain sample; the prediction result of the ith target domain sample for the second classifier; respectively an ith source domain sample and a jth target domain sample; to satisfy source domain distribution Mathematical expectations calculated for source domain samples of (2); To meet the target domain distribution Mathematical expectations calculated for the target domain samples of (2); The combined variable is the characteristic g of the ith source domain sample and the predicted result p; Is the joint variable of the feature g of the jth target domain sample and the predicted result p.
For multi-linear mapping, the extracted source domain features and target domain features and the prediction result of the double classifier are taken as input, a domain discriminator D is adopted, and the domain classification prediction result of the multi-linear mapping is outputThe domain classification error refers to the error between the prediction result of a certain sample judged to belong to a source domain or a target domain by a domain discriminator and the real result of the sample actually belonging to the source domain or the target domain.
A is a square matrix of size K x K,Is the element in the m-th row and n-th column of A, which represents the probability product of the first classifier classifying the sample into the m-th class and the second classifier classifying the sample into the n-th class, A can effectively evaluate the prediction correlation of the dual classifier on different classes.
To minimize classifier differences in prediction relevance, two predictions are made consistent and relevant, i.e., maximizing the diagonal elements of a, which also enables predictions to determine prediction categories with high confidence. Meanwhile, the off-diagonal elements of a may be regarded as confusion information of two classifiers.
In addition, T is a multi-linear mapping for overcoming the problem that multi-modal information conveyed in classifier prediction cannot be fully utilized to match multi-modal distribution in complex fields. Multi-linear mappingThe multi-modal structure behind a complex data distribution can be fully captured and thus used for o-computation of joint variables, noted:
;
Wherein g is the extracted source domain feature or target domain feature; and p is the prediction result of the classifier, which is the tensor product.
At this timeIn order to avoid dimensional explosion, when the dimensions of g and p areSatisfy the following requirementsWhen the randomization method is applied to the calculation of the joint variable o, it is noted that:
;
Wherein, the "; And Is a random matrix that is sampled only once and fixed during the training process, each elementFollowing a symmetrical distribution with a single variance, i.e.,Such as gaussian and uniform distributions.
The joint variable o, the input of the domain arbiter (multi-linear mapping), is then written as:
;
Wherein, the AndAre multi-linear mappings.
In this embodiment, the feature extractor G is trained by the classification result of the classifier and the classification result of the domain arbiter, so that the extracted features can have the same effect on the two classifiers as much as possible, and can confuse the domain arbiter as much as possible.
In addition, the above process does not directly align the features extracted by the feature extractor, thus introducing the maximum mean difference between the source domain features and the target domain featuresFurther aligning the features extracted by the feature extractor G;
The process is as follows:
;
wherein H is the regenerated core Hilbert space, Is a feature mapping function mapped to the regenerated kernel hilbert space; Is the weight of the maximum mean difference; features extracted for the ith source domain sample; features extracted for the ith target domain sample.
Extracting domain invariant features for knowledge migration through a feature extractor G by performing antagonism training on a feature extractor G, a domain discriminator D and a dual classifier in a dual-mode antagonism deep migration learning network; and finally, training the network until convergence, and classifying the target domain sample by using the trained feature extractor and the trained double classifier to realize the fault diagnosis of the rolling bearing under the cross-working condition.
In this embodiment, experiments were performed on bearing failure diagnosis data sets, and vibration signals of the bearings were collected by accelerometers mounted at the ends of the bearings. The data set includes three types of failure samples-an inner ring failure, an outer ring failure, and a rolling element failure. The vibration signals are collected in a set from the bearing end and the fan end, respectively. Each type of sample had four fault diameters (0.007 inch, 0.014 inch, 0.021 inch and 0.028 inch) and was tested at four different loads (0, 1 horsepower, 2 horsepower and 3 horsepower) and corresponding speeds.
The proposed sparse wavelet convolution module is applied to fault feature extraction of vibration signals. Three different wavelet convolution kernels of Laplace wavelet, mexh wavelet and Morlet wavelet are adopted, and the scale parameters on each channel of the convolution kernels are uniformly distributed in the ranges of (0.1,2), (0.1,3) and (0.1, 4.5) respectively. For each super parameter in the dual-mode countermeasure deep migration learning network, the super parameters are determined by a grid search method,,。
Experimental results show that the dual-mode anti-deep migration learning network provided by the embodiment can well extract fault characteristics, the multi-linear mapping and the deterministic difference of the dual classifier are utilized to retain category information, and the maximum mean difference is introduced to help better distribute Ji Yuanyu target domain characteristics.
According to the characteristic visual qualitative analysis, the characteristics extracted by the samples with the same fault type are clustered, and obvious boundaries exist among the characteristics extracted by the samples with different fault types, so that the model has good fault diagnosis capability, and an effective scheme is provided for fault diagnosis of the rolling bearing.
Of course, in other embodiments, the selection of the wavelet convolution kernel and the setting of the gating factor value may be modified according to the specific situation and requirement, and are not limited to the above values.
Example 2
The present embodiment provides a rolling bearing failure diagnosis system including:
The characteristic extraction module is configured to take an acquired vibration signal to be detected of the rolling bearing as a target domain sample, take a historical vibration signal as a source domain sample, and extract characteristics of the target domain sample and the source domain sample to obtain source domain characteristics and target domain characteristics;
The training module is configured to train the built dual-mode anti-deep migration learning network based on the source domain features and the target domain features, train the dual-classifier based on the source domain features and the target domain features to obtain a source domain classification error and a dual-classifier deterministic error, train the domain discriminator based on the multi-linear mapping formed by the prediction results of the source domain features, the target domain features and the dual-classifier to obtain a domain classification error, and then combine the maximum mean difference error of the source domain features and the target domain features to serve as a loss function in the training process;
the classification module is configured to classify the target domain samples by adopting the trained dual-mode anti-deep migration learning network to obtain fault diagnosis results.
It should be noted that the above modules correspond to the steps described in embodiment 1, and the above modules are the same as examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in embodiment 1. It should be noted that the modules described above may be implemented as part of a system in a computer system, such as a set of computer-executable instructions.
In further embodiments, there is also provided:
An electronic device comprising a memory and a processor and computer instructions stored on the memory and running on the processor, which when executed by the processor, perform the method described in embodiment 1. For brevity, the description is omitted here.
It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general purpose processors, digital signal processors DSP, application specific integrated circuits ASIC, off-the-shelf programmable gate array FPGA or other programmable logic device, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory may include read only memory and random access memory and provide instructions and data to the processor, and a portion of the memory may also include non-volatile random access memory. For example, the memory may also store information of the device type.
A computer readable storage medium storing computer instructions which, when executed by a processor, perform the method described in embodiment 1.
The method in embodiment 1 may be directly embodied as a hardware processor executing or executed with a combination of hardware and software modules in the processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory, and the processor reads the information in the memory and, in combination with its hardware, performs the steps of the above method. To avoid repetition, a detailed description is not provided herein.
A computer program product comprising a computer program which, when executed by a processor, implements the method described in embodiment 1.
The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product comprises computer executable instructions, such as instructions comprised in program modules, being executed in a device on a real or virtual processor of a target to perform the processes/methods as described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform particular tasks or implement particular abstract data types. In various embodiments, the functionality of the program modules may be combined or split between program modules as desired. Machine-executable instructions for program modules may be executed within local or distributed devices. In distributed devices, program modules may be located in both local and remote memory storage media.
Computer program code for carrying out methods of the present invention may be written in one or more programming languages. These computer program code may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the computer or other programmable data processing apparatus, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the computer, partly on the computer, as a stand-alone software package, partly on the computer and partly on a remote computer or entirely on the remote computer or server.
In the context of the present invention, computer program code or related data may be carried by any suitable carrier to enable an apparatus, device or processor to perform the various processes and operations described above. Examples of carriers include signals, computer readable media, and the like. Examples of signals may include electrical, optical, radio, acoustical or other form of propagated signals, such as carrier waves, infrared signals, etc.
Those of ordinary skill in the art will appreciate that the elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
While the foregoing description of the embodiments of the present invention has been presented in conjunction with the drawings, it should be understood that it is not intended to limit the scope of the invention, but rather, it is intended to cover all modifications or variations within the scope of the invention as defined by the claims of the present invention.

Claims (9)

1.一种滚动轴承故障诊断方法，其特征在于，包括：1. A rolling bearing fault diagnosis method, characterized by comprising: 以获取的滚动轴承的待测振动信号为目标域样本，以历史振动信号为源域样本，对目标域样本和源域样本进行特征提取，得到源域特征和目标域特征；The vibration signal of the rolling bearing to be measured is used as the target domain sample, and the historical vibration signal is used as the source domain sample, and features are extracted from the target domain sample and the source domain sample to obtain source domain features and target domain features; 基于源域特征和目标域特征对构建的双模式对抗深度迁移学习网络进行训练；其中，基于源域特征和目标域特征对双分类器进行训练，得到源域分类误差和双分类器确定性误差，基于由源域特征、目标域特征和双分类器的预测结果构成的多线性映射对域判别器进行训练，得到域分类误差，再结合源域特征和目标域特征的最大均值差异误差，作为训练过程的损失函数；The constructed dual-mode adversarial deep transfer learning network is trained based on the source domain features and the target domain features; wherein, the dual classifier is trained based on the source domain features and the target domain features to obtain the source domain classification error and the dual classifier deterministic error, and the domain discriminator is trained based on the multilinear mapping composed of the source domain features, the target domain features and the prediction results of the dual classifier to obtain the domain classification error, and then the maximum mean difference error of the source domain features and the target domain features is combined as the loss function of the training process; 采用训练后的双模式对抗深度迁移学习网络对目标域样本进行分类，得到故障诊断结果；The trained dual-mode adversarial deep transfer learning network is used to classify the target domain samples and obtain the fault diagnosis results; 对目标域样本和源域样本，采用稀疏小波卷积模块，分别进行特征提取；具体包括：稀疏小波卷积模块包括小波卷积层和由两个全连接层和两个激活函数构成的稀疏约束模块，且在小波卷积层中引入谱峭度约束项，在稀疏约束模块中引入稀疏约束项；小波卷积层采用具有平移参数和尺度参数的小波卷积核进行初始化，以初始化后的小波卷积层和稀疏约束模块对目标域样本和源域样本分别进行特征提取；具体为：For target domain samples and source domain samples, sparse wavelet convolution modules are used to extract features respectively; specifically, the sparse wavelet convolution module includes a wavelet convolution layer and a sparse constraint module composed of two fully connected layers and two activation functions, and a spectral kurtosis constraint term is introduced in the wavelet convolution layer, and a sparse constraint term is introduced in the sparse constraint module; the wavelet convolution layer is initialized with a wavelet convolution kernel with a translation parameter and a scale parameter, and the initialized wavelet convolution layer and sparse constraint module are used to extract features for target domain samples and source domain samples respectively; specifically, ；； ; ; 其中，u和s分别为平移参数和尺度参数，t为卷积核的索引，为所选择的小波基函数，为输入的样本，h为小波卷积层的输出，为卷积运算；Among them, u and s are the translation parameter and scale parameter respectively, t is the index of the convolution kernel, is the selected wavelet basis function, is the input sample, h is the output of the wavelet convolution layer, is the convolution operation; 在稀疏约束模块中，通过全连接层将一个随机向量变换为原长度的一半，然后用ReLU激活函数转换，随后通过另一个全连接层恢复到原来的长度，最后用sigmoid激活函数归一化到（0，1）范围内，得到最终稀疏向量V，将h乘以稀疏向量V，得到输出的特征。In the sparse constraint module, a random vector is transformed to half of its original length through a fully connected layer, then converted using the ReLU activation function, and then restored to its original length through another fully connected layer. Finally, it is normalized to the range of (0, 1) using the sigmoid activation function to obtain the final sparse vector V. h is multiplied by the sparse vector V to obtain the output features. 2.如权利要求1所述的一种滚动轴承故障诊断方法，其特征在于，谱峭度约束项为：2. A rolling bearing fault diagnosis method as claimed in claim 1, characterized in that the spectral kurtosis constraint term for: ；； ; ; 稀疏约束项为：；Sparse Constraints for: ; 其中，I表示中特征的个数，表示中的第i个特征，L表示的长度，表示的第l个元素，表示h的包络线；为核希尔伯特空间；为L1范数。Where I represents The number of features in express The i- th feature in L represents Length, express The lth element of represents the envelope of h ; is the nuclear Hilbert space; is the L1 norm. 3.如权利要求1所述的一种滚动轴承故障诊断方法，其特征在于，基于源域分类误差、双分类器确定性误差和域分类误差得到训练过程的损失函数为：3. A rolling bearing fault diagnosis method as claimed in claim 1, characterized in that based on the source domain classification error , dual classifier deterministic error and domain classification error The loss function of the training process is obtained as: ； ; 其中，分别为第一分类器、第二分类器和域判别器的网络参数；分别为双分类器确定性误差和域分类误差的权重；为源域样本个数；为目标域样本个数；为交叉熵损失；为第i个源域样本在第j个分类器下的预测结果；为第i个源域样本的实际标签；为计算双分类器确定性误差的算子；为多线性映射；为第一分类器对第i个目标域样本的预测结果；为第二分类器对第i个目标域样本的预测结果；为满足源域分布的源域样本所计算的数学期望；为满足目标域分布的目标域样本所计算的数学期望；分别为第i个源域样本和第j个目标域样本；为第i个源域样本的特征g和预测结果p的联合变量；为第j个目标域样本的特征g和预测结果p的联合变量。in, They are the first classifier, the second classifier and the domain discriminator respectively. Network parameters; are the weights of the dual classifier deterministic error and domain classification error, respectively; is the number of source domain samples; is the number of samples in the target domain; is the cross entropy loss; is the prediction result of the i- th source domain sample under the j -th classifier; is the actual label of the i- th source domain sample; is the operator for calculating the deterministic error of the dual classifier; is a multilinear mapping; is the prediction result of the first classifier for the i- th target domain sample; is the prediction result of the second classifier for the i- th target domain sample; To meet the source domain distribution The mathematical expectation calculated from the source domain samples; To meet the target domain distribution The mathematical expectation calculated from the target domain samples; are the i- th source domain sample and the j -th target domain sample respectively; is the joint variable of the feature g of the i- th source domain sample and the predicted result p ; It is the joint variable of the feature g and the prediction result p of the j -th target domain sample. 4.如权利要求1所述的一种滚动轴承故障诊断方法，其特征在于，域判别器的输入为：4. A rolling bearing fault diagnosis method as claimed in claim 1, characterized in that the input of the domain discriminator is: ； ; 其中，g为提取的源域特征或目标域特征；p为双分类器预测结果，为两个分类器分别预测后取平均值得到；和均为多线性映射；为张量积；⊙是元素积，为g和p的维度。Where g is the extracted source domain feature or target domain feature; p is the prediction result of the dual classifier, which is the average value of the predictions of the two classifiers. and They are all multilinear mappings; is the tensor product; ⊙ is the element-wise product, are the dimensions of g and p . 5.如权利要求1所述的一种滚动轴承故障诊断方法，其特征在于，最大均值差异误差用于对所提取的特征进行对齐；5. A rolling bearing fault diagnosis method according to claim 1, characterized in that the maximum mean difference error is used to align the extracted features; ； ; 其中，为特征提取器的网络参数；分别为双分类器确定性误差、域分类误差和最大均值差异误差的权重；为源域样本个数；为目标域样本个数；为计算双分类器确定性误差的算子；为第一分类器对第i个目标域样本的预测结果；为第二分类器对第i个目标域样本的预测结果；为满足源域分布的源域样本所计算的数学期望；为满足目标域分布的目标域样本所计算的数学期望；分别为第i个源域样本和第j个目标域样本；为第i个源域样本的特征g和预测结果p的联合变量；为第j个目标域样本的特征g和预测结果p的联合变量；为多线性映射；为域判别器；H是再生核希尔伯特空间，是映射到再生核希尔伯特空间的特征映射函数；为第i个源域样本所提取的特征；为第i个目标域样本所提取的特征。in, is the network parameter of the feature extractor; are the deterministic errors of the dual classifiers , domain classification error and the maximum mean difference error The weight of is the number of source domain samples; is the number of samples in the target domain; is the operator for calculating the deterministic error of the dual classifier; is the prediction result of the first classifier for the i- th target domain sample; is the prediction result of the second classifier for the i- th target domain sample; To meet the source domain distribution The mathematical expectation calculated from the source domain samples; To meet the target domain distribution The mathematical expectation calculated from the target domain samples; are the i- th source domain sample and the j -th target domain sample respectively; is the joint variable of the feature g of the i- th source domain sample and the predicted result p ; is the joint variable of the feature g and the prediction result p of the j -th target domain sample; is a multilinear mapping; is the domain discriminator; H is the reproducing kernel Hilbert space, is the characteristic mapping function mapped to the reproducing kernel Hilbert space; The features extracted from the i- th source domain sample; is the feature extracted from the i- th target domain sample. 6.一种滚动轴承故障诊断系统，其特征在于，包括：6. A rolling bearing fault diagnosis system, characterized by comprising: 特征提取模块，被配置为以获取的滚动轴承的待测振动信号为目标域样本，以历史振动信号为源域样本，对目标域样本和源域样本进行特征提取，得到源域特征和目标域特征；The feature extraction module is configured to use the vibration signal of the rolling bearing to be measured as the target domain sample and the historical vibration signal as the source domain sample, perform feature extraction on the target domain sample and the source domain sample, and obtain the source domain feature and the target domain feature; 对目标域样本和源域样本，采用稀疏小波卷积模块，分别进行特征提取；具体包括：稀疏小波卷积模块包括小波卷积层和由两个全连接层和两个激活函数构成的稀疏约束模块，且在小波卷积层中引入谱峭度约束项，在稀疏约束模块中引入稀疏约束项；小波卷积层采用具有平移参数和尺度参数的小波卷积核进行初始化，以初始化后的小波卷积层和稀疏约束模块对目标域样本和源域样本分别进行特征提取；具体为：For target domain samples and source domain samples, sparse wavelet convolution modules are used to extract features respectively; specifically, the sparse wavelet convolution module includes a wavelet convolution layer and a sparse constraint module composed of two fully connected layers and two activation functions, and a spectral kurtosis constraint term is introduced in the wavelet convolution layer, and a sparse constraint term is introduced in the sparse constraint module; the wavelet convolution layer is initialized with a wavelet convolution kernel with a translation parameter and a scale parameter, and the initialized wavelet convolution layer and sparse constraint module are used to extract features for target domain samples and source domain samples respectively; specifically, ；； ; ; 其中，u和s分别为平移参数和尺度参数，t为卷积核的索引，为所选择的小波基函数，为输入的样本，h为小波卷积层的输出，为卷积运算；Among them, u and s are the translation parameter and scale parameter respectively, t is the index of the convolution kernel, is the selected wavelet basis function, is the input sample, h is the output of the wavelet convolution layer, is the convolution operation; 在稀疏约束模块中，通过全连接层将一个随机向量变换为原长度的一半，然后用ReLU激活函数转换，随后通过另一个全连接层恢复到原来的长度，最后用sigmoid激活函数归一化到（0，1）范围内，得到最终稀疏向量V，将h乘以稀疏向量V，得到输出的特征；In the sparse constraint module, a random vector is transformed to half of its original length through a fully connected layer, then converted using the ReLU activation function, and then restored to its original length through another fully connected layer. Finally, it is normalized to the range of (0, 1) using the sigmoid activation function to obtain the final sparse vector V. h is multiplied by the sparse vector V to obtain the output features. 训练模块，被配置为基于源域特征和目标域特征对构建的双模式对抗深度迁移学习网络进行训练；其中，基于源域特征和目标域特征对双分类器进行训练，得到源域分类误差和双分类器确定性误差，基于由源域特征、目标域特征和双分类器的预测结果构成的多线性映射对域判别器进行训练，得到域分类误差，再结合源域特征和目标域特征的最大均值差异误差，作为训练过程的损失函数；A training module is configured to train a dual-mode adversarial deep transfer learning network constructed based on source domain features and target domain features; wherein the dual classifier is trained based on the source domain features and the target domain features to obtain a source domain classification error and a dual classifier deterministic error, and the domain discriminator is trained based on a multilinear mapping composed of the source domain features, the target domain features and the prediction results of the dual classifier to obtain a domain classification error, and then the maximum mean difference error of the source domain features and the target domain features is combined as a loss function of the training process; 分类模块，被配置为采用训练后的双模式对抗深度迁移学习网络对目标域样本进行分类，得到故障诊断结果。The classification module is configured to classify the target domain samples using the trained dual-mode adversarial deep transfer learning network to obtain fault diagnosis results. 7.一种电子设备，其特征在于，包括存储器和处理器以及存储在存储器上并在处理器上运行的计算机指令，所述计算机指令被处理器运行时，完成权利要求1-5任一项所述的方法。7. An electronic device, characterized in that it comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in any one of claims 1 to 5 is performed. 8.一种计算机可读存储介质，其特征在于，用于存储计算机指令，所述计算机指令被处理器执行时，完成权利要求1-5任一项所述的方法。8. A computer-readable storage medium, characterized in that it is used to store computer instructions, and when the computer instructions are executed by a processor, the method described in any one of claims 1-5 is completed. 9.一种计算机程序产品，其特征在于，包括计算机程序，所述计算机程序被处理器执行时实现完成权利要求1-5任一项所述的方法。9. A computer program product, characterized in that it comprises a computer program, and when the computer program is executed by a processor, it implements the method according to any one of claims 1 to 5.
CN202510549457.6A 2025-04-29 2025-04-29 Rolling bearing fault diagnosis method, system, device, medium and program product Active CN120086803B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202510549457.6A CN120086803B (en) 2025-04-29 2025-04-29 Rolling bearing fault diagnosis method, system, device, medium and program product

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202510549457.6A CN120086803B (en) 2025-04-29 2025-04-29 Rolling bearing fault diagnosis method, system, device, medium and program product

Publications (2)

Publication Number Publication Date
CN120086803A CN120086803A (en) 2025-06-03
CN120086803B true CN120086803B (en) 2025-07-15

Family

ID=95848944

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202510549457.6A Active CN120086803B (en) 2025-04-29 2025-04-29 Rolling bearing fault diagnosis method, system, device, medium and program product

Country Status (1)

Country Link
CN (1) CN120086803B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN120524302B (en) * 2025-07-23 2025-09-19 太原理工大学 Bearing fault diagnosis method based on cross-domain adaptive weighting
CN120890673B (en) * 2025-09-30 2025-12-09 山东大学 A Method and System for Fault Diagnosis of Rotating Machinery Based on Gaussian Boundary Constraint Networks

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116451150A (en) * 2023-04-04 2023-07-18 河北工业大学 Equipment fault diagnosis method based on semi-supervised small sample
CN116793682A (en) * 2023-07-07 2023-09-22 武汉理工大学 Bearing fault diagnosis method based on iCORAL-MMD and anti-migration learning

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111651937B (en) * 2020-06-03 2023-07-25 苏州大学 Fault Diagnosis Method of Intraclass Adaptive Bearing under Variable Working Conditions
CN118227957A (en) * 2022-12-20 2024-06-21 中国石油化工股份有限公司 Device fault diagnosis method, machine-readable storage medium and processor
CN116894187A (en) * 2023-07-12 2023-10-17 山东省计算中心（国家超级计算济南中心） A gearbox fault diagnosis method based on deep transfer learning
CN119322967A (en) * 2024-09-26 2025-01-17 北京工业大学 Vacuum dry pump bearing fault diagnosis method based on domain countermeasure and attention transfer learning

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116451150A (en) * 2023-04-04 2023-07-18 河北工业大学 Equipment fault diagnosis method based on semi-supervised small sample
CN116793682A (en) * 2023-07-07 2023-09-22 武汉理工大学 Bearing fault diagnosis method based on iCORAL-MMD and anti-migration learning

Also Published As

Publication number Publication date
CN120086803A (en) 2025-06-03

Similar Documents

Publication Publication Date Title
CN120086803B (en) Rolling bearing fault diagnosis method, system, device, medium and program product
CN112764024B (en) Radar target identification method based on convolutional neural network and Bert
CN106973057B (en) A kind of classification method suitable for intrusion detection
CN110738235B (en) Pulmonary tuberculosis judging method, device, computer equipment and storage medium
CN107643181B (en) A Fault Diagnosis Method for Rolling Bearings under Variable Working Conditions Based on Image Recognition
CN111612035A (en) Method for training transfer model, method and device for fault detection
CN107408209A (en) Automatic defect classification without sampling and feature selection
CN117036843B (en) Target detection model training method, target detection method and device
Kim et al. Label-preserving data augmentation for mobile sensor data
CN110717554A (en) Image recognition method, electronic device, and storage medium
CN110516305A (en) Intelligent fault diagnosis method under small sample based on attention mechanism meta-learning model
KR102734292B1 (en) Method and apparatus for classifying data, method and apparatus for training classifier
CN112132776A (en) Visual detection method and system, storage medium and device based on federated learning
CN114897764A (en) Pulmonary nodule false positive elimination method and device based on standardized channel attention
Shoohi et al. DCGAN for Handling Imbalanced Malaria Dataset based on Over-Sampling Technique and using CNN.
US20230104028A1 (en) System for failure prediction for industrial systems with scarce failures and sensor time series of arbitrary granularity using functional generative adversarial networks
CN113592071A (en) Equipment fault recognition model training and recognition method, system, device and medium
CN119558184B (en) Deep learning-based method and system for predicting residual life of aeroengine
Tasnim et al. Comparison of CNN and QCNN performance in binary classification of breast cancer histopathological images
CN113818191B (en) Washing control method, device, equipment and storage medium based on image recognition
JP2024066425A (en) Ensemble model-based labeling method and device
CN118965201B (en) A malware detection and classification method and system based on multimodal feature fusion
Son et al. Combination of facial recognition and interaction with academic portal in automatic attendance system
CN111598020A (en) Electromagnetic signal identification method based on semi-supervised generation countermeasure network
CN113553899B (en) Gun gesture perception recognition method and 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