CN111006865A - Motor bearing fault diagnosis method - Google Patents
Motor bearing fault diagnosis method Download PDFInfo
- Publication number
- CN111006865A CN111006865A CN201911121072.0A CN201911121072A CN111006865A CN 111006865 A CN111006865 A CN 111006865A CN 201911121072 A CN201911121072 A CN 201911121072A CN 111006865 A CN111006865 A CN 111006865A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- layer
- training
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
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
-
- 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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (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 relates to a motor bearing fault diagnosis method, which comprises the following steps: step S1: establishing a generation countermeasure network of small sample data categories by a GAN training method based on a discrimination model and a generation model, and generating a data set conforming to the characteristics of the small categories; step S2: adding the generated data set into an original subclass sample training set to form a balanced data set; step S3: constructing a discrimination model and a deep convolutional neural network for generating the model, wherein the deep convolutional neural network of the discrimination model comprises three convolutional layers and three corresponding pooling layers, two full-connection layers are arranged behind the third pooling layer, and an optimized balanced data set is used as a training set of the deep convolutional neural network; step S4: the training set learns the fault characteristics layer by layer in a self-adaptive mode from training data, and diagnosis and identification of different fault categories are completed through a classifier. Compared with the prior art, the method has the advantages of adaptively learning the fault characteristics, improving the diagnosis and identification rate of the fault with smaller data volume and the like.
Description
Technical Field
The invention relates to the field of mechanical equipment fault diagnosis, in particular to a motor bearing fault diagnosis method.
Background
The traditional fault diagnosis mainly processes data, and obtains characteristics including fault information by performing a series of time domain and frequency domain analysis on data such as current, vibration, electromagnetism and temperature of a motor, and then evaluates the characteristics, such as empirical mode decomposition, morphological filtering and wavelet packet transformation. With the development of intelligent algorithms such as machine learning, pattern recognition and the like, the intelligent algorithms are used for carrying out feature extraction on data, then various optimization algorithms are used, and finally a specific classifier is used for completing diagnosis and recognition of motor faults.
At the initial stage of motor fault, the influence on current and vibration signals is small, and the acquisition precision of acquisition equipment and the accuracy of a fault diagnosis method are required, so that certain cost problems and algorithm difficulty are caused. The traditional fault diagnosis method cannot realize the discrimination of weak signals. The machine learning algorithm relies on a signal processing technology and diagnosis experience to extract fault features and can finish fault diagnosis work on fixed equipment, but the complex industrial environment has great influence on equipment operation data, and the traditional method is difficult to finish the fault diagnosis work accurately.
Chinese patent CN201811590223.2 discloses a motor fault diagnosis method and system, which utilizes kernel density to process training data, and utilizes a supervised learning model to compare fault data with normal data to judge fault types, but has the defect of insufficient data feature extraction capability, and the supervised learning model is difficult to be applied to actual production and life, and does not consider a small number of fault types.
Chinese patent CN201811082332.3 discloses an interpolation method of wind measurement missing data of a hub of a wind turbine based on GAN, which applies a generated confrontation network to the field of fault diagnosis, inputs training samples into a generator model and a confrontation model for iterative training, and completes the interpolation work of data.
Disclosure of Invention
The invention aims to overcome the defects of the motor fault diagnosis method lacking unbalanced data and low diagnosis recognition rate of fault categories with small data volume in the prior art, and provides the unbalanced data motor fault diagnosis method based on the generation countermeasure network.
The purpose of the invention can be realized by the following technical scheme:
a motor bearing fault diagnosis method comprises the following steps:
step S1: establishing a generation countermeasure network of small sample data categories by a GAN training method based on a discrimination model and a generation model, and generating a data set conforming to the characteristics of the small categories;
step S2: adding the data set generated in the step S1 into an original subclass sample training set to form a balanced data set;
step S3: constructing a discrimination model and a deep convolutional neural network for generating the model, wherein the deep convolutional neural network of the discrimination model comprises three convolutional layers, a pooling layer is arranged behind each convolutional layer, two full-connection layers are arranged behind the pooling layer of the third convolutional layer, the balanced data set optimized in S2 is used as a training set of the deep convolutional neural network, and simultaneously, 20% of data is randomly selected from the training set to be used as a cross-validation data set;
step S4: the training set learns the fault characteristics layer by layer in a self-adaptive manner from the training data, records the learned fault characteristics into the classifier, and finally completes diagnosis and identification of different fault categories through the classifier.
The GAN training method specifically comprises the following steps:
wherein x represents a true sample; z represents random noise in the input generative model; d (x) represents the probability that the input sample is judged to be a real sample by the discriminant model; g (z) represents a sample generated after the generation model receives random noise; pdata(x) Representing the true data distribution; pz(z) represents generating a data distribution.
The discrimination model can accurately judge the authenticity of the input sample, so that D (x) is infinitely close to 1 and D (G (z)) is infinitely close to 0, and V (D, G) is enlarged at the moment, namely maxD is obtained.
The generative model can generate samples closer to the real data, such that D (G (z)) is infinitely close to 1, at which time V (D, G) becomes smaller, i.e., minG is found.
Preferably, in the network structure of the generative model, the output layer uses a Tanh activation function, and the rest uses a ReLU activation function to solve the situation of gradient disappearance and accelerate the convergence speed; in the network structure of the discrimination model, the output layer adopts a Sigmoid activation function, and the rest adopt LeakyReLU activation functions to compress data, so that the forward propagation of a convolution result is facilitated, and the situation of gradient disappearance is solved.
And the data of the layers of the generative model except the output layer are processed by batch standardization, and the data of the layers of the discriminant model except the input layer are processed by batch standardization.
The deep convolutional neural network of the discrimination model comprises three convolutional layers, a pooling layer is arranged behind each convolutional layer, and two full-connection layers are arranged behind the pooling layer of the third convolutional layer.
In the deep convolutional neural network of the generated model, batch normalization processing is required to be carried out between two deconvolution layers.
The training size of each batch in the deep convolutional neural network of the generative model is 10 samples.
The deep convolutional neural network is optimized by adopting an Adam method.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the invention, a small sample data category generation countermeasure network is established through a GAN training method based on a discrimination model and a generation model, the network of the generation model learns the distribution rule of real samples to generate new samples, the network of the discrimination model performs back propagation on the discrimination results of the real samples and the generated samples and optimizes internal parameters, so that the generator is prompted to generate more real data samples, and the generated data is supplemented to a training set with insufficient data, thereby improving the diagnosis and identification rate of the category faults.
2. The deep convolutional neural network adaptively learns the fault characteristics layer by performing characteristic extraction on the samples, realizes characteristic learning of different faults, and finally realizes diagnosis and identification of the fault categories of the small samples.
3. The method takes the balanced data set as a training set of the deep convolutional neural network, simultaneously randomly selects 20% of data from the training set as a cross validation data set, and improves the training speed and precision by utilizing a cross validation mode.
4. The batch processing method used in the invention can solve the problem of poor initialization effect, help gradient propagation to each layer of the network, accelerate model convergence, simultaneously effectively slow down the problem of model overfitting, avoid model collapse and prevent generated samples from converging to the same point.
5. The deep convolutional neural network is optimized by adopting an Adam method, the learning rate of each parameter is dynamically adjusted by utilizing first moment estimation and second moment estimation of the gradient, and the weight is updated to obtain a global optimal solution in a back propagation stage, so that a loss function reaches the minimum value.
Drawings
FIG. 1 is a schematic diagram of the structure of the generation of a countermeasure network according to the present invention;
FIG. 2 is a schematic diagram of the structure of the generative model of the present invention;
FIG. 3 is a schematic diagram of a discriminant model according to the present invention;
FIG. 4 is a schematic flow chart of the motor fault diagnosis of the present invention;
FIG. 5 is a schematic diagram of the generation of a countering network loss value according to the present invention;
FIG. 6 is a graph comparing the spectra of raw and generated data according to the present invention;
FIG. 7 is a graph illustrating the rate of change identified in the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
A motor bearing fault diagnosis method comprises the following steps:
step S1: establishing a generation countermeasure network of small sample data categories by a GAN training method based on a discrimination model and a generation model, and generating a data set conforming to the characteristics of the small categories;
step S2: adding the data set generated in the step S1 into an original subclass sample training set to form a balanced data set;
step S3: constructing a discrimination model and a deep convolutional neural network for generating the model, wherein the deep convolutional neural network of the discrimination model comprises three convolutional layers, a pooling layer is arranged behind each convolutional layer, two full-connection layers are arranged behind the pooling layer of the third convolutional layer, the balanced data set optimized in S2 is used as a training set of the deep convolutional neural network, and simultaneously, 20% of data is randomly selected from the training set to be used as a cross validation data set;
step S4: the training set learns the fault characteristics layer by layer in a self-adaptive mode from the training data, records the learned fault characteristics into the classifier, and finally completes diagnosis and identification of different fault categories through the classifier.
The GAN training method specifically comprises the following steps:
wherein x represents a true sample; z represents random noise in the input generative model; d (x) represents the probability that the input sample is judged to be a real sample by the discriminant model; g (z) represents a sample generated after the generation model receives random noise; pdata(x) Representing the true data distribution; pz(z) represents generating a data distribution.
The generative model can generate samples closer to the real data, such that D (G (z)) is infinitely close to 1, at which time V (D, G) becomes smaller, i.e., minG is obtained. As shown in fig. 2, in the generated model, first, 100 nodes are set at an input end, 1024 nodes are set at a first hidden layer, each node at the input end is multiplied by an initialized network weight, then an offset value is added, the first hidden layer processes results of all input nodes through an activation function to serve as an input value of a next layer, 3200 nodes of the next layer are set in the same method to achieve the purpose of data expansion, then data samples are reset, the expanded data is single-column data, calculation after the expansion is inconvenient, data flattening processing is needed, 3200 data generates single-side data of 25x128, then 5x5x128 multi-layer data of 5x5x128 layers are generated, and finally the data is reset to (5, 128) data to facilitate convolution operation. The reset sample passes through two layers of deconvolution layers, so that the number of channels of input data is reduced, the image size is increased, the characteristics of the data sample are extracted, and a (20,20,1) false sample is output. The convolution kernels of the deconvolution are all 4x4 in size, and the step sizes are all set to 2.
The discrimination model can accurately judge the authenticity of the input sample, so that D (x) is infinitely close to 1 and D (G (z)) is infinitely close to 0, and V (D, G) is increased at the moment, namely maxD is obtained. As shown in fig. 3, the input real data or false data is a sample of (20,20,1), the size of the input real data or false data becomes 3200 dimensions after the data is flattened through two convolution networks, then the input real data or false data passes through three full connection layers, so that the image size becomes smaller, the number of channels becomes larger, and finally one-dimensional data is output, that is, the input data is judged to be true or false.
The method comprises the steps that a random noise signal generates a fake sample through a generation model, the fake sample and a true sample are input into a discrimination network to respectively judge whether the random noise signal is true or false, a loss value is generated by utilizing discrimination error probability, the loss value is fed back to the generation model to improve various parameters of the generation model, a more vivid fake sample is generated for the next training, a generation countermeasure network with closed-loop feedback is finally formed until the network reaches nash balance, fake data which is infinitely close to the true data is generated, the fake data is input into a training set of a deep neural network, the sample number of the training set is expanded, the learning capacity of the deep neural network on fault characteristics is improved, the learned fault characteristics are recorded into a classifier, and the fault type in a test set is judged through the classifier.
In the network structure of the generated model, the output layer uses a Tanh activation function, and the rest uses a ReLU activation function to solve the situation of gradient disappearance and accelerate the convergence speed; in the network structure of the discrimination model, the output layer adopts a Sigmoid activation function, and the rest adopt LeakyReLU activation functions to carry out data compression, so that the forward propagation of a convolution result is facilitated, and the condition that the gradient disappears is solved.
The data of the layers of the generated model except the output layer are processed by batch standardization, and the data of the layers of the discriminant model except the input layer are processed by batch standardization.
The deep convolutional neural network of the discrimination model comprises three convolutional layers, a pooling layer is arranged behind each convolutional layer, two full-connection layers are arranged behind the pooling layer of the third convolutional layer, and the softmax function is used for final classification.
In the deep convolutional neural network of the generated model, batch normalization processing is required to be carried out between two deconvolution layers.
The training size of each batch in the deep convolutional neural network of the generative model was 10 samples.
And the deep convolutional neural network is optimized by adopting an Adam method.
Example one
The method comprises the steps that an open motor bearing data set provided by a western storage university laboratory is used as a training set, wherein the data set comprises motor normal bearing data, drive end bearing fault data and fan end bearing fault data, the bearing fault positions comprise rolling element faults, inner ring faults and outer ring faults, and the fault sizes are respectively 0.007 inches, 0.014 inches, 0.021 inches and 0.028 inches to simulate different fault degrees of a bearing; the rotating speeds of the motors are 1797, 1772, 1750 and 1730r/min respectively, the motors correspond to three loads of 0,1, 2 and 3hp respectively, and the sampling frequency of the motors is 12 kHz; the length of a single sample is 400, the updating times ratio of the discriminant model to the generative model is 1: 2, the learning rates of the generative model and the antagonistic model are 0.0001 and 0.00001 respectively, and 10 samples are read in each batch.
As shown in fig. 5, after 5400 times of training, the final loss value is kept around 0.5, which shows that the generated data is very close to the real data, and the convergence rate is also very fast.
As shown in fig. 6, the real data and the data generated by the generative model are subjected to fourier transform to generate a corresponding spectrogram, and the spectrogram of the generated data corresponding to the generative model has fluctuation characteristics identical to those of the real data in three positions of 50Hz, 100Hz and 300Hz in the overall trend, so that the data generated by the generative model can be considered to have the main characteristics of the original real data.
As shown in fig. 7, after 2500 cycles of the training, each parameter of the network model is optimized from the initial 20% recognition rate with the increase of the training times, the recognition rate is continuously increased, the recognition rate is 90% when the training reaches 500 times, and the final recognition rate is kept at about 95%.
In addition, it should be noted that the specific embodiments described in the present specification may have different names, and the above descriptions in the present specification are only illustrations of the structures of the present invention. Minor or simple variations in the structure, features and principles of the present invention are included within the scope of the present invention. Various modifications or additions may be made to the described embodiments or methods may be similarly employed by those skilled in the art without departing from the scope of the invention as defined in the appending claims.
Claims (9)
1. A motor bearing fault diagnosis method is characterized by comprising the following steps:
step S1: establishing a generation countermeasure network of small sample data categories by a GAN training method based on a discrimination model and a generation model, and generating a data set conforming to the characteristics of the small categories;
step S2: adding the data set generated in the step S1 into an original subclass sample training set to form a balanced data set;
step S3: constructing a discrimination model and a deep convolutional neural network for generating the model, wherein the deep convolutional neural network of the discrimination model comprises three convolutional layers, a pooling layer is arranged behind each convolutional layer, two full-connection layers are arranged behind the pooling layer of the third convolutional layer, the balanced data set optimized in S2 is used as a training set of the deep convolutional neural network, and simultaneously, 20% of data is randomly selected from the training set to be used as a cross-validation data set;
step S4: the training set learns the fault characteristics layer by layer in a self-adaptive manner from the training data, records the learned fault characteristics into the classifier, and finally completes diagnosis and identification of different fault categories through the classifier.
2. The motor bearing fault diagnosis method according to claim 1, wherein the GAN training method specifically comprises:
wherein x represents a true sample; z represents random noise in the input generative model; d (x) represents the probability that the input sample is judged to be a real sample by the discriminant model; g (z) represents a sample generated after the generation model receives random noise; pdata(x) Representing the true data distribution; pz(z) represents generating a data distribution.
3. The method as claimed in claim 2, wherein the discrimination model is capable of accurately determining the authenticity of the input sample, such that D (x) is infinitely close to 1 and D (G (z)) is infinitely close to 0, and when V (D, G) becomes large, maxD is obtained.
4. The method according to claim 2, wherein the generative model is capable of generating a sample closer to real data, such that D (G (z)) is infinitely close to 1, and V (D, G) is decreased to obtain minG.
5. The motor bearing fault diagnosis method according to claim 1, wherein in the network structure of the generative model, the output layer uses Tanh activation function, and the rest uses ReLU activation function to solve the situation of gradient disappearance and accelerate the convergence speed; in the network structure of the discrimination model, the output layer adopts a Sigmoid activation function, and the rest adopt LeakyReLU activation functions to compress data, so that the forward propagation of a convolution result is facilitated, and the situation of gradient disappearance is solved.
6. The method as claimed in claim 1, wherein the data of the layers of the generative model other than the output layer are processed by batch normalization, and the data of the layers of the discriminant model other than the input layer are processed by batch normalization.
7. The method of claim 1, wherein a batch normalization process is required between two deconvolution layers in the deep convolutional neural network of the generative model.
8. The method of claim 1, wherein the training size of each batch in the deep convolutional neural network of the generative model is 10 samples.
9. The motor bearing fault diagnosis method according to claim 1, characterized in that the deep convolutional neural network is optimized by using an Adam method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911121072.0A CN111006865A (en) | 2019-11-15 | 2019-11-15 | Motor bearing fault diagnosis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911121072.0A CN111006865A (en) | 2019-11-15 | 2019-11-15 | Motor bearing fault diagnosis method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111006865A true CN111006865A (en) | 2020-04-14 |
Family
ID=70112949
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911121072.0A Pending CN111006865A (en) | 2019-11-15 | 2019-11-15 | Motor bearing fault diagnosis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111006865A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112504678A (en) * | 2020-11-12 | 2021-03-16 | 重庆科技学院 | Motor bearing fault diagnosis method based on knowledge distillation |
| CN113032917A (en) * | 2021-03-03 | 2021-06-25 | 安徽大学 | Electromechanical bearing fault detection method based on generation countermeasure and convolution cyclic neural network and application system |
| CN113221946A (en) * | 2021-04-02 | 2021-08-06 | 中国人民解放军92578部队 | Method for diagnosing fault types of mechanical equipment |
| CN113255458A (en) * | 2021-04-29 | 2021-08-13 | 燕山大学 | Bearing fault diagnosis method based on multi-view associated feature learning |
| CN113376516A (en) * | 2021-06-07 | 2021-09-10 | 科润智能控制股份有限公司 | Medium-voltage vacuum circuit breaker operation fault self-diagnosis and early-warning method based on deep learning |
| CN113421262A (en) * | 2021-08-23 | 2021-09-21 | 深圳市信润富联数字科技有限公司 | Hub defect detection method and device, electronic equipment and storage medium |
| CN113486931A (en) * | 2021-06-21 | 2021-10-08 | 南京航空航天大学 | Rolling bearing enhancement diagnosis method based on PDA-WGANGP |
| CN113485863A (en) * | 2021-07-14 | 2021-10-08 | 北京航空航天大学 | Method for generating heterogeneous unbalanced fault samples based on improved generation countermeasure network |
| CN113887136A (en) * | 2021-10-08 | 2022-01-04 | 东北大学 | Improved GAN and ResNet based electric vehicle motor bearing fault diagnosis method |
| CN114323642A (en) * | 2020-09-29 | 2022-04-12 | 北京金风慧能技术有限公司 | Wind turbine generator vibration data processing system and data dilution method |
| CN114491823A (en) * | 2022-03-28 | 2022-05-13 | 西南交通大学 | Train bearing fault diagnosis method based on improved generation countermeasure network |
| CN115169506A (en) * | 2022-09-06 | 2022-10-11 | 中铁第四勘察设计院集团有限公司 | Method and system for rapidly diagnosing faults of power supply and transformation key equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108647786A (en) * | 2018-07-10 | 2018-10-12 | 电子科技大学 | The rotating machinery on-line fault monitoring method of neural network is fought based on depth convolution |
| EP3407292A1 (en) * | 2017-05-24 | 2018-11-28 | General Electric Company | Neural network point cloud generation system |
| CN109489946A (en) * | 2018-09-21 | 2019-03-19 | 华中科技大学 | A kind of fault diagnosis method and system of rotating machinery |
| CN109918999A (en) * | 2019-01-22 | 2019-06-21 | 西安交通大学 | Based on the mechanical equipment fault intelligent diagnosing method for generating model under a kind of Small Sample Database |
| CN110188822A (en) * | 2019-05-30 | 2019-08-30 | 盐城工学院 | A kind of domain is to the one-dimensional convolutional neural networks intelligent failure diagnosis method of anti-adaptive |
-
2019
- 2019-11-15 CN CN201911121072.0A patent/CN111006865A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3407292A1 (en) * | 2017-05-24 | 2018-11-28 | General Electric Company | Neural network point cloud generation system |
| CN108647786A (en) * | 2018-07-10 | 2018-10-12 | 电子科技大学 | The rotating machinery on-line fault monitoring method of neural network is fought based on depth convolution |
| CN109489946A (en) * | 2018-09-21 | 2019-03-19 | 华中科技大学 | A kind of fault diagnosis method and system of rotating machinery |
| CN109918999A (en) * | 2019-01-22 | 2019-06-21 | 西安交通大学 | Based on the mechanical equipment fault intelligent diagnosing method for generating model under a kind of Small Sample Database |
| CN110188822A (en) * | 2019-05-30 | 2019-08-30 | 盐城工学院 | A kind of domain is to the one-dimensional convolutional neural networks intelligent failure diagnosis method of anti-adaptive |
Non-Patent Citations (2)
| Title |
|---|
| 包萍 等: "不均衡数据集下基于生成对抗网络的改进深度模型故障识别研究", 《电子测量与仪器学报》 * |
| 张彬 等: "《图像复原优化算法》", 31 August 2019, 国防工业出版社 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114323642A (en) * | 2020-09-29 | 2022-04-12 | 北京金风慧能技术有限公司 | Wind turbine generator vibration data processing system and data dilution method |
| CN112504678A (en) * | 2020-11-12 | 2021-03-16 | 重庆科技学院 | Motor bearing fault diagnosis method based on knowledge distillation |
| CN113032917A (en) * | 2021-03-03 | 2021-06-25 | 安徽大学 | Electromechanical bearing fault detection method based on generation countermeasure and convolution cyclic neural network and application system |
| CN113221946A (en) * | 2021-04-02 | 2021-08-06 | 中国人民解放军92578部队 | Method for diagnosing fault types of mechanical equipment |
| CN113255458A (en) * | 2021-04-29 | 2021-08-13 | 燕山大学 | Bearing fault diagnosis method based on multi-view associated feature learning |
| CN113376516A (en) * | 2021-06-07 | 2021-09-10 | 科润智能控制股份有限公司 | Medium-voltage vacuum circuit breaker operation fault self-diagnosis and early-warning method based on deep learning |
| CN113486931A (en) * | 2021-06-21 | 2021-10-08 | 南京航空航天大学 | Rolling bearing enhancement diagnosis method based on PDA-WGANGP |
| CN113486931B (en) * | 2021-06-21 | 2023-05-26 | 南京航空航天大学 | Rolling bearing enhanced diagnosis method based on PDA-WGANGP |
| CN113485863A (en) * | 2021-07-14 | 2021-10-08 | 北京航空航天大学 | Method for generating heterogeneous unbalanced fault samples based on improved generation countermeasure network |
| CN113421262A (en) * | 2021-08-23 | 2021-09-21 | 深圳市信润富联数字科技有限公司 | Hub defect detection method and device, electronic equipment and storage medium |
| CN113887136A (en) * | 2021-10-08 | 2022-01-04 | 东北大学 | Improved GAN and ResNet based electric vehicle motor bearing fault diagnosis method |
| CN113887136B (en) * | 2021-10-08 | 2024-05-14 | 东北大学 | Electric automobile motor bearing fault diagnosis method based on improved GAN and ResNet |
| CN114491823A (en) * | 2022-03-28 | 2022-05-13 | 西南交通大学 | Train bearing fault diagnosis method based on improved generation countermeasure network |
| CN115169506A (en) * | 2022-09-06 | 2022-10-11 | 中铁第四勘察设计院集团有限公司 | Method and system for rapidly diagnosing faults of power supply and transformation key equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111006865A (en) | Motor bearing fault diagnosis method | |
| CN108334936B (en) | Fault prediction method based on migration convolutional neural network | |
| CN109299705B (en) | Rotary machine fault diagnosis method based on one-dimensional depth residual convolution neural network | |
| CN112183581A (en) | Semi-supervised mechanical fault diagnosis method based on self-adaptive migration neural network | |
| CN111351665B (en) | Rolling bearing fault diagnosis method based on EMD and residual error neural network | |
| CN111783841A (en) | Garbage classification method, system and medium based on transfer learning and model fusion | |
| CN113033309A (en) | Fault diagnosis method based on signal downsampling and one-dimensional convolution neural network | |
| Miao et al. | Deep sparse representation network for feature learning of vibration signals and its application in gearbox fault diagnosis | |
| CN113252790A (en) | Magnetic shoe internal defect detection method based on wide convolution and cyclic neural network | |
| CN114091553A (en) | Diagnosis method for rolling bearing fault | |
| CN115962946A (en) | Bearing fault diagnosis method based on improved WGAN-GP and Alxnet | |
| CN114491823B (en) | Train bearing fault diagnosis method based on improved generation countermeasure network | |
| CN112910812A (en) | Modulation mode identification method for deep learning based on space-time feature extraction | |
| Lv et al. | Deep transfer network with multi-kernel dynamic distribution adaptation for cross-machine fault diagnosis | |
| Peng et al. | Research on fault diagnosis method of rolling bearing based on 2DCNN | |
| Shi et al. | A novel rotating machinery fault diagnosis method based on adaptive deep belief network structure and dynamic learning rate under variable working conditions | |
| CN114186589A (en) | Superconducting cable partial discharge mode identification method based on residual error network Resnet50 | |
| Liu et al. | Fault diagnosis of satellite flywheel bearing based on convolutional neural network | |
| CN113420870B (en) | U-Net structure generation countermeasure network and method for underwater sound target recognition | |
| CN115082435A (en) | Defect detection method based on self-supervision momentum contrast | |
| CN114140736A (en) | Image anomaly detection method based on high-frequency and low-frequency reconstruction | |
| Tuo et al. | A deep learning model for bearing fault diagnosis based on convolution neural network with multi-channel and residual network | |
| Liu et al. | Generative adversarial network based multi-class imbalanced fault diagnosis of rolling bearing | |
| CN112686366A (en) | Bearing fault diagnosis method based on random search and convolutional neural network | |
| CN116625689B (en) | Rolling bearing fault diagnosis method and system based on SMDER |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200414 |
|
| RJ01 | Rejection of invention patent application after publication |