CN113032916A - Electromechanical device bearing fault prediction method based on Bayesian network of transfer learning - Google Patents
Electromechanical device bearing fault prediction method based on Bayesian network of transfer learning Download PDFInfo
- Publication number
- CN113032916A CN113032916A CN202110236878.5A CN202110236878A CN113032916A CN 113032916 A CN113032916 A CN 113032916A CN 202110236878 A CN202110236878 A CN 202110236878A CN 113032916 A CN113032916 A CN 113032916A
- Authority
- CN
- China
- Prior art keywords
- target domain
- bayesian network
- domain
- source domain
- fault
- 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.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Abstract
The invention discloses a method for predicting the bearing fault of electromechanical equipment based on a Bayesian network of transfer learning, which comprises the following specific steps: denoising an original acquisition signal by CEEMDAN and FastICA technologies to form a reconstructed original signal; introducing transfer learning, clustering original signals through a neural network to obtain a signal set classified according to fault types, taking the signal set as a target domain data set to become an input of a Bayesian network, and selecting a reference sample set in a source domain as a training set of the Bayesian network in the source domain; based on a covariate shift theory, the goal of minimizing the parameter loss value on the target domain is completed by utilizing the training data of the source domain, and the optimization of the maximum likelihood estimation of the Bayesian network on the target domain is realized; and returning a result through output of the Bayesian network on the target domain, performing visual display, giving a warning if the result is in an abnormal state, returning parameter information to the mechanical equipment in a declining state, and indicating the machine to adjust self-adjustable parameters such as rotating speed and the like within a certain range.
Description
Technical Field
The invention belongs to the technical field of fault diagnosis of electromechanical equipment, and particularly relates to a mine mechanical bearing fault diagnosis method based on a transfer learning maximum likelihood estimation Bayesian network.
Background
While the modern industry in China pursues high efficiency, intelligent monitoring becomes important technical force for ensuring the stability of the industry. Mining industry is as the abortion industry on the industrial chain, guarantees its personnel's life safety and the safety and stability of field device are the prerequisite, but often mining industry scene operational environment is relatively complicated, and the major part uses large-scale electromechanical device outdoors, in case the trouble problem often can cause unpredictable loss. The bearing consists of an inner ring, an outer ring, a rolling body, a retainer and lubricating grease. The bearing used in the mining machine is an important component in large-scale equipment, and the working state of the bearing influences the state of the whole machine. At present, large electromechanical equipment such as the large electromechanical equipment mainly depends on manual inspection, repair after fault judgment through manual experience, and a traditional online monitoring method. The problems of poor real-time performance, excessively long unplanned downtime, missing of optimal maintenance time, complex link installation and the like are often caused. Therefore, the effective and timely technical means solves the problem that the real-time state monitoring and early warning fault processing of the mining large-scale equipment become irreparable requirements of the industrial industry.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a Bayesian network mine electromechanical bearing fault degree prediction method based on transfer learning.
The technical scheme adopted by the invention is as follows: the Bayesian network mine field electromechanical bearing fault degree prediction method based on the transfer learning comprises the following steps:
firstly, denoising an original collected signal by CEEDAN and FastICA technologies to form a reconstructed original signal, and extracting a feature vector by utilizing LLE dimension reduction;
step two, introducing transfer learning, clustering the original signals through a neural network to obtain a signal set classified according to fault types, and taking the signal set as a target domain data set to become the input of the Bayesian network;
selecting a reference sample set in a source domain as a training set of the Bayesian network of the source domain; based on a covariate shift theory, the goal of minimizing the parameter loss value on the target domain is completed by utilizing the training data of the source domain, and the optimization of the maximum likelihood estimation of the Bayesian network on the target domain is realized;
and step four, returning results through output of the Bayesian network on the target domain, performing visual embodiment, completing degree prediction of a certain fault type, giving a warning if the fault type is abnormal, returning parameter information to mechanical equipment in a decline state, and indicating a machine to adjust self-adjustable parameters such as rotating speed and the like within a certain range.
In the first step, the method comprises the following steps:
1. because the vibration signal has obvious nonlinearity and the time domain analysis is difficult to accurately judge the specific fault type, the invention samples CEEMDAN empirical mode decomposition to extract the noise reduction characteristic of the signal, and obtains the corresponding characteristic space, and then utilizes LLE dimension reduction to obtain the usable dimension data set sample characteristic space.
2. The calculation formula of the fault characteristic frequency of the rolling bearing is as follows:
rotational frequency of rolling bearing:
frequency of outer ring failure:
frequency of inner ring failure:
frequency of rolling element failure:
frequency of inner and outer loop failures:
wherein dr is the radius of the rolling bodies, D omega is the radius of the pitch circle, Z is the number of the rolling bodies, the rotating speed of the Sp shaft and an alpha pressure angle.
By using the LLE linear local embedding dimension reduction technology, the linear relationship between samples in the sub-low-dimensional space is kept, that is, the weight Wi between the samples is kept unchanged, so that the low-dimensional space coordinate Zi corresponding to Xi can be calculated by the following formula:
obtaining corresponding feature vectors;
in the second step, the method comprises the following steps:
selecting a reference sample set from the source domain data samples by using a K nearest neighbor algorithm, namely selecting a migrated source domain training data object for the target domain;
calculating a similarity matrix by using manifold distance, introducing a transfer learning model mechanism, sharing weight parameters of SOM neural network nodes of a source domain and a target domain of a data sample, and optimizing a target domain clustering task function and neuron weight through a source domain training result;
and finishing the clustering task of the target domain samples through the target domain output layer result of the SOM neural network to obtain a clustering result and a signal set classified according to fault types.
In step three, the method comprises the following steps:
and taking the signal results of different clusters in the step two as the input of the Bayesian network, wherein the operation of the source domain and the target domain is the same.
By reasonably selecting the source domain data samples which are basically distributed the same as the target domain, the Bayesian network on the source domain and the target domain has the same structure, covariate shift exists between the source domain and the target domain, and a theoretical basis is established for transfer learning between the source domain and the target domain. According to the covariate shift assumption, there are:
however, due to the problem of model specification error in the transfer learning, that is, there is no parameter that can accurately describe the relationship between the covariate x and the dependent variable y, the specific expression is as follows:
at this time, direct migration is not possible, so a weighted likelihood function is used to define a parameter loss value function for the target domain bayesian network, a loss function on the target domain is constructed by assigning a weight to the source domain data sample loss function, and an optimal parameter combination of the bayesian network on the target domain is learned, so that the expectation of the loss function of the bayesian network under the parameter is minimum, namely:
the loss function for the entire bayesian network is then:
wherein J = S represents a loss function on the source domain, J = T represents a loss function on the destination domain,
for sample data D, there is a likelihood function:
wherein J = S represents the likelihood function on the source domain, J = T represents the likelihood function on the target domain,
the likelihood function of the random variable Xi on the data Dl is:
when the random variable I and the parent node value thereof appear in the sample, I =1, otherwise I = 0;
when m is
In the process, the maximum likelihood estimation on the source domain approaches to the value of a loss function minimization parameter formula on the source domain, namely the significance is the same under the limit condition, and the maximum likelihood estimation on the target domain is expressed as follows on the target domain in the same way:
and because the assumed conditions of covariate shift are satisfied on the source domain and the target domain, the following are:
substituting the limit formula to obtain:
therefore, a mathematical relation of a minimization loss function of a source domain and a target domain is established, namely when sample data is large enough and tends to infinity, weighted maximum likelihood estimation of a source domain network tends to maximum likelihood estimation on the target domain, in original data, the weight of source domain data is PT (dt), and the weight of target domain data is Ps (Ds), so that the source domain data and the target domain data are mixed and assigned again, wherein the source domain data is assigned with the weight PT (dt)/Ps (Ds), the target domain data is assigned with the weight of 1, and a maximum likelihood estimation formula obtained by calculation is as follows:
therefore, optimization of maximum likelihood estimation of the Bayesian network in the target domain is achieved, and the network analysis learning effect of the target domain under the condition of small sample number is improved.
In step four, the method comprises the following steps:
the system needs to complete the fault visualization and state early warning work on the basis of the configuration of an external alarm system and an internal feedback system.
According to the output result of the target domain Bayesian network, judging the four fault degrees of normal, slight decline, accelerated decline, fault and failure of the equipment;
when the state is judged to be normal, the external alarm system is dormant, and the internal feedback system is dormant
And when the early warning state is judged to be the initial decline early warning state or the rapid decline early warning state, the external alarm system is immediately activated to report the early warning state, and the early warning information comprises a pre-failure component, a pre-failure type and a pre-failure degree. And the internal feedback system judges whether the self-adaptive adjustment capability of the machine is within the range, and automatically adjusts parameters, such as automatically reducing the rotating speed and the like.
And when the state is judged to be a complete fault state, the external alarm system is activated to enter a fault emergency alarm state.
Drawings
FIG. 1 is a flowchart illustrating steps of a method for predicting a bearing fault of an electromechanical device based on a Bayesian network for transfer learning according to an embodiment of the present invention; .
FIG. 2 is a block diagram of an electromechanical diagnostic system in accordance with the present invention;
FIG. 3 is a block diagram illustrating an embodiment of the remote service module of FIG. 1;
FIG. 4 is a block diagram illustrating the traffic interface module of FIG. 1 in one embodiment;
FIG. 5 is a block diagram illustrating background data management of FIG. 1 in one embodiment;
fig. 6 is a schematic diagram showing an overall network architecture flow of a bayesian network in the method for predicting the bearing fault of the electromechanical device based on the bayesian network of the transfer learning according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described below in detail and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention.
Referring to fig. 2, a block diagram of an electromechanical diagnostic system according to the present invention is shown.
Referring to fig. 1, a flowchart illustrating steps of a method for predicting a bearing fault of an electromechanical device based on a bayesian network for transfer learning according to an embodiment of the present invention is shown. The implementation process can be divided into four steps:
firstly, denoising an original collected signal by CEEDAN and FastICA technologies to form a reconstructed original signal, and extracting a feature vector by utilizing LLE dimension reduction;
step two, introducing transfer learning, clustering the original signals through a neural network to obtain a signal set classified according to fault types, and taking the signal set as a target domain data set to become the input of the Bayesian network;
selecting a reference sample set in a source domain as a training set of the Bayesian network of the source domain; based on a covariate shift theory, the goal of minimizing the parameter loss value on the target domain is completed by utilizing the training data of the source domain, and the optimization of the maximum likelihood estimation of the Bayesian network on the target domain is realized;
and step four, returning results through output of the Bayesian network on the target domain, performing visual embodiment, completing degree prediction of a certain fault type, giving a warning if the fault type is abnormal, returning parameter information to mechanical equipment in a decline state, and indicating a machine to adjust self-adjustable parameters such as rotating speed and the like within a certain range.
In the first step, the method comprises the following steps:
1. because the vibration signal has obvious nonlinearity and the time domain analysis is difficult to accurately judge the specific fault type, the invention samples CEEMDAN empirical mode decomposition to extract the noise reduction characteristic of the signal, and obtains the corresponding characteristic space, and then utilizes LLE dimension reduction to obtain the usable dimension data set sample characteristic space.
2. The calculation formula of the fault characteristic frequency of the rolling bearing is as follows:
rotational frequency of rolling bearing:
frequency of outer ring failure:
frequency of inner ring failure:
frequency of rolling element failure:
frequency of inner and outer loop failures:
wherein dr is the radius of the rolling bodies, D omega is the radius of the pitch circle, Z is the number of the rolling bodies, the rotating speed of the Sp shaft and an alpha pressure angle.
3. By using the LLE linear local embedding dimension reduction technology, the linear relationship between samples in the sub-low-dimensional space is kept, that is, the weight Wi between the samples is kept unchanged, so that the low-dimensional space coordinate Zi corresponding to Xi can be calculated by the following formula:
obtaining corresponding feature vectors;
in the second step, the method comprises the following steps:
selecting a reference sample set from the source domain data samples by using a K nearest neighbor algorithm, namely selecting a migrated source domain training data object for the target domain;
calculating a similarity matrix by using manifold distance, introducing a transfer learning model mechanism, sharing weight parameters of SOM neural network nodes of a source domain and a target domain of a data sample, and optimizing a target domain clustering task function and neuron weight through a source domain training result;
and finishing the clustering task of the target domain samples through the target domain output layer result of the SOM neural network to obtain a clustering result and a signal set classified according to fault types.
In step three, the method comprises the following steps:
and taking the signal results of different clusters in the step two as the input of the Bayesian network, wherein the operation of the source domain and the target domain is the same.
By reasonably selecting the source domain data samples which are basically distributed the same as the target domain, the Bayesian network on the source domain and the target domain has the same structure, covariate shift exists between the source domain and the target domain, and a theoretical basis is established for transfer learning between the source domain and the target domain. According to the covariate shift assumption, there are:
however, due to the problem of model specification error in the transfer learning, that is, there is no parameter that can accurately describe the relationship between the covariate x and the dependent variable y, the specific expression is as follows:
at this time, direct migration is not possible, so a weighted likelihood function is used to define a parameter loss value function for the target domain bayesian network, a loss function on the target domain is constructed by assigning a weight to the source domain data sample loss function, and an optimal parameter combination of the bayesian network on the target domain is learned, so that the expectation of the loss function of the bayesian network under the parameter is minimum, namely:
the loss function for the entire bayesian network is then:
wherein J = S represents a loss function on the source domain, J = T represents a loss function on the destination domain,
for sample data D, there is a likelihood function:
wherein J = S represents the likelihood function on the source domain, J = T represents the likelihood function on the target domain,
the likelihood function of the random variable Xi on the data Dl is:
when the random variable I and the parent node value thereof appear in the sample, I =1, otherwise I = 0;
when m is
In the process, the maximum likelihood estimation on the source domain approaches to the value of a loss function minimization parameter formula on the source domain, namely the significance is the same under the limit condition, and the maximum likelihood estimation on the target domain is expressed as follows on the target domain in the same way:
and because the assumed conditions of covariate shift are satisfied on the source domain and the target domain, the following are:
substituting the limit formula to obtain:
therefore, a mathematical relation of a minimization loss function of a source domain and a target domain is established, namely when sample data is large enough and tends to infinity, weighted maximum likelihood estimation of a source domain network tends to maximum likelihood estimation on the target domain, in original data, the weight of source domain data is PT (dt), and the weight of target domain data is Ps (Ds), so that the source domain data and the target domain data are mixed and assigned again, wherein the source domain data is assigned with the weight PT (dt)/Ps (Ds), the target domain data is assigned with the weight of 1, and a maximum likelihood estimation formula obtained by calculation is as follows:
therefore, optimization of maximum likelihood estimation of the Bayesian network in the target domain is achieved, and the network analysis learning effect of the target domain under the condition of small sample number is improved.
Fig. 6 is a schematic diagram showing an overall network architecture flow of a bayesian network in the method for predicting a bearing fault of an electromechanical device based on a bayesian network of a transfer learning according to the present invention.
In step four, the method comprises the following steps:
referring to the figure, the system needs to complete fault visualization and state early warning work on the basis of an external alarm system and an internal feedback system.
According to the output result of the target domain Bayesian network, judging the four fault degrees of normal, slight decline, accelerated decline, fault and failure of the equipment;
when the state is judged to be normal, the external alarm system is dormant, and the internal feedback system is dormant
And when the early warning state is judged to be the initial decline early warning state or the rapid decline early warning state, the external alarm system is immediately activated to report the early warning state, and the early warning information comprises a pre-failure component, a pre-failure type and a pre-failure degree. And the internal feedback system judges whether the self-adaptive adjustment capability of the machine is within the range, and automatically adjusts parameters, such as automatically reducing the rotating speed and the like.
And when the state is judged to be a complete fault state, the external alarm system is activated to enter a fault emergency alarm state.
Referring to FIG. 3, a block diagram illustrating an embodiment of the remote service module of FIG. 2 is shown
Referring to fig. 4, a block diagram of a service interface module of fig. 2 in an embodiment is shown
Please refer to fig. 5, which is a block diagram illustrating an embodiment of the background data management module in fig. 2
The invention is not limited to the foregoing embodiments. The invention extends to any novel feature or any novel combination of features disclosed in this specification and any novel method or process steps or any novel combination of features disclosed. Those skilled in the art to which the invention pertains will appreciate that insubstantial changes or modifications can be made without departing from the spirit of the invention as defined by the appended claims.
Claims (5)
1. A method for predicting the bearing fault of an electromechanical device based on a Bayesian network of transfer learning is characterized by comprising the following steps:
firstly, denoising an original collected signal by CEEDAN and FastICA technologies to form a reconstructed original signal, and extracting a feature vector by utilizing LLE dimension reduction;
step two, introducing transfer learning, clustering the original signals through a neural network to obtain a signal set classified according to fault types, and taking the signal set as a target domain data set to become the input of the Bayesian network;
selecting a reference sample set in a source domain as a training set of the Bayesian network of the source domain; based on a covariate shift theory, the goal of minimizing the parameter loss value on the target domain is completed by utilizing the training data of the source domain, and the optimization of the maximum likelihood estimation of the Bayesian network on the target domain is realized;
and step four, returning results through output of the Bayesian network on the target domain, performing visual embodiment, completing degree prediction of a certain fault type, giving a warning if the fault type is abnormal, returning parameter information to mechanical equipment in a decline state, and indicating a machine to adjust self-adjustable parameters such as rotating speed and the like within a certain range.
2. The method for predicting the bearing fault of the electromechanical device based on the Bayesian network based on the migration learning of claim 1, wherein the step one comprises:
collecting raw vibration signals from a sensor;
decomposing the CEEMDAN self-adaptive noise by using a complete empirical mode decomposition method of the CEEMDAN self-adaptive noise to obtain a plurality of intrinsic mode functions IMF;
denoising by a FastICA algorithm, obtaining a new IMF' by utilizing ICA inverse transformation, and accumulating and restoring a plurality of new intrinsic signals to obtain a reconstructed signal;
and performing LLE embedded dimensionality reduction processing on the processed new original signal, and extracting a feature vector.
3. The electromechanical device bearing fault prediction method based on the transfer learning bayesian network according to claim 1, wherein the second step comprises:
taking the feature vector extracted in the step one as the input of a clustering neural network, adopting an SOM unsupervised self-adaptive neural network, taking an output signal as a neural network input layer neuron, and completing the weight parameter optimization of source domain data to a target domain SOM neural network through the weight sharing of a transfer learning source domain and a target domain;
and finishing the clustering task of the target domain samples through the SOM neural network target domain output layer result to obtain a clustering result, namely classifying the signals according to the mechanical fault diagnosis type.
4. The method for predicting the bearing fault of the electromechanical device based on the Bayesian network based on the transfer learning of claim 1, wherein the third step comprises:
taking the signal results of different clusters in the step two as the input of the Bayesian network, wherein the operation of the source domain is the same as that of the target domain;
by reasonably selecting source domain data samples which are basically the same as the target domain in distribution, the Bayesian network on the source domain and the target domain has the same structure, covariate shift exists between the source domain and the target domain, and a theoretical basis is established for transfer learning between the source domain and the target domain;
according to the covariate shift assumption, there are:
however, due to the problem of model specification error in the transfer learning, that is, there is no parameter that can accurately describe the relationship between the covariate x and the dependent variable y, the specific expression is as follows:
at this time, direct migration is not possible, so a weighted likelihood function is used to define a parameter loss value function for the target domain bayesian network, a loss function on the target domain is constructed by assigning a weight to the source domain data sample loss function, and an optimal parameter combination of the bayesian network on the target domain is learned, so that the expectation of the loss function of the bayesian network under the parameter is minimum, namely:
the loss function for the entire bayesian network is then:
when J = S represents a loss function on a source domain, J = T represents a loss function on a target domain, and for sample data D, a likelihood function is as follows:
wherein J = S represents the likelihood function on the source domain, J = T represents the likelihood function on the target domain,
the likelihood function of the random variable Xi on the data Dl is:
when the random variable I and the parent node value thereof appear in the sample, I =1, otherwise I = 0;
when m is
In the process, the maximum likelihood estimation on the source domain approaches to the value of a loss function minimization parameter formula on the source domain, namely the maximum likelihood estimation has the same meaning under the limit condition and is the same on the target domain,represented on the target domain as follows:
and because the assumed conditions of covariate shift are satisfied on the source domain and the target domain, the following are:
substituting the limit formula to obtain:
therefore, a mathematical relation of a minimization loss function of a source domain and a target domain is established, namely when sample data is large enough and tends to infinity, weighted maximum likelihood estimation of a source domain network tends to maximum likelihood estimation on the target domain, in original data, the weight of source domain data is PT (dt), and the weight of target domain data is Ps (Ds), so that the source domain data and the target domain data are mixed and assigned again, wherein the source domain data is assigned with the weight PT (dt)/Ps (Ds), the target domain data is assigned with the weight of 1, and a maximum likelihood estimation formula obtained by calculation is as follows:
therefore, the network analysis learning effect of the target domain under the condition of small number of samples is improved.
5. The method for predicting the bearing fault of the electromechanical device based on the Bayesian network based on the transfer learning of claim 1, wherein the fourth step comprises:
the system needs to complete fault visualization and state early warning work on the basis of an external alarm system and an internal feedback system;
according to the output result of the target domain Bayesian network, judging the four fault degrees of normal, slight decline, accelerated decline, fault and failure of the equipment;
when the state is judged to be normal, the external alarm system is dormant, and the internal feedback system is dormant
When the early warning state is judged to be a preliminary decline early warning state or a rapid decline early warning state, an external alarm system is immediately activated to report the early warning state, and early warning information comprises a pre-failure component, a pre-failure type and a pre-failure degree;
the internal feedback system judges whether parameters are automatically adjusted within the self-adaptive adjustment capability range of the machine, such as automatically reducing the rotating speed and the like;
and when the state is judged to be a complete fault state, the external alarm system is activated to enter a fault emergency alarm state.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110236878.5A CN113032916A (en) | 2021-03-03 | 2021-03-03 | Electromechanical device bearing fault prediction method based on Bayesian network of transfer learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110236878.5A CN113032916A (en) | 2021-03-03 | 2021-03-03 | Electromechanical device bearing fault prediction method based on Bayesian network of transfer learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113032916A true CN113032916A (en) | 2021-06-25 |
Family
ID=76466581
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110236878.5A Withdrawn CN113032916A (en) | 2021-03-03 | 2021-03-03 | Electromechanical device bearing fault prediction method based on Bayesian network of transfer learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113032916A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114169416A (en) * | 2021-11-30 | 2022-03-11 | 电子科技大学 | Short-term load prediction method under small sample set based on transfer learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107024352A (en) * | 2017-05-03 | 2017-08-08 | 哈尔滨理工大学 | A kind of Rolling Bearing Fault Character extracting method based on slip entropy ICA algorithm |
| CN108414226A (en) * | 2017-12-25 | 2018-08-17 | 哈尔滨理工大学 | Fault Diagnosis of Roller Bearings under the variable working condition of feature based transfer learning |
| CN110689130A (en) * | 2019-10-24 | 2020-01-14 | 陕西科技大学 | Bearing fault diagnosis method |
| CN110849627A (en) * | 2019-11-27 | 2020-02-28 | 哈尔滨理工大学 | Width migration learning network and rolling bearing fault diagnosis method based on same |
| CN111898095A (en) * | 2020-07-10 | 2020-11-06 | 佛山科学技术学院 | Deep migration learning intelligent fault diagnosis method and device, storage medium and equipment |
-
2021
- 2021-03-03 CN CN202110236878.5A patent/CN113032916A/en not_active Withdrawn
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107024352A (en) * | 2017-05-03 | 2017-08-08 | 哈尔滨理工大学 | A kind of Rolling Bearing Fault Character extracting method based on slip entropy ICA algorithm |
| CN108414226A (en) * | 2017-12-25 | 2018-08-17 | 哈尔滨理工大学 | Fault Diagnosis of Roller Bearings under the variable working condition of feature based transfer learning |
| CN110689130A (en) * | 2019-10-24 | 2020-01-14 | 陕西科技大学 | Bearing fault diagnosis method |
| CN110849627A (en) * | 2019-11-27 | 2020-02-28 | 哈尔滨理工大学 | Width migration learning network and rolling bearing fault diagnosis method based on same |
| CN111898095A (en) * | 2020-07-10 | 2020-11-06 | 佛山科学技术学院 | Deep migration learning intelligent fault diagnosis method and device, storage medium and equipment |
Non-Patent Citations (1)
| Title |
|---|
| 李子达: "贝叶斯网络参数迁移学习方法", 《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114169416A (en) * | 2021-11-30 | 2022-03-11 | 电子科技大学 | Short-term load prediction method under small sample set based on transfer learning |
| CN114169416B (en) * | 2021-11-30 | 2023-04-21 | 电子科技大学 | Short-term load prediction method based on migration learning under small sample set |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Simultaneous bearing fault recognition and remaining useful life prediction using joint-loss convolutional neural network | |
| CN110361176B (en) | Intelligent fault diagnosis method based on multitask feature sharing neural network | |
| Wu et al. | A multi-level-denoising autoencoder approach for wind turbine fault detection | |
| Chen et al. | Robust deep learning-based diagnosis of mixed faults in rotating machinery | |
| CN111579243B (en) | Rolling bearing intelligent diagnosis system based on deep migration learning | |
| Jia et al. | A deviation based assessment methodology for multiple machine health patterns classification and fault detection | |
| Filev et al. | An industrial strength novelty detection framework for autonomous equipment monitoring and diagnostics | |
| Alzghoul et al. | Increasing availability of industrial systems through data stream mining | |
| KR102383139B1 (en) | Industrial facility failure prediction diagnosis data learning labeling system and method based on artificial intelligence | |
| CN102252843A (en) | Assessment method for rolling bearing performance variation | |
| Martínez-Rego et al. | Fault detection via recurrence time statistics and one-class classification | |
| Peco Chacón et al. | State of the art of artificial intelligence applied for false alarms in wind turbines | |
| CN116484751B (en) | Fatigue life assessment method and device for wind turbine generator components | |
| CN111325410A (en) | General fault early warning system based on sample distribution and early warning method thereof | |
| CN113496061A (en) | SOM network clustering electromechanical device bearing fault analysis method based on transfer learning and manifold distance | |
| CN113032916A (en) | Electromechanical device bearing fault prediction method based on Bayesian network of transfer learning | |
| Liu et al. | A deep generative model based on CNN-CVAE for wind turbine condition monitoring | |
| Zhang et al. | A novel fault diagnosis method for wind turbine based on adaptive multivariate time-series convolutional network using SCADA data | |
| CN110378371A (en) | A kind of energy consumption method for detecting abnormality based on average nearest neighbor distance Outlier factor | |
| Colone et al. | Optimisation of data acquisition in wind turbines with data-driven conversion functions for sensor measurements | |
| Bach-Andersen et al. | Scalable systems for early fault detection in wind turbines: a data driven approach | |
| CN114486262B (en) | CNN-AT-LSTM-based bearing residual service life prediction method | |
| CN113807027B (en) | Wind turbine generator system health state evaluation model, method and system | |
| Mateus et al. | Comparing LSTM and GRU Models to Predict the Condition of a Pulp Paper Press. Energies 2021, 14, 6958 | |
| Yang et al. | Anomaly detection of automobile welding workstation based on stacked autoencoders and long short-term memory network |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210625 |
|
| WW01 | Invention patent application withdrawn after publication |