CN109813542A - The method for diagnosing faults of air-treatment unit based on production confrontation network - Google Patents
The method for diagnosing faults of air-treatment unit based on production confrontation network Download PDFInfo
- Publication number
- CN109813542A CN109813542A CN201910197936.0A CN201910197936A CN109813542A CN 109813542 A CN109813542 A CN 109813542A CN 201910197936 A CN201910197936 A CN 201910197936A CN 109813542 A CN109813542 A CN 109813542A
- Authority
- CN
- China
- Prior art keywords
- sample
- fault
- model
- data
- generates
- 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
Landscapes
- Automatic Analysis And Handling Materials Therefor (AREA)
Abstract
The invention discloses a kind of method for diagnosing faults of air-treatment unit based on production confrontation network, the following steps are included: data prediction, by to will lack and duplicate data delete, data set is normalized and Feature Selection, to accelerate the convergence rate that sample generates model, data dimension is reduced；Fault sample generates model construction, generates model by the true fault sample training fault sample sampled, learns the distribution of true fault sample；Samples Estimates model construction proposes to select using one assessment models of true fault sample training sample is generated, the degree of closeness of the distribution of fault sample and authentic specimen for training the generation of fault diagnosis model, to assess the quality for generating sample；Fault diagnosis model building, by using the sample training fault diagnosis model of above-mentioned generation, produced data are detected when running to air-treatment unit in reality, and whether diagnostic device breaks down.
Description
Technical field
The present invention relates to fault diagnosis technology fields, more specifically, being related to a kind of sky based on production confrontation network
The method for diagnosing faults of gas disposal unit.
Background technique
Fault detection and diagnosis (Fault detection and diagnosis, FDD) technology is as industrial a kind of heavy
The technology wanted plays an important role guaranteeing that industrial equipment normally, is efficiently run.By machine learning and big data skill
Art is applied to fault detection and diagnosis technology, and the staff for safeguarding industrial equipment can be helped more effectively to the state of equipment
Monitoring, to industrial equipment maintenance will also become simpler.The algorithm of machine learning is broadly divided into three kinds: having supervision
Learning algorithm, semi-supervised learning algorithm and unsupervised learning algorithm.Due to the quick hair of modern industrial technology in reality
Exhibition, many equipment are not easy failure problems frequent occurrence.Therefore, we are often difficult to obtain sufficient fault data training one
Supervised learning model is used for fault detection and diagnosis.
Confrontation network (Generative Adversarial Networks, GANs) is generated just to put forward as 2014
Unsupervised learning model, it has broad application prospects in fault detection and diagnosis field.It is calculated as a kind of machine learning
Method, he has several advantages that
1, can be learnt by the distribution to a small amount of fault data, to generate a large amount of fault sample.
2, model has only used backpropagation, without Markov Chain.
3, it does not need to do hidden variable when training and infer.
4, it can be combined well with deep learning model.
5, more preferable than traditional unsupervised learning algorithm effect.
GANs can be learnt by the distribution to a small amount of fault data, to generate the spy of a large amount of fault sample
Point is very suitable to solve to lack sufficient failure sample when industrially carrying out detection and diagnosis to the failure of large-size air conditioning system at present
This problem of.Since the time of current GANs technology proposition is later, still in starting developing stage, it is still had several drawbacks, than
It such as trains relatively difficult, generates that sample diversity is insufficient, and the sample of generation and true sample still have some gaps.
Summary of the invention
In view of this, the present invention proposes that a kind of building sample generates model, and reduces and generate between sample and authentic specimen
Gap based on production confrontation network air-treatment unit method for diagnosing faults, it is of the existing technology for solving
Fault sample is insufficient during fault detection and diagnosis, GANs training is more difficult, generates the sample that sample diversity is insufficient, generates
With the technical problems such as true sample.
The present invention provides a kind of method for diagnosing faults of air-treatment unit based on production confrontation network, including with
Lower step:
Data prediction, by the way that that will lack and duplicate data are deleted, data set is normalized and feature
It chooses, to accelerate the convergence rate that sample generates model, reduces data dimension；
Fault sample generates model construction, generates model by the true fault sample training fault sample sampled, learns
Practise the distribution of true fault sample；
Samples Estimates model construction proposes to choose using one assessment models of true fault sample training to sample is generated
Choosing, the degree of closeness of the distribution of fault sample and authentic specimen for training the generation of fault diagnosis model are generated with assessing
The quality of sample；
Fault diagnosis model building, by using the sample training fault diagnosis model of above-mentioned generation, to air in reality
Produced data are detected when processing unit operation, and whether diagnostic device breaks down.
Optionally, the Feature Selection refers to, selects using before the sequence of the cost-sensitive based on SVM classifier to feature
It selects algorithm and carries out Feature Selection, since a scheduled character subset, until selecting most important feature.
Optionally, a large amount of fault samples are generated for dividing the data of air-treatment unit in fault diagnosis model
Class, being realized using support vector machines makes same failure point in one kind.
Optionally, after Samples Estimates model evaluation, Samples Estimates model is added in the high generation sample of the quality picked out
Training set assessment models are further trained, be further continued for generate sample select.
Optionally, carry out the normalized to data set to refer to, be standardized using=deviation, to initial data into
Row linear transformation is mapped to end value between [0,1] using following formula；
Wherein, Min=0, Max=1.
Optionally, it is that condition Wasserstein generates confrontation network model that the fault sample, which generates model, pass through by
Wasserstein generates confrontation network and conditional generates confrontation network integration set up the condition Wasserstein and generates confrontation net
Network model, including generator and arbiter；
The training set of confrontation network model is generated using the data pre-processed as condition Wasserstein, if x and z points
It is not the distribution P from training setdata(x) and a priori noise profile Pz(z) authentic specimen and noise sampled out in；In order to
Learn Pdata(x) distribution, by priori noise profile Pz(z) a mapping space G (z is constructed；θg)；Corresponding arbiter
Mapping function is D (x；θd), output x is truthful data probability, and objective function characterization obtains the generation sample closest to authentic specimen
This, is shown below；
Wherein, it is expected thatIn, D (x | c) indicate that the sample x of arbiter judgement input is under conditions of c
The probability of authentic specimen；It is expectedMiddle x indicates that the sample that generation network generates, D (xc) indicate that arbiter is sentenced
Medium well is the probability of authentic specimen at sample,For penalty term, λ is penalty factor, generator
Final goal be to allow the sample infinite approach authentic specimen of generation；And the purpose of arbiter be as far as possible by authentic specimen and
Sample is generated to distinguish.
Using the present invention, compared with prior art, has the advantage that invention generates model using fault sample, generate
A large amount of fault sample data, and training sample assessment models select mass high generation sample for training fault diagnosis
Model realizes and trains the higher model of fault detection and diagnosis accuracy rate in the case where fault sample is less.
Detailed description of the invention
Fig. 1 is the flow diagram that the method for diagnosing faults of air-treatment unit of network is fought based on production；
Fig. 2 is the flow diagram that sample generates model；
Fig. 3 is the flow diagram of Samples Estimates model；
Fig. 4 is the flow diagram of the feature selecting of cost-sensitive.
Specific embodiment
The preferred embodiment of the present invention is described in detail below in conjunction with attached drawing, but the present invention is not restricted to these
Embodiment.The present invention covers any substitution made in the spirit and scope of the present invention, modification, equivalent method and scheme.
In order to make the public have thorough understanding to the present invention, it is described in detail in the following preferred embodiment of the present invention specific
Details, and the present invention can also be understood completely in description without these details for a person skilled in the art.
The present invention is more specifically described by way of example referring to attached drawing in the following passage.It should be noted that attached drawing is adopted
With more simplified form and using non-accurate ratio, only to facilitate, lucidly aid in illustrating the embodiment of the present invention
Purpose.
The present invention provides a kind of method for diagnosing faults of air-treatment unit based on production confrontation network, including following
Step:
Data prediction, by the way that that will lack and duplicate data are deleted, data set is normalized and feature
It chooses, to accelerate the convergence rate that sample generates model, reduces data dimension；
Fault sample generates model construction, generates model by the true fault sample training fault sample sampled, learns
Practise the distribution of true fault sample；
Samples Estimates model construction proposes to choose using one assessment models of true fault sample training to sample is generated
Choosing, the degree of closeness of the distribution of fault sample and authentic specimen for training the generation of fault diagnosis model are generated with assessing
The quality of sample；
Fault diagnosis model building, by using the sample training fault diagnosis model of above-mentioned generation, to air in reality
Produced data are detected when processing unit operation, and whether diagnostic device breaks down.
Optionally, the Feature Selection refers to, selects using before the sequence of the cost-sensitive based on SVM classifier to feature
It selects algorithm and carries out Feature Selection, since a scheduled character subset, until selecting most important feature.
Optionally, a large amount of fault samples are generated for dividing the data of air-treatment unit in fault diagnosis model
Class, being realized using support vector machines makes same failure point in one kind.
Optionally, after Samples Estimates model evaluation, Samples Estimates model is added in the high generation sample of the quality picked out
Training set assessment models are further trained, be further continued for generate sample select.
Optionally, carry out the normalized to data set to refer to, be standardized using=deviation, to initial data into
Row linear transformation is mapped to end value between [0,1] using following formula；
Wherein, Min=0, Max=1.
Optionally, it is that condition Wasserstein generates confrontation network model that the fault sample, which generates model, pass through by
Wasserstein generates confrontation network and conditional generates confrontation network integration set up the condition Wasserstein and generates confrontation net
Network model, including generator and arbiter；
The training set of confrontation network model is generated using the data pre-processed as condition Wasserstein, if x and z points
It is not the distribution P from training setdata(x) and a priori noise profile Pz(z) authentic specimen and noise sampled out in；In order to
Learn Pdata(x) distribution, by priori noise profile Pz(z) a mapping space G (z is constructed；θg)；Corresponding arbiter
Mapping function is D (x；θd), output x is truthful data probability, and objective function characterization obtains the generation sample closest to authentic specimen
This, is shown below；
Wherein, it is expected thatIn, D (x | c) indicate that the sample x of arbiter judgement input is under conditions of c
The probability of authentic specimen；It is expectedMiddle x indicates that the sample that generation network generates, D (xc) indicate that arbiter is sentenced
Medium well is the probability of authentic specimen at sample,For penalty term, λ is penalty factor, generator
Final goal be to allow the sample infinite approach authentic specimen of generation；And the purpose of arbiter be as far as possible by authentic specimen and
Sample is generated to distinguish.
The generation Samples Estimates model construction: generation sample is assessed by using a small amount of authentic specimen Training valuation model
Originally the higher sample of mass is selected, then, then the training set for generating sample and assessment models being added that the quality that will be singled out is high
Assessment models are further trained, are further continued for selecting generation sample, detailed process is as shown in Figure 3.
Implementation process of the invention can be divided into following steps progress:
1, sample collection.It has been downloaded from existing database as provided by ASHRAE project NO.1312-RP
Data.The quantity of normal sample is 21600 in the data set, and the sample number of every kind of failure is 1440 totally 17 class data.This
In pick wherein in relatively common 6 failure as experimental data.
2, data prediction.The data that will acquire are taken out missing values and repeated sample, and place is then normalized
Reason, to accelerate the training speed of model, makes network be more easier to restrain.
3, feature selecting.Feature selecting is carrying out an important step in machine learning task, it can reduce number
According to dimension, the operational efficiency of program is improved as far as possible in the case where guaranteeing machine learning model performance.In original data set
It is inappropriate certainly with data one model of training of such higher-dimension containing more than 140 a features.Therefore, by using being based on
It is selected before the sequence of the cost-sensitive of SVM classifier to feature of the feature selecting algorithm to original data set, as shown in Figure 4.
4, fault sample is generated.A large amount of failure is generated by using above-mentioned proposed C-WGANs model in this
Sample, the training sample of C-WGANs model use sample good handled by previous step.
5, assessment models construct.The training set of assessment models equally uses handled good sample in (3) step.So
Afterwards, then the high generation sample of the quality that will be singled out is added the training sets of assessment models and further trains to assessment models, followed by
It is continuous that generation sample is selected.
6, it picks out and generates sample similar in authentic specimen.Using above-mentioned proposed Samples Estimates model to generation sample
This progress is assessed one by one, selects that quality is higher similar with authentic specimen to generate sample.
7, fault diagnosis model.We will construct a support vector machines (Support Vector in this step
Machine, SVM) disaggregated model, the generation sample that previous step is picked out is trained into fault detection and diagnosis mould as training set
Type.Here using supervised learning model, it can choose SVM, random depth woods or K- arest neighbors (k-
NearestNeighbor, KNN) algorithm.
Although embodiment is separately illustrated and is illustrated above, it is related to the common technology in part, in ordinary skill
Personnel apparently, can be replaced and integrate between the embodiments, be related to one of embodiment and the content recorded is not known, then
It can refer to another embodiment on the books.
Embodiments described above does not constitute the restriction to the technical solution protection scope.It is any in above-mentioned implementation
Made modifications, equivalent substitutions and improvements etc., should be included in the protection model of the technical solution within the spirit and principle of mode
Within enclosing.
Claims (6)
1. a kind of method for diagnosing faults of the air-treatment unit based on production confrontation network, it is characterised in that: including following
Step:
Data prediction, by the way that that will lack and duplicate data are deleted, data set is normalized and Feature Selection,
To accelerate the convergence rate that sample generates model, data dimension is reduced；
Fault sample generates model construction, generates model by the true fault sample training fault sample sampled, study is true
The distribution of real fault sample；
Samples Estimates model construction proposes to select using one assessment models of true fault sample training sample is generated,
For training the degree of closeness of the distribution of the fault sample and authentic specimen of the generation of fault diagnosis model, to assess generation sample
Quality；
Fault diagnosis model building, by using the sample training fault diagnosis model of above-mentioned generation, to air-treatment in reality
Produced data are detected when unit operation, and whether diagnostic device breaks down.
2. the method for diagnosing faults of the air-treatment unit according to claim 1 based on production confrontation network, special
Sign is: the Feature Selection refers to, using before the sequence of the cost-sensitive based on SVM classifier to feature selecting algorithm into
Row Feature Selection, since a scheduled character subset, until selecting most important feature.
3. the method for diagnosing faults of the air-treatment unit according to claim 1 or 2 based on production confrontation network,
It is characterized in that: generating a large amount of fault samples for classifying in fault diagnosis model to the data of air-treatment unit, use
Support vector machines makes same failure point in one kind to realize.
4. the method for diagnosing faults of the air-treatment unit according to claim 1 based on production confrontation network, special
Sign is: after Samples Estimates model evaluation, the high training set for generating sample and Samples Estimates model being added of the quality picked out
Assessment models are further trained, are further continued for selecting generation sample.
5. the fault diagnosis of -4 air-treatment unit based on production confrontation network described in any one according to claim 1
Method, it is characterised in that: the normalized is carried out to data set and is referred to, using=deviation standardization, to initial data
Linear transformation is carried out, is mapped to end value between [0,1] using following formula；
Wherein, Min=0, Max=1.
6. the fault diagnosis of -4 air-treatment unit based on production confrontation network described in any one according to claim 1
Method, it is characterised in that: it is that condition Wasserstein generates confrontation network model that the fault sample, which generates model, pass through by
Wasserstein generates confrontation network and conditional generates confrontation network integration set up the condition Wasserstein and generates confrontation net
Network model, including generator and arbiter；
The training set that confrontation network model is generated using the data pre-processed as condition Wasserstein, if x and z are respectively
From the distribution P of training setdata(x) and a priori noise profile Pz(z) authentic specimen and noise sampled out in；In order to learn
Pdata(x) distribution, by priori noise profile Pz(z) a mapping space G (z is constructed；θg)；The mapping of corresponding arbiter
Function is D (x；θd), output x is truthful data probability, and objective function characterization obtains the generation sample closest to authentic specimen, such as
Shown in following formula；
Wherein, it is expected thatIn, D (x | c) indicate that the sample x of arbiter judgement input is true under conditions of c
The probability of sample；It is expectedThe sample that middle x expression generation network generates, and D (x | c) indicate that arbiter judgement is given birth to
It is the probability of authentic specimen at sample,For penalty term, λ is penalty factor, and generator is most
Whole target is to allow the sample infinite approach authentic specimen of generation；And the purpose of arbiter is as far as possible by authentic specimen and generation
Sample distinguishes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910197936.0A CN109813542A (en) | 2019-03-15 | 2019-03-15 | The method for diagnosing faults of air-treatment unit based on production confrontation network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910197936.0A CN109813542A (en) | 2019-03-15 | 2019-03-15 | The method for diagnosing faults of air-treatment unit based on production confrontation network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109813542A true CN109813542A (en) | 2019-05-28 |
Family
ID=66609098
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910197936.0A Pending CN109813542A (en) | 2019-03-15 | 2019-03-15 | The method for diagnosing faults of air-treatment unit based on production confrontation network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109813542A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110542819A (en) * | 2019-09-25 | 2019-12-06 | 贵州电网有限责任公司 | transformer fault type diagnosis method based on semi-supervised DBNC |
| CN110567720A (en) * | 2019-08-07 | 2019-12-13 | 东北电力大学 | method for diagnosing depth confrontation of fault of fan bearing under unbalanced small sample scene |
| CN111310791A (en) * | 2020-01-17 | 2020-06-19 | 电子科技大学 | Dynamic progressive automatic target identification method based on small sample number set |
| CN111445147A (en) * | 2020-03-27 | 2020-07-24 | 中北大学 | Generative confrontation network model evaluation method for mechanical fault diagnosis |
| CN111723851A (en) * | 2020-05-30 | 2020-09-29 | 同济大学 | Production line fault detection method |
| CN112966429A (en) * | 2020-08-11 | 2021-06-15 | 中国矿业大学 | Non-linear industrial process modeling method based on WGANs data enhancement |
| CN113505876A (en) * | 2021-06-11 | 2021-10-15 | 国网浙江省电力有限公司嘉兴供电公司 | High-voltage circuit breaker fault diagnosis method based on generation type countermeasure network |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107563510A (en) * | 2017-08-14 | 2018-01-09 | 华南理工大学 | A kind of WGAN model methods based on depth convolutional neural networks |
| AU2018100325A4 (en) * | 2018-03-15 | 2018-04-26 | Nian, Xilai MR | A New Method For Fast Images And Videos Coloring By Using Conditional Generative Adversarial Networks |
| CN108680807A (en) * | 2018-05-17 | 2018-10-19 | 国网山东省电力公司青岛供电公司 | The Diagnosis Method of Transformer Faults and system of network are fought based on condition production |
| CN108711138A (en) * | 2018-06-06 | 2018-10-26 | 北京印刷学院 | A kind of gray scale picture colorization method based on generation confrontation network |
| CN108763857A (en) * | 2018-05-29 | 2018-11-06 | 浙江工业大学 | A kind of process soft-measuring modeling method generating confrontation network based on similarity |
| CN109086658A (en) * | 2018-06-08 | 2018-12-25 | 中国科学院计算技术研究所 | A kind of sensing data generation method and system based on generation confrontation network |
-
2019
- 2019-03-15 CN CN201910197936.0A patent/CN109813542A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107563510A (en) * | 2017-08-14 | 2018-01-09 | 华南理工大学 | A kind of WGAN model methods based on depth convolutional neural networks |
| AU2018100325A4 (en) * | 2018-03-15 | 2018-04-26 | Nian, Xilai MR | A New Method For Fast Images And Videos Coloring By Using Conditional Generative Adversarial Networks |
| CN108680807A (en) * | 2018-05-17 | 2018-10-19 | 国网山东省电力公司青岛供电公司 | The Diagnosis Method of Transformer Faults and system of network are fought based on condition production |
| CN108763857A (en) * | 2018-05-29 | 2018-11-06 | 浙江工业大学 | A kind of process soft-measuring modeling method generating confrontation network based on similarity |
| CN108711138A (en) * | 2018-06-06 | 2018-10-26 | 北京印刷学院 | A kind of gray scale picture colorization method based on generation confrontation network |
| CN109086658A (en) * | 2018-06-08 | 2018-12-25 | 中国科学院计算技术研究所 | A kind of sensing data generation method and system based on generation confrontation network |
Non-Patent Citations (3)
| Title |
|---|
| CHAOWEN ZHONG ET AL: "Energy Efficiency Solutions for Buildings: Automated Fault Diagnosis of Air Handling Units Using Generative Adversarial Networks", 《ENERGIES》 * |
| SHARAD RAWAT ET AL: "A Novel Topology Optimization Approach using Conditional Deep Learning", 《MACHINE LEARNING》 * |
| 钟超文等: "半监督支持向量机的空气处理机组夏季故障诊断", 《中国计量大学学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110567720A (en) * | 2019-08-07 | 2019-12-13 | 东北电力大学 | method for diagnosing depth confrontation of fault of fan bearing under unbalanced small sample scene |
| CN110542819A (en) * | 2019-09-25 | 2019-12-06 | 贵州电网有限责任公司 | transformer fault type diagnosis method based on semi-supervised DBNC |
| CN110542819B (en) * | 2019-09-25 | 2022-03-22 | 贵州电网有限责任公司 | Transformer fault type diagnosis method based on semi-supervised DBNC |
| CN111310791A (en) * | 2020-01-17 | 2020-06-19 | 电子科技大学 | Dynamic progressive automatic target identification method based on small sample number set |
| CN111445147A (en) * | 2020-03-27 | 2020-07-24 | 中北大学 | Generative confrontation network model evaluation method for mechanical fault diagnosis |
| CN111723851A (en) * | 2020-05-30 | 2020-09-29 | 同济大学 | Production line fault detection method |
| CN112966429A (en) * | 2020-08-11 | 2021-06-15 | 中国矿业大学 | Non-linear industrial process modeling method based on WGANs data enhancement |
| CN113505876A (en) * | 2021-06-11 | 2021-10-15 | 国网浙江省电力有限公司嘉兴供电公司 | High-voltage circuit breaker fault diagnosis method based on generation type countermeasure network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109813542A (en) | The method for diagnosing faults of air-treatment unit based on production confrontation network | |
| CN110361176B (en) | Intelligent fault diagnosis method based on multitask feature sharing neural network | |
| CN109000930B (en) | Turbine engine performance degradation evaluation method based on stacking denoising autoencoder | |
| CN112989710B (en) | Industrial control sensor numerical value abnormality detection method and device | |
| CN109800875A (en) | Chemical industry fault detection method based on particle group optimizing and noise reduction sparse coding machine | |
| CN109829236A (en) | A kind of Compressor Fault Diagnosis method based on XGBoost feature extraction | |
| CN111241673B (en) | Health state prediction method for industrial equipment in noisy environment | |
| CN108647707B (en) | Probabilistic neural network creation method, failure diagnosis method and apparatus, and storage medium | |
| Gai et al. | A Parameter‐Optimized DBN Using GOA and Its Application in Fault Diagnosis of Gearbox | |
| CN108182445A (en) | Procedure fault recognition methods based on big data intelligence core independent component analysis | |
| CN114553475A (en) | Network attack detection method based on network flow attribute directed topology | |
| CN111126820A (en) | Electricity stealing prevention method and system | |
| CN113537277A (en) | Determining interpretation of a classification | |
| CN112784920A (en) | Cloud-side-end-coordinated dual-anti-domain self-adaptive fault diagnosis method for rotating part | |
| CN115345222A (en) | Fault classification method based on TimeGAN model | |
| Ruderman et al. | Uncovering surprising behaviors in reinforcement learning via worst-case analysis | |
| CN117056678B (en) | Machine pump equipment operation fault diagnosis method and device based on small sample | |
| CN113343123A (en) | Training method and detection method for generating confrontation multiple relation graph network | |
| CN115356599B (en) | Multi-mode urban power grid fault diagnosis method and system | |
| CN112346956A (en) | Test method for neuron variation aiming at deep learning program | |
| CN115931359A (en) | Turbine pump bearing fault diagnosis method and device | |
| Wang et al. | Match-reinforcement learning with time frequency selection for bearing fault diagnosis | |
| CN115392434A (en) | Depth model reinforcement method based on graph structure variation test | |
| CN114565051A (en) | Test method of product classification model based on neuron influence degree | |
| CN115314287A (en) | Confrontation anomaly detection system based on deep clustering |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190528 |