CN114817856B - Beam-pumping unit fault diagnosis method based on structural information retention domain adaptation network - Google Patents
Beam-pumping unit fault diagnosis method based on structural information retention domain adaptation network Download PDFInfo
- Publication number
- CN114817856B CN114817856B CN202210397562.9A CN202210397562A CN114817856B CN 114817856 B CN114817856 B CN 114817856B CN 202210397562 A CN202210397562 A CN 202210397562A CN 114817856 B CN114817856 B CN 114817856B
- Authority
- CN
- China
- Prior art keywords
- domain
- target domain
- pumping unit
- source domain
- membership matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000006978 adaptation Effects 0.000 title claims abstract description 32
- 238000005086 pumping Methods 0.000 title claims abstract description 24
- 238000003745 diagnosis Methods 0.000 title claims abstract description 23
- 230000014759 maintenance of location Effects 0.000 title claims abstract description 10
- 238000012549 training Methods 0.000 claims abstract description 20
- 238000010586 diagram Methods 0.000 claims abstract description 16
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 14
- 238000004364 calculation method Methods 0.000 claims abstract description 7
- 238000005259 measurement Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 34
- 238000006073 displacement reaction Methods 0.000 claims description 10
- 230000003044 adaptive effect Effects 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 7
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 239000003208 petroleum Substances 0.000 abstract description 3
- 238000012360 testing method Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 238000011084 recovery Methods 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012800 visualization Methods 0.000 description 2
- 235000014653 Carica parviflora Nutrition 0.000 description 1
- 241000243321 Cnidaria Species 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000003129 oil well Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000000243 solution Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Pure & Applied Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Mathematical Physics (AREA)
- Evolutionary Biology (AREA)
- Mathematical Optimization (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Computation (AREA)
- Operations Research (AREA)
- Artificial Intelligence (AREA)
- Algebra (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention relates to the technical field of petroleum industry, in particular to a pumping unit fault diagnosis method based on a structural information retention domain adaptation network, which comprises the steps of collecting pumping unit indicator diagram data to obtain a source domain data set and a target domain data set, extracting source domain features and target domain features by using a feature generator, performing supervised training by using source domain features and source domain sample labels in the source domain to update the feature generator and a classifier, clustering the target domain features by using an unsupervised clustering algorithm to obtain target domain pseudo-labels, inputting the source domain features and the source domain labels and the target domain features and the target domain pseudo-labels into a local maximum mean difference measurement formula, and performing distribution difference calculation to optimize a feature generator. The structural information holding domain adaptation network provided by the invention can be used for fault diagnosis of the oil pumping unit, can reduce the excessive requirement of fault diagnosis model training on data annotation, and can improve the generalization performance of the fault diagnosis model in different application scenes.
Description
Technical Field
The invention relates to the technical field of petroleum industry, in particular to a pumping unit fault diagnosis method based on a structural information retention domain adaptation network.
Background
The energy problem is a great problem related to the stable society and the sustainable development of national economy and health, and petroleum plays an important role in the energy structure. Oil recovery can be divided into two modes according to different oil recovery modes: mechanical oil recovery and self-injection oil recovery. The most common use in China is mechanical oil extraction. The pumping unit works in thousands of meters underground throughout the year, and has the problems of severe environment, complex working condition, abrasion, corrosion, mechanical fatigue, functional failure and the like, thereby bringing hidden trouble and uncertainty to the pumping unit.
The current fault diagnosis of the pumping unit based on the artificial intelligence algorithm obtains good performance in simulation experiments. Machine learning algorithms (e.g., artificial Neural Networks (ANNs), support Vector Machines (SVMs), etc.) have difficulty in processing dimensions and dynamically monitoring data. The deep learning algorithm can solve more complex problems and is widely applied to fault diagnosis. Deep learning algorithms such as Deep Belief Networks (DBNs), sparse Automatic Encoders (SAE), convolutional Neural Networks (CNNs) and the like have outstanding performance in terms of fault diagnosis.
However, most of the current researches stay in the experimental simulation stage, and have certain limitations, and most of the researches do not consider the problems of insufficient model generalization capability and reduced recognition accuracy caused by different model application scenes.
Disclosure of Invention
The invention aims to provide a pumping unit fault diagnosis method based on a structural information retention domain adaptation network, and aims to solve the problems that most of current researches do not consider insufficient model generalization capability and low recognition accuracy caused by different model application scenes.
In order to achieve the above purpose, the invention provides a pumping unit fault diagnosis method based on a structural information retention domain adaptation network, which comprises the steps of collecting pumping unit indicator diagram data to obtain a source domain data set and a target domain data set;
Extracting source domain features and target domain features using a feature generator;
performing supervised training in a source domain by using source domain features and source domain sample labels to update a feature generator and a classifier;
clustering the target domain features by using an unsupervised clustering algorithm to obtain target domain pseudo tags;
and (3) inputting the source domain characteristics and the source domain labels into local maximum mean differences of the target domain characteristics and the target domain pseudo labels to perform distribution difference calculation to optimize the characteristic generator.
Wherein the indicator diagram refers to a closed graph formed by discrete points of displacement and load.
The feature generator is a feature extraction model formed by a convolutional neural network.
The specific steps of obtaining the target domain pseudo tag comprise:
randomly initializing a membership matrix according to the data set;
Calculating an optimal membership matrix and a clustering center;
And obtaining the pseudo tag of the target domain by using the membership matrix and the clustering center.
The specific steps of calculating the optimal membership matrix and the clustering center comprise:
updating the clustering center by using the membership matrix;
Calculating the Euclidean distance from the sample to the clustering center by using the clustering center;
updating the membership matrix by using the cluster number, the cluster center distance and the fuzzy weight index;
Judging whether an iteration stop condition is met according to whether the maximum iteration times and the errors are reached;
Obtaining an optimal membership matrix and a clustering center of an algorithm;
the specific steps of obtaining the pseudo tag of the target domain are as follows:
obtaining a sample clustering label according to the membership matrix;
Aiming at the samples under each cluster label, a classifier is utilized to obtain a classification label;
Counting the number of each classification label;
selecting the label with the largest number of classified labels as the label of the clustering class, and updating the clustering center;
updating the membership matrix by using a new cluster center;
And obtaining the target domain pseudo tag according to the new membership matrix.
The pumping unit fault diagnosis method based on the structural information retention domain adaptation network can be used for fault diagnosis of the pumping unit, can reduce excessive requirements of fault diagnosis model training on data annotation, and can improve generalization performance of the fault diagnosis model in different application scenes.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flow chart of a pumping unit fault diagnosis method based on a structural information holding domain adaptive network.
Fig. 2 is a flowchart of specific steps of target domain pseudo tag acquisition.
FIG. 3 is a confusion matrix for the results of the task D1→D2 test set.
Fig. 4 is a graph of the loss of SIP-DAN during task d1→d2 training.
FIG. 5 is a graph of target domain test accuracy during training of various methods.
FIG. 6 is a t-SNE plot of the results of the task D1→D2 test set.
Detailed Description
Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present invention and should not be construed as limiting the invention.
The specific embodiment of the invention is as follows:
referring to the drawings, the invention provides a pumping unit fault diagnosis method based on a structural information holding domain adaptation network (Structure Information PreservingDomainAdaptationNetwork, SIP-DAN), which comprises the following steps:
s101, acquiring pumping unit indicator diagram data to obtain a source domain data set and a target domain data set;
And collecting the indicator diagram data of the oil pumping unit, and dividing the indicator diagram data into a source domain data set and a target domain data set.
To avoid the effect of size differences between the data on the calculation, the raw data of the indicator diagram is normalized. Each indicator diagram consists of a set of displacement x and load y discrete points { (x i,yi) }.
The displacement normalization formula is as follows:
Wherein M is the number of samples displaced; n is the number of displacement points in one displacement sample;
x ik is the kth sample value of the ith displacement sample before normalization;
a kth sample value that is the normalized ith displacement sample;
xi,min=min{xik|1≤k≤N};
xi,max=max{xik|1≤k≤N}。
the load normalization formula is as follows:
M is the number of samples of the load; n is the number of load points in one load sample;
y ik is the kth sample value of the ith payload sample before normalization;
A kth sample value that is the normalized ith load sample;
yi,min=min{yik|1≤k≤N};
yi,max=max{yik|1≤k≤N}。
And drawing an indicator diagram by using the normalized load and displacement, and drawing an indicator diagram by using the displacement as an abscissa and the load as an ordinate, and storing the indicator diagram as a picture sample.
S102, extracting a source domain feature F (x s) and a target domain feature F (x t) by using a feature generator;
The feature generator is a convolutional neural network.
S103, performing supervised training in a source domain by using source domain features and source domain sample labels to update a feature generator and a classifier;
The optimization objective of the fault classifier is to minimize classification errors under supervision of the source domain data labels. In this way, identifiable fault signatures can be extracted. The output of the feature extractor is denoted by F (-) and the output of the classifier is denoted by C (-). Classification loss uses a cross entropy loss function:
Wherein y s represents one-hot form of the source domain sample tag; is the output of the classifier in the source domain,
S104, clustering the target domain features by using an unsupervised clustering algorithm to obtain a target domain pseudo tag;
The features F (x t) of the target domain are clustered using a Fuzzy C-Means clustering algorithm (FCM) to obtain cluster labels, and then a classifier is used to assist in the alignment of the cluster labels. The method comprises the following steps:
S401 randomly initializes a membership matrix u= [ U ij]N×M.
Where N represents the number of samples, M represents the number of clusters, u ij represents the membership degree of the ith (i=1, 2, …, N) sample belonging to the jth (j=1, 2, …, M) class, satisfying the following relationship:
S402, calculating the optimal membership matrix and clustering center by using the membership matrix
(1) Calculating a clustering center:
Wherein c j denotes the jth cluster center, N denotes the number of samples, f i denotes the ith sample, and m is a fuzzy weight index.
(2) The Euclidean distance from the ith sample f i to the jth cluster center c j is calculated:
(3) Updating the membership matrix u ij:
Wherein M is the cluster number, d ij is the distance from the ith sample to the jth cluster center, d ik is the distance from the ith sample to the kth cluster center, and M is the fuzzy weight index.
(4) Judging whether any of the following iteration stop conditions is satisfied
Condition 1: judging whether the maximum iteration number max_iter is reached
Condition 2: determination of error
Wherein,For the membership matrix of the current iteration,/>And epsilon is an error threshold value for the membership matrix of the last iteration.
And if one of the conditions 1 and 2 is satisfied, the iteration can be stopped.
(5) And (4) repeating the steps (2) to (4) when the iteration stop condition of the step (4) is not satisfied. Stopping iteration when the iteration stopping condition of (4) is satisfied,
Obtaining the optimal membership matrix U and the clustering center C of the algorithm.
S403, obtaining pseudo labels of target domains by using membership matrix and clustering center
(1) And obtaining the clustering category of each sample according to the membership matrix U.
(2) The samples of the ith (i=1, 2, …, k) cluster category are input into the classifier C to obtain the classification label of the classifier.
(3) The number counter of each category in the category label is counted.
(4) Updating the cluster center of the current cluster category as follows: c new i =c (max). Repeating the steps (1) to (3) so that each cluster center is updated to obtain a new cluster center C new, and updating the membership matrix U according to the formula (3) in the step S402 by using the updated cluster center C new to obtain a new membership matrix U new.
Obtaining pseudo tags for a target domainS105 optimizes the feature generator by inputting the source domain features and source domain labels, and the target domain features and target domain pseudo labels into Local maximum mean differences (Local maximum MEAN DISCREPANCY, LMMD) for distribution difference calculation.
Obtaining target domain pseudo tagsThereafter, the source domain feature F (x s) and tag y s, and the target domain feature F (x t) and pseudo tag/>Input LMMD performs a distribution difference calculation.
LMMD the basic idea is to divide the two domains into multiple subfields by class labels and then fine-grained domain adaptation of the relevant subfields. Assume that each sample belongs to each class with a weight of w c. LMMD is defined as follows:
Wherein H is the Regenerated Kernel Hilbert Space (RKHS) for a given feature kernel k; phi (·) represents the nonlinear mapping from the original feature space to RKHS; c is the number of fault categories; The weight of the ith sample in the source domain belonging to subclass c; /(I) The j-th sample in the target domain is given the weight of subclass c.
Since φ (-) cannot be directly calculated, the formula expands as follows:
the method for calculating the weight w c is as follows:
Where y ic is the probability that the ith sample belongs to class c and n is the number of samples. For the source domain samples, the one-hot version of y s is used directly; for the target domain, the sample pseudo tag y t obtained in step (5) is used.
In this embodiment:
(1) Data set
Three data sets (named D1, D2, D3) are composed according to the data sources, each data set including normal conditions and 6 types of faults, numbered: normal, fault#1, fault#2, fault#3, fault#4, fault#5, fault#6. The specific situation of the data set is shown in table 1, and most pumping units are in normal operation, so that the collection of fault samples is difficult. In the data set obtained by analysis, normal samples are very many, but samples of fault types are fewer (such as airlocks), and the number of samples of each working condition is limited to 500 in order to reduce the influence of data unbalance.
Table 1 dataset sample case
Typical indicator diagram shapes of the data sets are shown in table 2, and the shapes of the same working condition under different data sets are different due to inconsistent acquisition conditions of original data of the different data sets. The condition factors may be different machine models, different sensor models, different oil well geology, etc., which may cause a distribution difference between the data sets, so that the classification capability is reduced when the model trained on one data set is applied to another data set for testing, and the model does not have good generalization performance.
TABLE 2 typical indicator diagram shape for each dataset
(2) Experimental setup
The domain adaptation tasks (d1→d2, d1→d3, d2→d1, d2→d3, D3→d1, D3→d2) of 6 source domain to target domain are set on three data sets of D1, D2 and D3, and the specific cases of the data samples are shown in table 3. As represented by D1→D2: d1 is a tagged source domain and D2 is an untagged destination domain.
Table 3 field adaptive task settings
In order to verify the validity of the SIP-DAN method, several supervised learning algorithms and domain adaptation methods and proposed methods are compared and analyzed, including:
and (3) SVM: support vector machines, the classical method of machine learning, are widely used in various fields.
ResNet 18A 18: a typical depth residual convolutional neural network serves as a benchmark for evaluating the domain adaptation task feature extraction capability. Because ResNet is a supervised learning network, training is performed using only labeled source domain data, and not unlabeled target domain data.
DAN: a depth domain adaptive method based on distance measurement uses the maximum mean difference (Maximummean discrepancy, MMD) as a distribution difference metric, which is a global domain adaptive method.
D-CORAL: a depth domain adaptive method based on distance measurement uses CORAL as a distribution difference metric. The method is a global domain adaptation method by performing linear transformation on a source domain and a target domain to align second-order statistical information of the source domain and the target domain.
For the SVM, the picture reading dimension is 100 x 100, and the direct leveling dimension is 10000, so that the picture is used as the input of the SVM. Experiments were performed using SVM under sklearn library, parameter settings: c=1.0, γ=auto, kenel =rbf.
For SIP-DAN, the balance parameters are dynamically set, and the calculation formula is as follows: λ=2/(1+exp (-10 x ep/epoch)) -1 (where ep is the current iteration step number and epoch is the total number of iterations).
All the above depth model methods use ResNet as the basic feature extraction network, and the general experimental parameters are set as follows:
Table 4 experimental parameter settings
(3) Comparative analysis experiment of fault diagnosis performance
The methods of SIP-DAN, etc. were implemented on 6 domain adaptation tasks, and table 5 summarizes the average diagnostic accuracy of all comparison methods.
Firstly, the comparison of the traditional machine learning method SVM and the depth method ResNet can see that the accuracy of the cross-domain fault diagnosis of the SVM is very low because the traditional method does not specifically design the feature extraction method of the indicator diagram. The depth method ResNet is subjected to end-to-end self-adaptive feature extraction and training, and the final accuracy is far higher than that of the SVM, so that the superiority of the depth method is demonstrated.
DAN and D-CORAL are representative methods of global domain adaptation, the average accuracy rate for six tasks is 86.96% and 88.33%, respectively, and the accuracy rate of the global domain adaptation method is improved by 1.57% and 3.17% compared with ResNet without domain adaptation, and the accuracy rate of the global domain adaptation method is found to be limited, because the global domain adaptation method ignores the importance of category information in the domain adaptation process, so that obvious class discrimination features are not extracted.
The SIP-DAN is fine-grained domain adaptation provided by the invention, the category information in the domain adaptation process is considered, and meanwhile, the maintenance of the target domain characteristic information is also considered, so that better accuracy is obtained. The optimal accuracy is obtained on 4 tasks in 6 tasks, the average accuracy is improved by 10.59% relative to ResNet, and a better source domain knowledge migration effect is achieved.
TABLE 5 results of various methods
(4) Confusion matrix contrast
Taking task d1→d2 as an example, the confusion matrix result of the target domain test set is shown in fig. 3. From the result of ResNet18, the distribution difference of the D1-D2 tasks is mainly reflected in categories Normal and Falut #2, resNet misidentifies 55% of Normal samples as Fault#1, and 13% of Normal samples as Fault#6; resNet18 misrecognizes 32% of the Fault#2 samples as Fault#1 and 9% of the Fault#2 samples as Fault#3. The misrecognition is not improved after DAN domain adaptation, and the Normal recognition is improved after D-CORAL domain adaptation, but the recognition effect on Fault#2 is poorer. As shown in (d), the SIP-DAN method greatly improves the fault diagnosis capability of the model.
(5) Algorithm convergence analysis experiment
Taking task d1→d2 as an example, the change of the loss value along with the training iteration step number in the SIP-DAN training process is shown in fig. 4. It can be found that the classification loss converges very rapidly at the beginning of training, and the initial domain adaptation balance parameter λ is small, and the domain adaptation loss occupies a small total loss value, so the total loss is also small. As the training proceeds, λ gradually increases, the LMMD-domain adaptation loss increases in the ratio of the total loss, and the total loss curve shows an increasing trend in the training front stage, but as the domain adaptation of the network trains, the distribution difference (LMMD) between the source domain and the target domain becomes smaller and smaller, so that the total loss curve shows a decreasing trend in the middle stage of training, and finally converges in the middle and later stages of training.
Taking task D1→D2 as an example, the change of the accuracy of the test set along with the number of training steps in the training process of the four methods is shown in FIG. 5. It can be seen that the DAN converges slowest and the final accuracy is lowest. The convergence speed difference of the other three methods is not large, the final accuracy is about 80%, the accuracy of the SIP-DAN provided by the invention is the best, and the training process is stable and has no large fluctuation.
(6) Feature visualization experiment
In order to observe the feature distribution situation after domain adaptation, the last feature layer of the depth network is reduced to 2 dimensions by using t-SNE, and then visualized. The t-SNE dimension reduction visualization of the task D1→D2 test set is shown in FIG. 6, and it can be seen that the feature distribution of ResNet is significantly shifted, and the feature distribution is not completely aligned after DAN and D-CORAL global domain adaptation, but has a certain effect. The SIP-DAN can obtain better performance by distributing and aligning the subclasses, and the characteristic distribution alignment degree of 7 classes is obviously higher than that of other three methods.
The above disclosure is only a preferred embodiment of the present invention, and it should be understood that the scope of the invention is not limited thereto, and those skilled in the art will appreciate that all or part of the procedures described above can be performed according to the equivalent changes of the claims, and still fall within the scope of the present invention.
Claims (4)
1. A pumping unit fault diagnosis method based on a structural information retention domain adaptation network is characterized in that,
Collecting oil pumping unit indicator diagram data to obtain a source domain data set and a target domain data set;
Extracting source domain features and target domain features using a feature generator;
performing supervised training in a source domain by using source domain features and source domain sample labels to update a feature generator and a classifier;
clustering the target domain features by using an unsupervised clustering algorithm to obtain target domain pseudo tags;
Inputting the source domain features and the source domain labels and the target domain features and the target domain pseudo labels into a local maximum mean difference measurement formula to perform distribution difference calculation so as to optimize a feature generator;
the specific steps for obtaining the target domain pseudo tag comprise:
randomly initializing a membership matrix according to the data set;
Calculating an optimal membership matrix and a clustering center;
obtaining a pseudo tag of the target domain by using the membership matrix and the clustering center;
the specific steps of obtaining the pseudo tag of the target domain are as follows:
obtaining a sample clustering label according to the membership matrix;
Aiming at the samples under each cluster label, a classifier is utilized to obtain a classification label;
Counting the number of each classification label;
selecting the label with the largest number of classified labels as the label of the clustering class, and updating the clustering center;
updating the membership matrix by using a new cluster center;
And obtaining the target domain pseudo tag according to the new membership matrix.
2. The method for diagnosing the failure of the pumping unit based on the structural information retention domain adaptive network according to claim 1, wherein,
The indicator diagram refers to a closed graph formed by displacement and load discrete points.
3. The method for diagnosing the failure of the pumping unit based on the structural information retention domain adaptive network according to claim 1, wherein,
The feature generator is a feature extraction model formed by a convolutional neural network.
4. A pumping unit fault diagnosis method based on a structural information holding domain adaptive network as defined in claim 1, wherein,
The specific steps of calculating the optimal membership matrix and the clustering center comprise:
updating the clustering center by using the membership matrix;
Calculating the Euclidean distance from the sample to the clustering center by using the clustering center;
updating the membership matrix by using the cluster number, the cluster center distance and the fuzzy weight index;
Judging whether an iteration stop condition is met according to whether the maximum iteration times and the errors are reached;
obtaining the optimal membership matrix and clustering center of the algorithm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210397562.9A CN114817856B (en) | 2022-04-15 | 2022-04-15 | Beam-pumping unit fault diagnosis method based on structural information retention domain adaptation network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210397562.9A CN114817856B (en) | 2022-04-15 | 2022-04-15 | Beam-pumping unit fault diagnosis method based on structural information retention domain adaptation network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114817856A CN114817856A (en) | 2022-07-29 |
CN114817856B true CN114817856B (en) | 2024-05-28 |
Family
ID=82537062
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210397562.9A Active CN114817856B (en) | 2022-04-15 | 2022-04-15 | Beam-pumping unit fault diagnosis method based on structural information retention domain adaptation network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114817856B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115906949B (en) * | 2022-11-22 | 2023-06-20 | 东北石油大学三亚海洋油气研究院 | Petroleum pipeline fault diagnosis method and system, storage medium and petroleum pipeline fault diagnosis equipment |
CN117173350B (en) * | 2023-08-09 | 2024-07-05 | 中国科学技术大学 | Geological modeling method, system and medium based on active domain adaptive learning |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109582003A (en) * | 2018-12-03 | 2019-04-05 | 东北林业大学 | Based on pseudo label semi-supervised kernel part Fei Sheer discriminant analysis bearing failure diagnosis |
CA3128957A1 (en) * | 2019-03-04 | 2020-03-03 | Bhaskar Bhattacharyya | Near real-time detection and classification of machine anomalies using machine learning and artificial intelligence |
CN112729835A (en) * | 2021-01-30 | 2021-04-30 | 温州大学 | Multi-domain semi-supervised fault diagnosis method and device for axial plunger pump bearing |
CN112819059A (en) * | 2021-01-26 | 2021-05-18 | 中国矿业大学 | Rolling bearing fault diagnosis method based on popular retention transfer learning |
CN113792758A (en) * | 2021-08-18 | 2021-12-14 | 中国矿业大学 | Rolling bearing fault diagnosis method based on self-supervision learning and clustering |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190384255A1 (en) * | 2018-06-19 | 2019-12-19 | Honeywell International Inc. | Autonomous predictive real-time monitoring of faults in process and equipment |
-
2022
- 2022-04-15 CN CN202210397562.9A patent/CN114817856B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109582003A (en) * | 2018-12-03 | 2019-04-05 | 东北林业大学 | Based on pseudo label semi-supervised kernel part Fei Sheer discriminant analysis bearing failure diagnosis |
CA3128957A1 (en) * | 2019-03-04 | 2020-03-03 | Bhaskar Bhattacharyya | Near real-time detection and classification of machine anomalies using machine learning and artificial intelligence |
CN112819059A (en) * | 2021-01-26 | 2021-05-18 | 中国矿业大学 | Rolling bearing fault diagnosis method based on popular retention transfer learning |
CN112729835A (en) * | 2021-01-30 | 2021-04-30 | 温州大学 | Multi-domain semi-supervised fault diagnosis method and device for axial plunger pump bearing |
CN113792758A (en) * | 2021-08-18 | 2021-12-14 | 中国矿业大学 | Rolling bearing fault diagnosis method based on self-supervision learning and clustering |
Non-Patent Citations (3)
Title |
---|
Bearing Fault Diagnosis Under Variable Working Conditions Based on Domain Adaptation Using Feature Transfer Learning;Zhe Tong等;《IEEE Access 》;20181123;第6卷;76187 - 76197 * |
基于无监督域适应的变负载下滚动轴承状态评估方法研究;邹佳悦;《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》;20210215(第02期);C029-294 * |
基于领域自适应的抽油机故障诊断方法研究;卢飞;《重庆科技学院硕士专业学位毕业论文》;20240402;1-65 * |
Also Published As
Publication number | Publication date |
---|---|
CN114817856A (en) | 2022-07-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109710636B (en) | Unsupervised industrial system anomaly detection method based on deep transfer learning | |
CN113935406B (en) | Mechanical equipment unsupervised fault diagnosis method based on countermeasure flow model | |
CN114817856B (en) | Beam-pumping unit fault diagnosis method based on structural information retention domain adaptation network | |
CN106124175B (en) | A kind of compressor valve method for diagnosing faults based on Bayesian network | |
CN112964469B (en) | Online fault diagnosis method for rolling bearing under variable load of transfer learning | |
CN113988126B (en) | Rolling bearing fault diagnosis method based on few-label data feature migration | |
Subbotin | The neuro-fuzzy network synthesis and simplification on precedents in problems of diagnosis and pattern recognition | |
CN111639678B (en) | INS/GPS integrated navigation fault detection and diagnosis method based on integrated neural network | |
CN112756759B (en) | Spot welding robot workstation fault judgment method | |
CN114295377B (en) | CNN-LSTM bearing fault diagnosis method based on genetic algorithm | |
CN115600150A (en) | Multi-mode gearbox fault diagnosis method based on deep migration learning | |
CN116894187A (en) | Gear box fault diagnosis method based on deep migration learning | |
CN113255573B (en) | Pedestrian re-identification method based on mixed cluster center label learning and storage medium | |
CN113609569B (en) | Distinguishing type generalized zero sample learning fault diagnosis method | |
CN108803555B (en) | Sub-health online identification and diagnosis method based on performance monitoring data | |
CN113378998B (en) | Stratum lithology while-drilling identification method based on machine learning | |
CN114354194B (en) | Rolling bearing fault diagnosis method based on full convolution self-encoder and optimized support vector machine | |
CN113887342A (en) | Equipment fault diagnosis method based on multi-source signals and deep learning | |
CN108009571A (en) | A kind of semi-supervised data classification method of new direct-push and system | |
CN114118138A (en) | Bearing composite fault diagnosis method based on multi-label field self-adaptive model | |
CN115791174B (en) | Rolling bearing abnormality diagnosis method, system, electronic equipment and storage medium | |
WO2019178930A1 (en) | Fault diagnosis method for mechanical device | |
CN116542170A (en) | Drainage pipeline siltation disease dynamic diagnosis method based on SSAE and MLSTM | |
CN115165366A (en) | Variable working condition fault diagnosis method and system for rotary machine | |
Saufi et al. | Machinery fault diagnosis based on a modified hybrid deep sparse autoencoder using a raw vibration time-series signal |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |