CN113807444A - Fault diagnosis method based on constraint-confrontation convolutional self-coding memory fusion network - Google Patents
Fault diagnosis method based on constraint-confrontation convolutional self-coding memory fusion network Download PDFInfo
- Publication number
- CN113807444A CN113807444A CN202111107279.XA CN202111107279A CN113807444A CN 113807444 A CN113807444 A CN 113807444A CN 202111107279 A CN202111107279 A CN 202111107279A CN 113807444 A CN113807444 A CN 113807444A
- Authority
- CN
- China
- Prior art keywords
- data
- network
- module
- self
- fusion
- 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.)
- Granted
Links
- 230000004927 fusion Effects 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000003745 diagnosis Methods 0.000 title claims abstract description 19
- 238000001514 detection method Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 11
- 230000005284 excitation Effects 0.000 claims abstract description 3
- 238000012360 testing method Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 4
- 230000001133 acceleration Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000004806 packaging method and process Methods 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims description 2
- 230000003042 antagnostic effect Effects 0.000 claims 1
- 238000013528 artificial neural network Methods 0.000 abstract description 5
- 238000009826 distribution Methods 0.000 abstract description 4
- 238000001125 extrusion Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 13
- 230000008569 process Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 4
- 238000013136 deep learning model Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 238000002679 ablation Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000007499 fusion processing Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000001105 regulatory effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000002902 bimodal effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention provides a fault diagnosis method based on a constraint-confrontation convolutional self-coding memory fusion network, which comprises the following steps of: s1, establishing a data fusion fault detection framework; s2, collecting a data set, preprocessing the data S3, inputting the processed vibration and current data into a convolution self-coding and training to obtain independent characteristics of two modes; s4, after the coded outputs of the two convolution self-encoders are subjected to attention weighting and gradient inversion layer through an SE module, the coded outputs are input to a mode discriminator to be trained so as to obtain mode invariance; and S5, performing feature splicing on the output of the SE (extrusion excitation) module, inputting the output into a long and short memory neural network (LSTM) for fusion, and using the output of the LSTM for deducing the network (Inf). According to the invention, by designing the constraint-based anti-convolution self-coding memory fusion network, the invariance of each mode (vibration and current) can be effectively captured to reduce the mode distribution difference, and simultaneously multi-mode data are fused for a fault diagnosis task.
Description
Technical Field
The invention relates to a fault diagnosis method based on multi-mode data fusion, in particular to a fault diagnosis method based on a constraint-confrontation convolutional self-coding memory fusion network.
Background
With the development of intelligent technology, revolutionary changes are brought to industrial equipment, and intelligent equipment becomes the leading edge of high-end equipment and the core of manufacturing industry, and is also an important mark for measuring the level of national technological innovation and high-end manufacturing industry. The intelligent equipment is electromechanical equipment with deep integration of information technology and artificial intelligence technology, and if a very small fault cannot be processed in time in the operation process, the coordinated operation state of the equipment can be damaged, the equipment is shut down, even the equipment is damaged, personal safety accidents are caused, and direct economic loss is caused to enterprises. Therefore, how to effectively detect the health of the equipment is a key for ensuring the normal operation of the equipment. In practical situations, a large number of sensors are arranged to monitor the fusion, which provides a necessary condition for the multi-sensor fusion technology. However, the diversity of the distribution of these sensing signals (e.g., vibration, current, etc.) presents difficulties for multi-sensor fusion. Therefore, how to effectively fuse the multi-modal sensing signals to obtain more comprehensive state information is the key to solving the problem of detecting the health state of the device.
At present, the deep learning technology is the mainstream of the current sensor fusion technology, such as a convolutional neural network, an autoencoder and the like. However, most methods independently extract features of data of different modalities before fusion, and do not consider interaction features between modalities, namely modality invariance, which is not beneficial to reducing differences between modality distributions in the fusion process.
Disclosure of Invention
Aiming at the problems, the invention provides a fault diagnosis method based on a constraint-confrontation convolutional self-coding memory fusion network, which can effectively step the interaction relation among the modes before fusion, reduce the calculation burden of the fusion process and obtain good performance in a fault diagnosis task.
In order to achieve the purpose, the invention adopts the following technical scheme:
s1, establishing a data fusion fault detection framework;
s2, collecting a data set and preprocessing the data;
s3, inputting the processed vibration and current data into a convolution self-coding to obtain independent characteristics of two modes;
s4, after the coded outputs of the two convolution self-encoders are subjected to attention weighting and gradient inversion layer through an SE module, the coded outputs are input to a mode discriminator to be trained so as to obtain mode invariance;
and S5, performing feature splicing on the output of the SE module, inputting the output of the SE module into the LSTM network for fusion, and using the output of the SE module in the inference network (Inf).
Further, in step S1, the data fusion fault detection framework includes six modules, which are connected in sequence, including a data acquisition module, a preprocessing module, a convolutional self-coding representation module, a constraint confrontation attention representation module, a memory fusion inference module, and a network confrontation training module.
Furthermore, the data acquisition module acquires current and vibration signals as a training data set and a test data set through the current sensor and the acceleration sensor.
Further, the data preprocessing in step S2 includes a data normalization process, which normalizes the original vibration and current data respectively to eliminate the dimensional influence.
Furthermore, after the normalized vibration and current data of the packaging equipment are segmented and intercepted, the 1-D data are recombined into a 2-D grid matrix form in a segmentation mode.
Further, the processed data is input into different convolutional self-encoding in step S3, and an encoded representation and a decoded representation are obtained.
Further, after the encoded output passes through the squeeze excitation module (SE Block) and the gradient inversion layer in step S4, the encoded output is input to the mode discriminator for counterlearning to obtain the invariance attention representation.
Further, in step S5, the invariance notice representation is spliced and then the memory fusion representation is acquired by the LSTM network, and then input to the inference network (Inf) to acquire a network inference output.
The invention has the beneficial effects that:
the invention establishes a data fusion fault detection framework and designs a new joint regular loss function to train the anti-convolution self-coding memory fusion network, thereby helping the network to learn corresponding characteristics; and acquiring network hyper-parameters and loss function weight parameters through a grid search algorithm, thereby acquiring an optimal model. Finally, the network achieves the achievement exceeding the traditional deep learning model in the fault diagnosis performance.
Drawings
FIG. 1 is a block diagram of the overall process of the method of the present invention;
FIG. 2 is a detailed parameter topology diagram of a network component;
FIG. 3 is a visual graph of a clustering fusion effect T-SNE hierarchy of a network on a test set;
FIG. 4 is a graph of confusion matrices for different methods of networks on a test set;
FIG. 5 is a graph of ROC curves for different methods of a network on a test set;
fig. 6 is a graph comparing the performance of different methods on different data sets by the network.
Detailed Description
In order to facilitate an understanding of the invention, preferred embodiments of the invention are set forth below. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used in the description is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
A fault diagnosis method based on a constraint-confrontation convolutional self-coding memory fusion network comprises the following specific steps:
and S1, establishing a data fusion fault detection framework.
As shown in fig. 1, the data fusion fault detection framework includes a data acquisition module, a preprocessing module, a convolutional self-coding representation module, a constraint confrontation attention representation module, a memory fusion inference module, and a network confrontation training module, which are connected in sequence. A training data set and a testing data set are obtained through an acceleration sensor and a current sensor in a mechanical equipment data acquisition module.
And S2, acquiring data and preprocessing the acquired data.
The data preprocessing comprises data normalization and a data dimension-increasing process for converting 1-D data into 2-D grid matrixes. The specific flow is that firstly, input data is normalized to eliminate the influence caused by dimension difference among different data; and then carrying out data rearrangement on the normalized data from 1-D to 2-D to prepare for convolution input.
S3, obtaining an encoded and decoded representation of the input data from the encoder by convolution.
The vibration and current data processed in step S2 are input to a convolutional self-encoder with two parameters not shared, and an encoded representation and a decoded representation of the two modal data are obtained, respectively (the encoded representation is used as a back-end input, and the decoded representation is used for loss function constraint learning). This section is represented as follows:
input (x) of a vibration signal using two convolutional self-encoding networks (CAEs), respectivelyv) Current signal input (x)c) Performing encoding-decoding learning to extract their independent features, whose encoding representation and decoding representation are respectively:
in the formula: CAEm、hm、UmAndthe CAE network of m-mode, the encoded representation, the decoded (reconstructed) representation, the network parameters of the CAE network, respectively.
S4, obtaining the anti-attention representation of the input data through SE Block, gradient inversion layer, modality discriminator.
The encoded representations of the two modalities acquired in step S3 are sequentially input into SE Block, gradient inversion layer, modality discriminator to acquire an attention representation of the two modalities and an output of the modality discriminator (the attention representation is used for back-end inference, and the modality discriminator input is used for a loss function for counterlearning). Wherein the attention representation and the modality discriminator output are respectively as follows:
in the formula: SEm、Respectively representing an SE module, SE network parameters and attention weighting of an m mode; d denotes a mode discriminator, p denotes an inferred set of vibration and current signals with respect to the mode discriminator, thetaDFor the mode discriminator network parameters, Q is the gradient inversion layer (GRL) function (output is the same as the identity function, but the gradient direction is opposite).
S5, fusing the attention representations of the two modalities through LSTM and performing inference tasks.
And splicing the attention representation outputs of the two modalities in the step S4, inputting the spliced attention representation outputs into the LSTM network, acquiring a fused memory fusion representation by virtue of the memory capacity of the LSTM network, and inputting the fused memory representation into the inference network to acquire a final inference representation, wherein the representation process is as follows:
first, two attention weighted representation features are stitched:
in the formula:a vector splicing operation is represented as a vector splicing operation,a weighted attention representation of the two modality encoding is represented.
The stitched joint representation is then input to the LSTM, resulting in the following representation:
R=LSTM(hSE;θLstm) (6)
in the formula: LSTM, thetaLstmRespectively representing the long and short memory neural networks and network parameters thereof.
Finally, the memory fusion representation is input to an inference neural network (Inf) for fault inference as follows:
Y=Inf(R;θInf) (7)
in the formula: y is an inferred representation, θInfTo infer network parameters. The network consists of two fully connected layers activated by leak Relu.
And (3) learning strategy:
the weighting constraint of the proposed method is mainly reflected in the design of the loss function, and the new loss function is designed as follows:
Ltotal=Ltask+αLrecon_v+βLrecon_c+δLadv+ηLsim (8)
in the formula: alpha, beta, delta and eta are loss function regulating factors for regulating the contribution of each loss, and the network is trained by aiming at minimizing the loss functions. Wherein L istask、Lrecon_v、Lrecon_c、Ladv、LsimRespectively cross entropy loss, reconstruction loss, countermeasure loss, and similarity loss. The definitions are as follows:
the task loss is defined as follows:
in the formula: y isiIs a true label for sample i, YiThe result is inferred for the network for sample i.
The reconstruction loss is defined as follows:
in the formula: x is the number ofmAnd UmRepresenting the original input data and the reconstructed data separately,represents the square L2And (4) norm.
The loss of confrontation is as follows:
assuming that there are N vibration samples and N current representative samples, respectively, the loss L is combatedadvThe following were used:
in the formula: p represents the output of the vibration and current signals with respect to the mode discriminator, diE {0,1} represents the modal signature of the vibration and current signal.
Loss of similarity is shown below:
let X and Y be bounded random samples in tight intervals [ a, b ]]NHaving respective probability distributions p and q, the center-to-center deviation regularizer CMDKThe empirical estimate defined as the CMD metric is as follows:
wherein C iskAnd E (X) is represented as follows:
in the formula: e (X) is the empirically expected vector, C, for sample Xk(X) is the vector of all k-th order sample central moments in the X coordinate. The adopted CMD similarity loss is as follows:
experimental analysis:
in order to verify the detection precision and the detection effect of the model, an experimental program is realized through python programming, and experimental equipment comprises: (1) a Processor (AMD Ryzen 52600X Six-Core Processor,3.60 GHz); (2) operating a memory (16G); (3) display card (NVIDIA GeForce GTX 1660, 6G); (4) code operating environment (Pytorch 1.2.0, Python 3.7.9).
In order to verify the effectiveness of the model, motor bearing test data including voltage and current data measured by the bearing in five different states are adopted, as shown in table 1:
TABLE 1 introduction of data set
Where each set of data contains 160000 discrete points. In addition, for effective training, the test network divides each set of data into a training set and a test set in a 3:1 ratio.
It should be noted that when the data preprocessing portion needs to normalize the data, the data needs to be mapped to the range of [ -1,1] in order to better conform to the characteristics of the vibration and current data. In the iterative training process, the experiment adopts Adam optimizer training network with max epoch 500, Batch size 100 and learning rate 0.01, and each component specific parameter of the network is as shown in fig. 2. Furthermore, the weight over-parameter of the loss function is set as follows: α ═ 0.6, β ═ 0.3, δ ═ 0.1, and η ═ 0.1.
In order to verify the effectiveness and superiority of the invention, the experiment that the monomodal traditional deep learning model comprises a Convolutional Neural Network (CNN) and a convolutional self-encoding neural network (CAE) and the bimodal traditional deep learning fusion model comprises a convolutional fusion neural network (CNN-FN), a convolutional self-encoding fusion network (CAE-FN) and a convolutional multiple attention fusion network (CNN-MAFN) is designed, and the result is shown in Table 2:
TABLE 2 comparative experiment
As can be seen from the results in Table 2, compared with the traditional single-modal deep learning model and the traditional dual-modal deep learning fusion model, the method has superior performance in fault diagnosis.
Then, in order to verify the reasonability of the model in the invention, the model process is visually analyzed, and the model structure and the loss function are subjected to ablation research experiments. The model process result is shown in table 3, and as can be seen from fig. 3, as the hierarchy of the constraint-resisting convolution self-coding memory fusion network disclosed by the invention is deepened, the fusion clustering effect on samples of the same category is better, and the rationality of the model is verified visually. The results of the ablation study are shown in Table 3, from which it can be seen that the lack of any component in the design of the network and the loss function results in a degradation of the network performance, which numerically validates the rationality of the model
TABLE 3 ablation study
In addition, in order to verify the classification performance of the network, the network is subjected to visualization of the confusion matrix and the ROC curve, and is compared with other traditional deep learning models, and as can be seen from fig. 4 and 5, compared with the traditional deep learning network, the network classification method has the advantages that the good effect and the optimal classification performance are achieved on the expression of the confusion matrix and the ROC curve.
Finally, in order to verify the generalization performance of the network of the present invention, the data sets from different sources were used to verify the network, and the results are shown in fig. 6. From fig. 6, it can be seen that the present invention achieves the optimal effect on different data sets compared with the conventional model, and thus it can be seen that the present invention has good generalization performance on equipment fault detection.
The embodiments of the present invention are described only for the preferred embodiments of the present invention, and not for the limitation of the concept and scope of the present invention, and various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the design concept of the present invention shall fall within the protection scope of the present invention, and the technical contents of the present invention which are claimed are all described in the claims.
Claims (8)
1. A fault diagnosis method based on a constraint-confrontation convolutional self-coding memory fusion network is characterized by comprising the following steps:
s1, establishing a data fusion fault detection framework;
s2, collecting a data set and preprocessing the data;
s3, inputting the processed vibration and current data into a convolution self-coding to obtain independent characteristics of two modes;
s4, after the coded outputs of the two convolution self-encoders are subjected to attention weighting and gradient inversion layer through an SE module, the coded outputs are input to a mode discriminator to be trained so as to obtain mode invariance;
and S5, performing feature splicing on the output of the SE module, inputting the output of the SE module into the LSTM network for fusion, and outputting the output of the SE module for deducing the network.
2. The fault diagnosis method based on the constraint-confronted convolutional self-coding memory fusion network of claim 1, wherein the data fusion fault detection framework in step S1 comprises a data acquisition module, a preprocessing module, a convolutional self-coding representation module, a constraint-confronted attention representation module, a memory fusion inference module and a network confrontation training module which are connected in sequence.
3. The method for fault diagnosis based on the constraint-based deconvolution self-coding memory fusion network according to claim 2, wherein the data acquisition module acquires current and vibration signals as a training data set and a test data set through a current sensor and an acceleration sensor.
4. The method for fault diagnosis based on constrained deconvolution self-coding memory fusion network of claim 1, wherein the data preprocessing in step S2 includes a data normalization procedure, in which the original vibration and current data are normalized respectively to eliminate dimensional influence.
5. The method for fault diagnosis based on the constraint-based anti-convolution self-coding memory fusion network is characterized in that after the normalized vibration and current data of the packaging equipment are segmented and intercepted, the 1-D data are recombined into a 2-D grid matrix form in a segmentation mode.
6. The method for fault diagnosis based on constraint-based antagonistic convolutional self-coding memory fusion network of claim 1, wherein the processed data is inputted into different convolutional self-coding to obtain coded representation and decoded representation in step S3.
7. The method for fault diagnosis based on constrained deconvolution self-coding memory fusion network of claim 1, wherein the encoded output is inputted to a mode discriminator for counterlearning after passing through the squeeze excitation module and the gradient inversion layer in step S4 to obtain the invariant attention representation.
8. The method for fault diagnosis based on constraint-based deconvolution self-coding memory fusion network of claim 1, wherein the invariance attention representation is spliced in step S5, and then the memory fusion representation is obtained through LSTM network, and then input into inference network to obtain network inference output.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111107279.XA CN113807444B (en) | 2021-09-22 | 2021-09-22 | Fault diagnosis method based on constraint countermeasure convolution self-coding memory fusion network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111107279.XA CN113807444B (en) | 2021-09-22 | 2021-09-22 | Fault diagnosis method based on constraint countermeasure convolution self-coding memory fusion network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113807444A true CN113807444A (en) | 2021-12-17 |
CN113807444B CN113807444B (en) | 2024-08-23 |
Family
ID=78939905
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111107279.XA Active CN113807444B (en) | 2021-09-22 | 2021-09-22 | Fault diagnosis method based on constraint countermeasure convolution self-coding memory fusion network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113807444B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114692694A (en) * | 2022-04-11 | 2022-07-01 | 合肥工业大学 | Equipment fault diagnosis method based on feature fusion and integrated clustering |
CN116243683A (en) * | 2023-03-15 | 2023-06-09 | 青岛澎湃海洋探索技术有限公司 | Method for diagnosing faults of propulsion system based on torque and multi-head self-encoder |
CN117150408A (en) * | 2023-09-11 | 2023-12-01 | 元始智能科技(南通)有限公司 | Motor fault diagnosis method based on multi-mode comparison learning |
CN117520590A (en) * | 2024-01-04 | 2024-02-06 | 武汉理工大学三亚科教创新园 | Ocean cross-modal image-text retrieval method, system, equipment and storage medium |
CN117725529A (en) * | 2024-02-18 | 2024-03-19 | 南京邮电大学 | Transformer fault diagnosis method based on multi-mode self-attention mechanism |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110929765A (en) * | 2019-11-06 | 2020-03-27 | 北京工业大学 | Convolution self-coding fault monitoring method based on batch imaging |
CN112329685A (en) * | 2020-11-16 | 2021-02-05 | 常州大学 | Method for detecting crowd abnormal behaviors through fusion type convolutional neural network |
CN112528548A (en) * | 2020-11-27 | 2021-03-19 | 东莞市汇林包装有限公司 | Self-adaptive depth coupling convolution self-coding multi-mode data fusion method |
-
2021
- 2021-09-22 CN CN202111107279.XA patent/CN113807444B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110929765A (en) * | 2019-11-06 | 2020-03-27 | 北京工业大学 | Convolution self-coding fault monitoring method based on batch imaging |
WO2021088377A1 (en) * | 2019-11-06 | 2021-05-14 | 北京工业大学 | Convolutional auto-encoding fault monitoring method based on batch imaging |
CN112329685A (en) * | 2020-11-16 | 2021-02-05 | 常州大学 | Method for detecting crowd abnormal behaviors through fusion type convolutional neural network |
CN112528548A (en) * | 2020-11-27 | 2021-03-19 | 东莞市汇林包装有限公司 | Self-adaptive depth coupling convolution self-coding multi-mode data fusion method |
Non-Patent Citations (1)
Title |
---|
丁石川;厉雪衣;杭俊;王尹江;王群京;: "深度学习理论及其在电机故障诊断中的研究现状与展望", 电力***保护与控制, no. 08, 15 April 2020 (2020-04-15) * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114692694A (en) * | 2022-04-11 | 2022-07-01 | 合肥工业大学 | Equipment fault diagnosis method based on feature fusion and integrated clustering |
CN114692694B (en) * | 2022-04-11 | 2024-02-13 | 合肥工业大学 | Equipment fault diagnosis method based on feature fusion and integrated clustering |
CN116243683A (en) * | 2023-03-15 | 2023-06-09 | 青岛澎湃海洋探索技术有限公司 | Method for diagnosing faults of propulsion system based on torque and multi-head self-encoder |
CN116243683B (en) * | 2023-03-15 | 2024-02-13 | 青岛澎湃海洋探索技术有限公司 | Method for diagnosing faults of propulsion system based on torque and multi-head self-encoder |
CN117150408A (en) * | 2023-09-11 | 2023-12-01 | 元始智能科技(南通)有限公司 | Motor fault diagnosis method based on multi-mode comparison learning |
CN117520590A (en) * | 2024-01-04 | 2024-02-06 | 武汉理工大学三亚科教创新园 | Ocean cross-modal image-text retrieval method, system, equipment and storage medium |
CN117520590B (en) * | 2024-01-04 | 2024-04-26 | 武汉理工大学三亚科教创新园 | Ocean cross-modal image-text retrieval method, system, equipment and storage medium |
CN117725529A (en) * | 2024-02-18 | 2024-03-19 | 南京邮电大学 | Transformer fault diagnosis method based on multi-mode self-attention mechanism |
CN117725529B (en) * | 2024-02-18 | 2024-05-24 | 南京邮电大学 | Transformer fault diagnosis method based on multi-mode self-attention mechanism |
Also Published As
Publication number | Publication date |
---|---|
CN113807444B (en) | 2024-08-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113807444A (en) | Fault diagnosis method based on constraint-confrontation convolutional self-coding memory fusion network | |
Czimmermann et al. | Visual-based defect detection and classification approaches for industrial applications—a survey | |
Hao et al. | A steel surface defect inspection approach towards smart industrial monitoring | |
CN112926641B (en) | Three-stage feature fusion rotating machine fault diagnosis method based on multi-mode data | |
CN109766992B (en) | Industrial control abnormity detection and attack classification method based on deep learning | |
Yu et al. | Improved Butterfly Optimizer‐Configured Extreme Learning Machine for Fault Diagnosis | |
CN114970715B (en) | Variable working condition fault diagnosis method and system under constraint of small sample and unbalanced data | |
CN114972213A (en) | Two-stage mainboard image defect detection and positioning method based on machine vision | |
Miao et al. | Adaptive densely connected convolutional auto-encoder-based feature learning of gearbox vibration signals | |
Tong et al. | Multi-sensor information fusion and coordinate attention-based fault diagnosis method and its interpretability research | |
Yang et al. | An intelligent singular value diagnostic method for concrete dam deformation monitoring | |
Meister et al. | Cross-evaluation of a parallel operating SVM–CNN classifier for reliable internal decision-making processes in composite inspection | |
CN112528548A (en) | Self-adaptive depth coupling convolution self-coding multi-mode data fusion method | |
CN114970326A (en) | Oil well fault diagnosis method based on improved capsule neural network | |
CN117075582A (en) | Industrial process generalized zero sample fault diagnosis method based on DSECMR-VAE | |
CN118070682B (en) | Spiral bolt hanging ring damage assessment method and device based on artificial intelligence | |
CN115424074A (en) | Classification method, device and equipment applied to industrial detection | |
Liu et al. | A dual-branch balance saliency model based on discriminative feature for fabric defect detection | |
Zhang et al. | A flexible monitoring framework via dynamic-multilayer graph convolution network | |
Sun et al. | Physics-inspired multimodal machine learning for adaptive correlation fusion based rotating machinery fault diagnosis | |
CN115618225A (en) | Traditional Chinese medicine state identification method based on graph attention network | |
CN114926702A (en) | Small sample image classification method based on depth attention measurement | |
CN114387524A (en) | Image identification method and system for small sample learning based on multilevel second-order representation | |
Castillo et al. | A hybrid fuzzy‐fractal approach for time series analysis and plant monitoring | |
Shi et al. | Transfer Learning Using Cluster Centers for Surface Defect Identification of Piston Rods in Pneumatic Cylinders |
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 |