CN115410069A - Fault detection method and system based on multiple attention mechanism - Google Patents
Fault detection method and system based on multiple attention mechanism Download PDFInfo
- Publication number
- CN115410069A CN115410069A CN202211028528.0A CN202211028528A CN115410069A CN 115410069 A CN115410069 A CN 115410069A CN 202211028528 A CN202211028528 A CN 202211028528A CN 115410069 A CN115410069 A CN 115410069A
- Authority
- CN
- China
- Prior art keywords
- feature map
- feature
- classification
- neural network
- attention mechanism
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Multimedia (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Biophysics (AREA)
- Entrepreneurship & Innovation (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Image Analysis (AREA)
Abstract
The application relates to the field of intelligent fault diagnosis of mechanical equipment, and particularly discloses a fault detection method and a fault detection system based on a multiple attention mechanism.
Description
Technical Field
The present disclosure relates to the field of intelligent diagnosis of faults of mechanical devices, and more particularly, to a fault detection method and system based on a multiple attention mechanism.
Background
In recent years, with the development of advanced technologies such as artificial intelligence, 5G, internet of things, block chains and the like, the upgrading of the traditional manufacturing industry by intelligent manufacturing, intelligent production lines and the like is actively carried out. For intelligent manufacturing or intelligent production lines, an important technology is fault detection and diagnosis for mechanical equipment. Among them, it is difficult to detect and diagnose the fault of the rotating machinery (e.g. a tilter, a rotating cantilever crane, a mechanical turntable, etc.) because the rotating machinery has a complicated structure and a harsh working environment, and the fault signal thereof often exhibits weak property, nonlinearity and coupling property, and is submerged in strong background noise and interference signals. Therefore, a fault detection method optimally directed to a rotating mechanical device is desired.
In recent years, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks also exhibit a level close to or even exceeding that of humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
In recent years, deep learning and development of neural networks provide solutions and schemes for fault diagnosis and early warning of rotating mechanical equipment.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a fault detection method and a system thereof based on a multi-attention mechanism, wherein a variation modal decomposition and Hilbert transform method is adopted in a data source domain to process vibration signals of rotating mechanical equipment into time-frequency images in the same dimension as infrared images, convolutional neural network models with a channel attention mechanism and a space attention mechanism are respectively used for coding the time-frequency images and the infrared images so as to extract specific local time-frequency implicit associated features which are more concerned with fault detection in the time-frequency images and more focus on a temperature implicit associated mode capable of representing faults, and derivative information hyperconvexity factors between feature maps are further calculated when the two features are fused, so that manifold differences between the feature maps can adapt to convex monotonicity on projection of sub-dimensions, and the precision and the accuracy of fault classification are improved.
According to an aspect of the present application, there is provided a fault detection method based on a multiple attention mechanism, including:
a training phase comprising:
acquiring a vibration signal and an infrared image of mechanical equipment through a vibration sensor and an infrared camera;
converting the vibration signal into a time-frequency image with the same size as the infrared image by a variational modal decomposition and Hilbert transform method;
passing the time-frequency image through a first convolution neural network model using a channel attention mechanism to obtain a first feature map;
passing the infrared image through a second convolutional neural network using a spatial attention mechanism to obtain a second feature map;
calculating a derivative information hyperconvexity measurement factor between the first feature map and the second feature map, wherein the derivative information hyperconvexity measurement factor is a logarithmic function value of a weighted sum of distances between feature values of each position in the first feature map and corresponding feature values of each position in the second feature map;
fusing the first feature map and the second feature map to obtain a classification feature map;
passing the classification feature map through a classifier to obtain a classification loss function value;
calculating a weighted sum between the derivative information hyperconvexity metric factor and the classification loss function value as a loss function value to train the first convolutional neural network model and the second convolutional neural network model; and
an inference phase comprising:
acquiring a vibration signal and an infrared image of mechanical equipment to be detected through a vibration sensor and an infrared camera;
converting the vibration signal into a time-frequency image with the same size as the infrared image by a variational modal decomposition and Hilbert transform method;
passing the time-frequency image through the first convolution neural network model which is trained in a training phase and uses a channel attention mechanism to obtain a first feature map;
passing the infrared image through the second convolutional neural network which is trained by the training stage and uses the spatial attention mechanism to obtain a second feature map;
fusing the first feature map and the second feature map to obtain a classification feature map; and enabling the classification characteristic diagram to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the mechanical equipment to be detected has faults or not.
According to the fault detection method and the fault detection system based on the multiple attention mechanism, a variation modal decomposition method and a Hilbert transform method are adopted in a data source domain to process vibration signals of rotating mechanical equipment into time-frequency images which are in the same dimension as infrared images, convolutional neural network models with channel attention mechanisms and space attention mechanisms are respectively used for coding the time-frequency images and the infrared images so as to extract specific local time-frequency implicit associated features which are more concerned in the time-frequency images and suitable for fault detection and temperature implicit associated modes which can represent faults, and derivative information hyperconvexity factors between feature maps are further calculated when the two features are fused, so that the manifold difference between the feature maps can adapt to the convexity monotonicity on projection of each sub-dimension, and the precision and the accuracy of fault classification are improved.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
Fig. 1 is a schematic view of a scenario of a fault detection method based on a multiple attention mechanism according to an embodiment of the present application.
Fig. 2A is a flowchart of a training phase in a fault detection method based on a multiple attention mechanism according to an embodiment of the present application.
FIG. 2B is a flowchart of an inference stage in a multiple attention mechanism based fault detection method according to an embodiment of the application.
Fig. 3A is a schematic diagram illustrating a training phase of a fault detection method based on multiple attention mechanisms according to an embodiment of the present disclosure.
Fig. 3B is a schematic diagram of an inference stage in a fault detection method based on multiple attention mechanisms according to an embodiment of the present application.
FIG. 4 is a block diagram of a multiple attention mechanism based fault detection system according to an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of scenes
As described above, in recent years, with the development of advanced technologies such as artificial intelligence, 5G, internet of things, and block chains, the upgrading of the conventional manufacturing industry such as intelligent manufacturing and intelligent production line has been actively carried out. For intelligent manufacturing or intelligent production lines, an important technology is fault detection and diagnosis of mechanical equipment. Among them, it is difficult to detect and diagnose the fault of the rotating machinery (e.g. a tilter, a rotating cantilever crane, a mechanical turntable, etc.) because the rotating machinery has a complicated structure and a harsh working environment, and the fault signal thereof often exhibits weak property, nonlinearity and coupling property, and is submerged in strong background noise and interference signals. Therefore, a fault detection method optimally directed to a rotating mechanical device is desired.
In recent years, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks also exhibit a level close to or even exceeding that of humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
In recent years, deep learning and development of neural networks provide solutions and schemes for fault diagnosis and early warning of rotating mechanical equipment.
Specifically, the inventors of the present application have attempted to use a two-stage classification model for fault detection and early warning of rotating mechanical equipment. In particular, considering that the source information of the rotating mechanical equipment has the characteristics of weakness, nonlinearity, coupling and the like, a technical route of multi-source information feature fusion is further adopted to extract classification features for fault detection. In the embodiment of the application, the multi-source information comprises an infrared image of the rotating mechanical equipment and a vibration signal of the rotating mechanical equipment when the rotating mechanical equipment vibrates. It should be noted that the infrared image is two-dimensional non-structural data, has a temperature-sensitive characteristic, and can well identify single faults, coupling faults, fault positions, severity and the like with temperature rise changes.
In the technical scheme of the application, a Variational Modal Decomposition (VMD) method and a Hilbert Transform (HT) method are adopted in a data source domain to process a vibration signal into a time-frequency image which has the same dimension as an infrared image. That is, the vibration signal is subjected to data isomerisation in a data source domain to convert the vibration signal into a data representation having the same data structure as the infrared image. Meanwhile, the time-frequency image can represent the time-domain characteristics, the frequency-domain characteristics and the association between the time-domain characteristics and the frequency-domain characteristics of the vibration of the rotating mechanical equipment.
The time-frequency image is then encoded using a convolutional neural network model to extract implicitly correlated features of the vibration of the rotating mechanical device in both the time dimension and the space dimension. Considering that if the rotating mechanical device fails, it may exhibit a specific vibration mode, e.g. high-frequency vibration, on the vibration signal, in the technical solution of the present application, the convolutional neural network model further integrates a channel attention mechanism to pay attention to a specific local time-frequency implicit correlation feature in the time-frequency image suitable for failure detection.
Meanwhile, the infrared image itself contains abundant image information, that is, the infrared image information can capture the temperature change of each position of the rotating mechanical equipment and the mode correlation between the temperatures of each position. Correspondingly, the infrared image passes through a second convolutional neural network, and spatial convolutional coding is carried out on the infrared image through the second convolutional neural network so as to extract the distribution implicit correlation characteristics of the temperature change of the rotary mechanical equipment in the space. In order to enable the second convolutional neural network to focus more on a temperature implicit correlation mode capable of characterizing a fault, in the technical solution of the present application, a spatial attention mechanism is further integrated into the second convolutional neural network model.
Further, the first feature map and the second feature map are fused to obtain a classification feature map, and then the classification can be carried out by using a classifier. However, considering the respective dimension distribution characteristics based on the time domain image and the infrared image, the feature extraction is performed by using the first convolutional neural network with channel attention and the second convolutional neural network with space attention in the present application, and since the feature extraction is performed by using the first convolutional neural network and the second convolutional neural network respectively in different dimensions (channel dimension and space dimension), the feature distribution of the obtained feature map can be converged in different dimensions, so that sparsity between dimensions can be generated during fusion, and the quality of the fused feature map is affected.
Therefore, in the present application, the first characteristic diagram F is referred to 1 And a second characteristic diagram F 2 Training first and second convolutional neural networks based on channel attention and spatial attention, respectively, by computing derivative information hyperconvexity factors between feature maps as a loss function, thereby optimizing the fused feature maps.
The derivative information hyperconvexity loss function is expressed as:
wherein
Feature values representing respective positions in the first feature map,
a feature value representing each position in the second feature map, | · | represents a distance between the first feature map and the second feature map.
The derivative information hyper-convexity degree loss function is information measurement between internal element sub-dimensions of the feature map in a high-dimensional space, and can be used for derivative representation of the hyper-convex consistency of manifold, so that the first feature map F can be enabled by training a model by the loss function 1 And a second characteristic diagram F 2 The inter-manifold difference can adapt to the convex monotonicity on the projection of each sub-dimension, so that the classification effect of the fused feature map is improved.
Based on this, the present application proposes a fault detection method based on a multiple attention mechanism, which includes a training phase including: acquiring a vibration signal and an infrared image of mechanical equipment through a vibration sensor and an infrared camera; converting the vibration signal into a time-frequency image with the same size as the infrared image by a variational modal decomposition and Hilbert transform method; passing the time-frequency image through a first convolutional neural network model using a channel attention mechanism to obtain a first feature map; passing the infrared image through a second convolutional neural network using a spatial attention mechanism to obtain a second feature map; calculating a derivative information hyperconvexity metric between the first feature map and the second feature map, wherein the derivative information hyperconvexity metric is a logarithmic function value of a weighted sum of distances between feature values of various positions in the first feature map and corresponding feature values of various positions in the second feature map; fusing the first feature map and the second feature map to obtain a classification feature map; passing the classification feature map through a classifier to obtain a classification loss function value; calculating a weighted sum between the derivative information hyperconvexity metric factor and the classification loss function value as a loss function value to train the first convolutional neural network model and the second convolutional neural network model; and, an inference phase comprising: acquiring a vibration signal and an infrared image of mechanical equipment to be detected through a vibration sensor and an infrared camera; converting the vibration signal into a time-frequency image with the same size as the infrared image by a variational modal decomposition and Hilbert transform method; passing the time-frequency image through the first convolution neural network model which is trained in a training phase and uses a channel attention mechanism to obtain a first feature map; passing the infrared image through a second convolutional neural network which is trained by a training stage and uses a spatial attention mechanism to obtain a second feature map; fusing the first feature map and the second feature map to obtain a classification feature map; and the classification characteristic diagram is processed by a classifier to obtain a classification result, and the classification result is used for indicating whether the mechanical equipment to be detected has faults or not.
Fig. 1 illustrates a scene schematic diagram of a fault detection method based on a multiple attention mechanism according to an embodiment of the present application. As shown in fig. 1, in this application scenario, in a training phase, first, a vibration signal and an infrared image of a rotating mechanical device (e.g., T as illustrated in fig. 1) are acquired by a vibration sensor (e.g., V as illustrated in fig. 1) and an infrared camera (e.g., C as illustrated in fig. 1) disposed at the mechanical device. Then, the vibration signal and the infrared image of the mechanical device are input into a server (e.g., S as illustrated in fig. 1) deployed with a fault detection algorithm based on a multiple attention mechanism, wherein the server is capable of training the first convolutional neural network model and the second convolutional neural network model of the fault detection method based on the multiple attention mechanism with the vibration signal and the infrared image of the mechanical device based on the fault detection algorithm based on the multiple attention mechanism.
After training is completed, in the inference phase, first, a vibration signal and an infrared image of a rotating mechanical device (e.g., T as illustrated in fig. 1) are acquired by a vibration sensor (e.g., V as illustrated in fig. 1) and an infrared camera (e.g., C as illustrated in fig. 1) disposed at the mechanical device. Then, the vibration signal and the infrared image of the mechanical device are input into a server (for example, S as illustrated in fig. 1) deployed with a fault detection algorithm based on a multiple attention mechanism, wherein the server can process the vibration signal and the infrared image of the mechanical device by the fault detection algorithm based on the multiple attention mechanism to generate a classification result for indicating whether the mechanical device to be detected has a fault.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary method
FIG. 2A illustrates a flow diagram of a training phase in a fault detection method based on a multiple attention mechanism according to an embodiment of the present application. FIG. 2B illustrates a flow diagram of an inference stage in a multiple attention mechanism based fault detection method according to an embodiment of the application. As shown in fig. 2A, a fault detection method based on a multiple attention mechanism according to an embodiment of the present application includes: a training phase comprising: s110, acquiring a vibration signal and an infrared image of mechanical equipment through a vibration sensor and an infrared camera; s120, converting the vibration signal into a time-frequency image with the same size as the infrared image through a variational modal decomposition and Hilbert transform method; s130, enabling the time-frequency image to pass through a first convolution neural network model using a channel attention mechanism to obtain a first feature map; s140, passing the infrared image through a second convolutional neural network using a spatial attention mechanism to obtain a second feature map; s150, calculating a derivative information hyperconvexity metric between the first feature map and the second feature map, where the derivative information hyperconvexity metric is a logarithmic function value of a weighted sum of distances between feature values of respective positions in the first feature map and corresponding feature values of respective positions in the second feature map; s160, fusing the first feature map and the second feature map to obtain a classification feature map; s170, enabling the classification characteristic map to pass through a classifier to obtain a classification loss function value; s180, calculating a weighted sum between the derivative information hyperconvexity metric factor and the classification loss function value as a loss function value, and training the first convolutional neural network model and the second convolutional neural network model.
As shown in fig. 2B, the method for detecting a fault based on multiple attention mechanisms according to the embodiment of the present application further includes: an inference phase comprising: s210, acquiring a vibration signal and an infrared image of mechanical equipment to be detected through a vibration sensor and an infrared camera; s220, converting the vibration signal into a time-frequency image with the same size as the infrared image through a variation modal decomposition method and a Hilbert transform method; s230, enabling the time-frequency image to pass through the first convolution neural network model which is trained in the training stage and uses the channel attention mechanism to obtain a first feature map; s240, passing the infrared image through the second convolutional neural network which is trained in the training stage and uses the spatial attention mechanism to obtain a second feature map; s250, fusing the first feature map and the second feature map to obtain a classification feature map; and S260, passing the classification characteristic diagram through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the mechanical equipment to be detected has faults or not.
Fig. 3A illustrates an architecture diagram of a training phase in a multiple attention mechanism-based fault detection method according to an embodiment of the present application. As shown in fig. 3A, in the training phase, first, in the network architecture, the obtained vibration signal (e.g., P1 as illustrated in fig. 3) is converted into a time-frequency image (e.g., P3 as illustrated in fig. 3) having the same size as the infrared image (e.g., P2 as illustrated in fig. 3) by a variational modal decomposition and hilbert transform method; then, passing the time-frequency image through a first convolution neural network model (e.g., CNN1 as illustrated in fig. 3) using a channel attention mechanism to obtain a first feature map (e.g., F1 as illustrated in fig. 3); then, the infrared image is passed through a second convolutional neural network (e.g., CNN2 as illustrated in fig. 3) using a spatial attention mechanism to obtain a second feature map (e.g., F2 as illustrated in fig. 3); then, calculating a derivative information hyper-convexity metric factor (e.g., SML as illustrated in fig. 3) between the first feature map and the second feature map; then, fusing the first feature map and the second feature map to obtain a classification feature map (e.g., FC as illustrated in fig. 3); then, passing the classification feature map through a classifier (e.g., a classifier as illustrated in fig. 3) to obtain a classification loss function value (e.g., a CLV as illustrated in fig. 3); finally, the first convolutional neural network model and the second convolutional neural network model are trained as loss function values (e.g., LV as illustrated in fig. 3) by computing a weighted sum between the derived information hyperconvexity measure factor and the classification loss function values.
Fig. 3B illustrates an architectural diagram of an inference phase in a fault detection method based on a multiple attention mechanism according to an embodiment of the application. As shown in fig. 3B, in the inference phase, first, in the network structure, the obtained vibration signal (e.g., P1 as illustrated in fig. 3) is converted into a time-frequency image (e.g., P3 as illustrated in fig. 3) having the same size as the infrared image (e.g., P2 as illustrated in fig. 3) by a variation modal decomposition and hilbert transform method; then, passing the time-frequency image through the first convolution neural network model (e.g., CN1 as illustrated in fig. 3) using channel attention mechanism after training by a training phase to obtain a first feature map (e.g., F1 as illustrated in fig. 3); then, passing the infrared image through a second convolutional neural network (e.g., CN2 as illustrated in fig. 3) using a spatial attention mechanism after training by a training phase to obtain a second feature map (e.g., F2 as illustrated in fig. 3); then, fusing the first feature map and the second feature map to obtain a classification feature map (e.g., FC as illustrated in fig. 3); and finally, passing the classification characteristic diagram through a classifier (for example, a classifier as illustrated in fig. 3) to obtain a classification result, wherein the classification result is used for indicating whether the mechanical equipment to be detected has a fault or not.
More specifically, in the training phase, in steps S110 and S120, a vibration signal and an infrared image of the mechanical device are acquired through a vibration sensor and an infrared camera, and the vibration signal is converted into a time-frequency image having the same size as the infrared image through a variational modal decomposition and a hilbert transform method. As described above, in the technical solution of the present application, it is desirable to use a two-stage classification model to perform fault detection and early warning of a rotating mechanical device. In particular, in consideration of the characteristics of the weakness, nonlinearity, coupling and the like of the source information of the rotating mechanical equipment, a technical route of multi-source information feature fusion is further adopted to extract classification features for fault detection. In the embodiment of the application, the multi-source information comprises an infrared image of the rotating mechanical equipment and a vibration signal of the rotating mechanical equipment when the rotating mechanical equipment vibrates. It should be noted that the infrared image is two-dimensional non-structural data, has a temperature-sensitive characteristic, and can well identify single faults, coupling faults, fault positions, severity and the like with temperature rise changes.
That is, specifically, in the technical solution of the present application, a vibration signal and an infrared image of a mechanical device are first acquired by a vibration sensor and an infrared camera disposed in the rotating mechanical device. Then, the vibration signals are processed into time-frequency images with the same dimension as the infrared images in a data source domain by adopting a Variational Modal Decomposition (VMD) method and a Hilbert Transform (HT) method. That is, the vibration signal is data-isomerized in the data-source domain to convert the vibration signal into a data representation having the same data structure as the infrared image. Meanwhile, the time-frequency image can represent the time-domain characteristics, the frequency-domain characteristics and the association between the time-domain characteristics and the frequency-domain characteristics of the vibration of the rotating mechanical equipment.
Specifically, in this embodiment of the present application, a process of converting the vibration signal into a time-frequency image having the same size as the infrared image by using a variational modal decomposition and hilbert transform method includes: firstly, the vibration signal and the infrared image are subjected to variation modal decomposition to be converted into a linear stable signal. Then, the linearly stationary signal is subjected to hilbert transform to obtain time-frequency images having the same size.
More specifically, in the training phase, in step S130, the time-frequency image is passed through a first convolutional neural network model using a channel attention mechanism to obtain a first feature map. That is, in the technical solution of the present application, the time-frequency image is then encoded using a convolutional neural network model to extract implicit correlation features of the vibration of the rotating mechanical device in a time dimension and a space dimension. However, considering that if the rotating mechanical device fails, it may exhibit a specific vibration mode, such as high-frequency vibration, on the vibration signal, in the technical solution of the present application, the convolutional neural network model further integrates a channel attention mechanism to pay attention to a specific local time-frequency implicit correlation feature in the time-frequency image suitable for fault detection.
Specifically, in this embodiment of the present application, a process of obtaining a first feature map from the time-frequency image through a first convolution neural network model using a channel attention mechanism includes: each layer of the first convolutional neural network model performs input data in forward transmission of a layer: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram; pooling the convolved feature map to generate a pooled feature map; performing activation processing on the pooled feature map to generate an activated feature map; calculating a global mean of each feature matrix along a channel dimension of the activation feature map to obtain a channel feature vector; calculating the ratio of the feature value of each position in the channel feature vector to the weighted sum of the feature values of all the positions of the channel feature vector to obtain a channel weighted feature vector; performing point multiplication on a feature matrix of the activation feature map along the channel dimension by taking the feature value of each position of the channel weighted feature vector as a weight to obtain a generated feature map; wherein the generated feature map output by the last layer of the first convolutional neural network model is the first feature map.
More specifically, in the training phase, in step S140, the infrared image is passed through a second convolutional neural network using a spatial attention mechanism to obtain a second feature map. It should be understood that the infrared image itself is considered to contain abundant image information, that is, the infrared image information can capture the temperature change of each position of the rotating mechanical device and the mode correlation between the temperatures of each position. Correspondingly, in the technical solution of the present application, the infrared image is further subjected to a second convolutional neural network, so that the second convolutional neural network performs spatial convolutional coding on the infrared image to extract a spatial distribution implicit correlation characteristic of the temperature variation of the rotating mechanical device. In order to enable the second convolutional neural network to focus more on a temperature implicit correlation mode capable of characterizing a fault, in the technical solution of the present application, a spatial attention mechanism is further integrated into the second convolutional neural network model.
Specifically, in this embodiment of the present application, the process of passing the infrared image through a second convolutional neural network using a spatial attention mechanism to obtain a second feature map includes: each layer of the second convolutional neural network model performs, in forward pass of the layer, on input data: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram; pooling the convolved feature map to generate a pooled feature map; performing activation processing on the pooled feature map to generate an activated feature map; performing global average pooling of the activation signature along a channel dimension to obtain a spatial signature matrix; performing convolution processing and activation processing on the spatial feature matrix to generate a weight vector; weighting each feature matrix of the activation feature map by using the weight value of each position in the weight vector to obtain a generated feature map; wherein the generated feature map output by the last layer of the second convolutional neural network model is the second feature map.
More specifically, in the training phase, in steps S150 and S160, a derivative information hyper-convexity metric between the first feature map and the second feature map is calculated, the derivative information hyper-convexity metric being a logarithmic function value of a weighted sum of distances between feature values of respective positions in the first feature map and corresponding feature values of respective positions in the second feature map. It should be understood that, in the technical solution of the present application, further, after the first feature map and the second feature map are fused to obtain a classification feature map, a classifier may be used for classification. However, considering that the feature distribution characteristics of the time domain image and the infrared image are based on, in the technical solution of the present application, feature extraction is performed by using the first convolutional neural network with channel attention and the second convolutional neural network with spatial attention, and since feature extraction is performed by using the first convolutional neural network and the second convolutional neural network respectively in different dimensions (channel dimension and spatial dimension), the obtained feature distribution of the feature map can be converged in different dimensions, so that sparsity between dimensions is generated during fusion, and the quality of the fused feature map is affected.
Therefore, in the technical solution of the present application, the first feature map F is referred to 1 And said second characteristic diagram F 2 And training the first and second convolutional neural networks respectively based on channel attention and space attention by calculating derivative information hyper-convexity coefficients between the feature maps as a loss function, thereby improving the density of feature distribution of the fused feature maps in a high-dimensional feature space.
Specifically, in this embodiment of the present application, the process of calculating the derivative information hyperconvexity factor between the first feature map and the second feature map includes: calculating the derivative information hyperconvexity measure factor between the first feature map and the second feature map in the following formula;
wherein the formula is:
wherein
Feature values representing respective positions in the first feature map,
a feature value indicating each position in the second feature map, | · | indicates a distance between the first feature map and the second feature map.
It should be understood that the derived information hyper-convex metric loss function is an information metric between the internal element sub-dimensions of the feature map in the high-dimensional space, and can be used for carrying out the derived representation of the hyper-convex consistency of the manifold, so that the first feature map F can be enabled by training the model with the loss function 1 And said second characteristic diagram F 2 The inter-manifold difference can adapt to the convex monotonicity on the projection of each sub-dimension, thereby improving the classification effect of the fused feature map.
More specifically, in the training phase, in steps S160, S170 and S180, the first feature map and the second feature map are fused to obtain a classification feature map, the classification feature map is passed through a classifier to obtain a classification loss function value, and a weighted sum between the derived information hyperconvexity parameter and the classification loss function value is calculated as a loss function value to train the first convolutional neural network model and the second convolutional neural network model. That is, in the technical solution of the present application, the first feature map and the second feature map are further fused to perform classification processing, so as to obtain a classification loss function value, so that the first convolutional neural network model and the second convolutional neural network model can be trained as a loss function value based on a weighted sum between the derivative information hyper-convexity metric factor and the classification loss function value. It should be understood that in this way, the manifold difference between the feature maps can be adapted to the convex monotonicity on the projection of each sub-dimension, so as to improve the precision and accuracy of fault classification.
Specifically, the process of passing the classification feature map through a classifier to obtain a classification loss function value includes: first, the classifier is used to pair the following formulasThe classification feature map is processed to generate a classification result, wherein the formula is as follows: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) L Project (F) }, where Project (F) denotes the projection of the classification feature map as a vector, W 1 To W n As a weight matrix for all connected layers of each layer, B 1 To B n A bias matrix representing the fully connected layers of each layer. Then, a cross entropy value between the classification result and a true value is calculated as the classification loss function value.
After training is completed, the inference phase is entered. That is, after the training of the first convolutional neural network model and the second convolutional neural network model is completed in the training stage, the obtained first convolutional neural network model and the second convolutional neural network model are used in the actual inference stage. And then, repeating the steps to obtain the classification characteristic diagram so as to obtain a classification result for indicating whether the mechanical equipment to be detected has faults or not through the classifier.
In summary, a fault detection method based on a multi-attention mechanism is illustrated in an embodiment of the present application, which processes a vibration signal of a rotating mechanical device into a time-frequency image in the same dimension as an infrared image by using a variational modal decomposition and a hilbert transform method in a data source domain, and encodes the time-frequency image and the infrared image by using a convolutional neural network model with a channel attention mechanism and a spatial attention mechanism, respectively, to extract a specific local time-frequency implicit association feature suitable for fault detection in the time-frequency image and to focus on a temperature implicit association pattern capable of characterizing a fault, and further calculates a derivative information hyperconvexity parameter factor between the feature maps when the two features are fused, so that a manifold difference between the feature maps can adapt to a convex monotonicity on projections of sub-dimensions, so as to improve precision and accuracy of fault classification.
Exemplary System
FIG. 4 illustrates a block diagram of a multiple attention mechanism based fault detection system in accordance with an embodiment of the present application. As shown in fig. 4, a fault detection system 400 based on a multiple attention mechanism according to an embodiment of the present application includes: a training module 410 and an inference module 420.
As shown in fig. 4, the training module 410 includes: a data acquisition unit 411 for acquiring a vibration signal and an infrared image of the mechanical device through the vibration sensor and the infrared camera; a time-frequency image conversion unit 412, configured to convert the vibration signal obtained by the data obtaining unit 411 into a time-frequency image having the same size as the infrared image obtained by the data obtaining unit 411 by using a variational modal decomposition and hilbert transform method; a channel attention extracting unit 413, configured to pass the time-frequency image obtained by the time-frequency image converting unit 412 through a first convolutional neural network model using a channel attention mechanism to obtain a first feature map; a spatial attention extracting unit 414, configured to pass the infrared image obtained by the data obtaining unit 411 through a second convolutional neural network using a spatial attention mechanism to obtain a second feature map; a derived information hyperconvexity degree factor calculation unit 415 configured to calculate a derived information hyperconvexity degree factor between the first feature map obtained by the channel attention extraction unit 413 and the second feature map obtained by the spatial attention extraction unit 414, the derived information hyperconvexity degree factor being a logarithmic function value of a weighted sum of distances between feature values of respective positions in the first feature map and corresponding respective positions in the second feature map; a fusion unit 416, configured to fuse the first feature map obtained by the channel attention extraction unit 413 and the second feature map obtained by the spatial attention extraction unit 414 to obtain a classification feature map; a classification loss function value calculation unit 417, configured to pass the classification feature map obtained by the fusion unit 416 through a classifier to obtain a classification loss function value; a training unit 418 configured to calculate a weighted sum between the derivative information hyperconvexity degree factor obtained by the derivative information hyperconvexity degree factor calculation unit 415 and the classification loss function value obtained by the classification loss function value calculation unit 417 to train the first convolutional neural network model and the second convolutional neural network model as a loss function value.
As shown in fig. 4, the inference module 420 includes: the inferred data acquisition unit 421 is used for acquiring a vibration signal and an infrared image of the mechanical equipment to be detected through the vibration sensor and the infrared camera; a conversion unit 422, configured to convert the vibration signal obtained by the inferred data obtaining unit 421 into a time-frequency image having the same size as the infrared image obtained by the inferred data obtaining unit 421 through a variational modal decomposition and hilbert transform method; a first convolution unit 423, configured to pass the time-frequency image obtained by the conversion unit 422 through the first convolution neural network model using channel attention mechanism after training in the training phase to obtain a first feature map; a second convolution unit 424, configured to pass the infrared image obtained by the inferred data obtaining unit 421 through the second convolution neural network using the spatial attention mechanism after training in the training phase to obtain a second feature map; a classification feature map generating unit 425 configured to fuse the first feature map obtained by the first convolution unit 423 and the second feature map obtained by the second convolution unit 424 to obtain a classification feature map; and a classification unit 426, configured to pass the classification feature map obtained by the classification feature map generation unit 425 through a classifier to obtain a classification result, where the classification result is used to indicate whether the mechanical equipment to be detected has a fault.
Here, it can be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described multiple attention mechanism-based fault detection system 400 have been described in detail in the above description of the multiple attention mechanism-based fault detection method with reference to fig. 1 to 3B, and thus, a repetitive description thereof will be omitted.
As described above, the multiple attention mechanism-based failure detection system 400 according to the embodiment of the present application may be implemented in various terminal devices, such as a server based on a multiple attention mechanism failure detection algorithm, and the like. In one example, the multiple attention mechanism based fault detection system 400 according to embodiments of the present application may be integrated into a terminal device as one software module and/or hardware module. For example, the multiple attention mechanism based fault detection system 400 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the multiple attention mechanism based fault detection system 400 may also be one of many hardware modules of the terminal device.
Alternatively, in another example, the multiple attention system based fault detection system 400 and the terminal device may also be separate devices, and the multiple attention system based fault detection system 400 may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information in an agreed data format.
The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the foregoing disclosure is not intended to be exhaustive or to limit the disclosure to the precise details disclosed.
The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by one skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.
Claims (10)
1. A fault detection method based on a multiple attention mechanism is characterized by comprising the following steps:
a training phase comprising:
acquiring a vibration signal and an infrared image of mechanical equipment through a vibration sensor and an infrared camera;
converting the vibration signal into a time-frequency image with the same size as the infrared image by a variational modal decomposition and Hilbert transform method;
passing the time-frequency image through a first convolution neural network model using a channel attention mechanism to obtain a first feature map;
passing the infrared image through a second convolutional neural network using a spatial attention mechanism to obtain a second feature map;
calculating a derivative information hyperconvexity measurement factor between the first feature map and the second feature map, wherein the derivative information hyperconvexity measurement factor is a logarithmic function value of a weighted sum of distances between feature values of each position in the first feature map and corresponding feature values of each position in the second feature map;
fusing the first feature map and the second feature map to obtain a classification feature map;
passing the classification feature map through a classifier to obtain a classification loss function value;
calculating a weighted sum between the derivative information hyperconvexity metric factor and the classification loss function value as a loss function value to train the first convolutional neural network model and the second convolutional neural network model; and
an inference phase comprising:
acquiring a vibration signal and an infrared image of mechanical equipment to be detected through a vibration sensor and an infrared camera;
converting the vibration signal into a time-frequency image with the same size as the infrared image by a variational modal decomposition and Hilbert transform method;
passing the time-frequency image through the first convolution neural network model which is trained in a training stage and uses a channel attention mechanism to obtain a first feature map;
passing the infrared image through a second convolutional neural network which is trained by a training stage and uses a spatial attention mechanism to obtain a second feature map;
fusing the first feature map and the second feature map to obtain a classification feature map; and
and passing the classification characteristic diagram through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the mechanical equipment to be detected has faults or not. .
2. The multiple attention mechanism based fault detection method of claim 1, wherein converting the vibration signal into a time-frequency image having the same size as the infrared image by a variational modal decomposition and hilbert transform method comprises:
carrying out variation modal decomposition on the vibration signal and the infrared image so as to convert the vibration signal and the infrared image into linear stationary signals;
and performing Hilbert transform on the linear stationary signals to obtain time-frequency images with the same size.
3. The multiple attention mechanism-based fault detection method of claim 2, wherein passing the time-frequency image through a first convolutional neural network model using a channel attention mechanism to obtain a first feature map comprises: each layer of the first convolutional neural network model performs input data in forward transfer of the layer:
performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram;
pooling the convolution feature map to generate a pooled feature map;
performing activation processing on the pooled feature map to generate an activated feature map;
calculating a global mean of each feature matrix of the activation feature map along a channel dimension to obtain a channel feature vector;
calculating the ratio of the feature value of each position in the channel feature vector to the weighted sum of the feature values of all positions of the channel feature vector to obtain a channel weighted feature vector; and
performing point multiplication on a feature matrix of the activation feature map along a channel dimension by taking the feature value of each position of the channel weighted feature vector as a weight to obtain a generated feature map;
wherein the generated feature map output by the last layer of the first convolutional neural network model is the first feature map.
4. The multiple attention mechanism based fault detection method of claim 2, wherein passing the infrared image through a second convolutional neural network using a spatial attention mechanism to obtain a second signature comprises: each layer of the second convolutional neural network model performs, in forward pass of the layer, on input data:
performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram;
pooling the convolution feature map to generate a pooled feature map;
performing activation processing on the pooled feature map to generate an activated feature map;
performing global average pooling of the activation signature along a channel dimension to obtain a spatial signature matrix;
performing convolution processing and activation processing on the spatial feature matrix to generate a weight vector; and
weighting each feature matrix of the activation feature map by the weight value of each position in the weight vector to obtain a generated feature map;
wherein the generated feature map output by the last layer of the second convolutional neural network model is the second feature map.
5. The multiple attention mechanism-based fault detection method of claim 4, wherein calculating a derivative information hyper-convexity metric between the first feature map and the second feature map comprises:
calculating the derivative information hyperconvexity measure factor between the first feature map and the second feature map in the following formula;
wherein the formula is:
wherein
Feature values representing respective positions in the first feature map,
a feature value indicating each position in the second feature map, | · | indicates a distance between the first feature map and the second feature map.
6. The multi-attention mechanism-based fault detection method of claim 5, wherein passing the classification feature map through a classifier to obtain classification loss function values comprises:
the classifier processes the classification feature map to generate a classification result according to the following formula, wherein the formula is as follows: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) L Project (F) }, where Project (F) denotes the projection of the classification feature map as a vector, W 1 To W n As a weight matrix for each fully connected layer, B 1 To B n A bias matrix representing the layers of the fully-connected layer; and
and calculating a cross entropy value between the classification result and the real value as the classification loss function value.
7. A multiple attention mechanism based fault detection system, comprising:
a training module comprising:
the data acquisition unit is used for acquiring a vibration signal and an infrared image of the mechanical equipment through the vibration sensor and the infrared camera;
the time-frequency image conversion unit is used for converting the vibration signals obtained by the data acquisition unit into time-frequency images with the same size as the infrared images obtained by the data acquisition unit through a variation modal decomposition method and a Hilbert transform method;
the channel attention extracting unit is used for enabling the time-frequency image obtained by the time-frequency image converting unit to pass through a first convolution neural network model using a channel attention mechanism so as to obtain a first feature map;
a spatial attention extracting unit for passing the infrared image obtained by the data obtaining unit through a second convolutional neural network using a spatial attention mechanism to obtain a second feature map;
a derived information hyperconvexity factor calculation unit configured to calculate a derived information hyperconvexity factor between the first feature map obtained by the channel attention extraction unit and the second feature map obtained by the spatial attention extraction unit, where the derived information hyperconvexity factor is a logarithmic function value of a weighted sum of distances between feature values of respective positions in the first feature map and corresponding respective positions in the second feature map;
a fusion unit configured to fuse the first feature map obtained by the channel attention extraction unit and the second feature map obtained by the spatial attention extraction unit to obtain a classification feature map;
the classification loss function value calculation unit is used for enabling the classification feature map obtained by the fusion unit to pass through a classifier so as to obtain a classification loss function value;
a training unit configured to calculate a weighted sum between the derivative information hyperconvexity degree factor obtained by the derivative information hyperconvexity degree factor calculation unit and the classification loss function value obtained by the classification loss function value calculation unit, and train the first convolutional neural network model and the second convolutional neural network model as a loss function value; and
an inference module comprising:
the inferred data acquisition unit is used for acquiring a vibration signal and an infrared image of the mechanical equipment to be detected through the vibration sensor and the infrared camera;
the conversion unit is used for converting the vibration signals obtained by the inferred data acquisition unit into time-frequency images with the same size as the infrared images obtained by the inferred data acquisition unit through a variation modal decomposition method and a Hilbert transform method;
the first convolution unit is used for enabling the time-frequency image obtained by the conversion unit to pass through the first convolution neural network model which is trained in the training stage and uses the channel attention mechanism so as to obtain a first feature map;
a second convolution unit, configured to pass the infrared image obtained by the inferred data obtaining unit through a second convolution neural network using a spatial attention mechanism that is trained by a training phase to obtain a second feature map;
a classification feature map generating unit, configured to fuse the first feature map obtained by the first convolution unit and the second feature map obtained by the second convolution unit to obtain a classification feature map; and
and the classification unit is used for enabling the classification characteristic diagram obtained by the classification characteristic diagram generation unit to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether the mechanical equipment to be detected has faults or not.
8. The multi-attention mechanism-based fault detection method of claim 7, wherein the channel attention extraction unit is further configured to: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram; pooling the convolved feature map to generate a pooled feature map; performing activation processing on the pooled feature map to generate an activated feature map; calculating a global mean of each feature matrix along a channel dimension of the activation feature map to obtain a channel feature vector; calculating the ratio of the feature value of each position in the channel feature vector to the weighted sum of the feature values of all the positions of the channel feature vector to obtain a channel weighted feature vector; performing point multiplication on a feature matrix of the activation feature map along the channel dimension by taking the feature value of each position of the channel weighted feature vector as a weight to obtain a generated feature map; wherein the generated feature map output by the last layer of the first convolutional neural network model is the first feature map.
9. The multiple attention mechanism based fault detection method of claim 8, wherein the spatial attention extraction unit is further configured to: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram; pooling the convolution feature map to generate a pooled feature map; performing activation processing on the pooled feature map to generate an activated feature map; performing global average pooling along a channel dimension on the activation feature map to obtain a spatial feature matrix; performing convolution processing and activation processing on the spatial feature matrix to generate a weight vector; weighting each feature matrix of the activation feature map by using the weight value of each position in the weight vector to obtain a generated feature map; wherein the generated feature map output by the last layer of the second convolutional neural network model is the second feature map.
10. The multiple attention mechanism-based fault detection method of claim 9, wherein the derived information hyperconvexity factor calculation unit is further configured to: calculating the derivative information hyperconvexity measure factor between the first feature map and the second feature map in the following formula;
wherein the formula is:
wherein
Feature values representing respective positions in the first feature map,
a feature value representing each position in the second feature map, | · | represents a distance between the first feature map and the second feature map.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211028528.0A CN115410069A (en) | 2022-08-25 | 2022-08-25 | Fault detection method and system based on multiple attention mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211028528.0A CN115410069A (en) | 2022-08-25 | 2022-08-25 | Fault detection method and system based on multiple attention mechanism |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115410069A true CN115410069A (en) | 2022-11-29 |
Family
ID=84161406
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211028528.0A Withdrawn CN115410069A (en) | 2022-08-25 | 2022-08-25 | Fault detection method and system based on multiple attention mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115410069A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116050807A (en) * | 2023-03-30 | 2023-05-02 | 东莞正大康地饲料有限公司 | Feed production equipment defect identification method and system based on video monitoring inspection |
| CN116580444A (en) * | 2023-07-14 | 2023-08-11 | 广州思林杰科技股份有限公司 | Method and equipment for testing long-distance running timing based on multi-antenna radio frequency identification technology |
| CN116934304A (en) * | 2023-08-03 | 2023-10-24 | 广东省俊耀控股有限公司 | Intelligent power distribution room equipment operation maintenance management system and method thereof |
| CN117197029A (en) * | 2023-04-28 | 2023-12-08 | 浙江工交机械股份有限公司 | Automatic production line and method for ball pins |
| CN117710166A (en) * | 2024-02-06 | 2024-03-15 | 吉林大学 | Teaching guidance detection system and method for pediatric neural nursing |
-
2022
- 2022-08-25 CN CN202211028528.0A patent/CN115410069A/en not_active Withdrawn
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116050807A (en) * | 2023-03-30 | 2023-05-02 | 东莞正大康地饲料有限公司 | Feed production equipment defect identification method and system based on video monitoring inspection |
| CN116050807B (en) * | 2023-03-30 | 2023-06-23 | 东莞正大康地饲料有限公司 | Feed production equipment defect identification method and system based on video monitoring inspection |
| CN117197029A (en) * | 2023-04-28 | 2023-12-08 | 浙江工交机械股份有限公司 | Automatic production line and method for ball pins |
| CN117197029B (en) * | 2023-04-28 | 2024-05-03 | 浙江工交机械股份有限公司 | Automatic production line and method for ball pins |
| CN116580444A (en) * | 2023-07-14 | 2023-08-11 | 广州思林杰科技股份有限公司 | Method and equipment for testing long-distance running timing based on multi-antenna radio frequency identification technology |
| CN116934304A (en) * | 2023-08-03 | 2023-10-24 | 广东省俊耀控股有限公司 | Intelligent power distribution room equipment operation maintenance management system and method thereof |
| CN116934304B (en) * | 2023-08-03 | 2024-06-07 | 广东省俊耀控股有限公司 | Intelligent power distribution room equipment operation maintenance management system and method thereof |
| CN117710166A (en) * | 2024-02-06 | 2024-03-15 | 吉林大学 | Teaching guidance detection system and method for pediatric neural nursing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115410069A (en) | Fault detection method and system based on multiple attention mechanism | |
| US11715190B2 (en) | Inspection system, image discrimination system, discrimination system, discriminator generation system, and learning data generation device | |
| CN115235612A (en) | Intelligent fault diagnosis system and method for servo motor | |
| CN115013298B (en) | Real-time performance online monitoring system and monitoring method of sewage pump | |
| Yang et al. | Spatio-temporal domain awareness for multi-agent collaborative perception | |
| JPWO2021049005A1 (en) | Information processing device and electronic equipment equipped with it | |
| CN116030097B (en) | Target tracking method and system based on dual-attention feature fusion network | |
| CN115424204A (en) | Pedestrian detection method and system based on information fusion | |
| CN115601318B (en) | Intelligent production method and system for quick-absorption low-reverse-osmosis paper diaper | |
| CN115577248A (en) | Fault diagnosis system and method for wind generating set | |
| CN114580515A (en) | Neural network training method for intelligent detection of semiconductor desoldering | |
| CN115526202A (en) | Offshore wind turbine fault diagnosis system based on data driving and diagnosis method thereof | |
| CN114800229A (en) | Double-surface double-glass surface polishing device and polishing method thereof | |
| CN115146676A (en) | Circuit fault detection method and system | |
| CN114821169A (en) | Method-level non-intrusive call link tracking method under micro-service architecture | |
| CN117131348B (en) | Data quality analysis method and system based on differential convolution characteristics | |
| CN115762558B (en) | Performance detection system and method for escalator production | |
| CN116402777B (en) | Power equipment detection method and system based on machine vision | |
| CN116584951B (en) | Electrocardiosignal detection and positioning method based on weak supervision learning | |
| CN112800217A (en) | Vector relevance matrix-based intelligent assessment method for government affair transaction processing report | |
| CN115143128B (en) | Fault diagnosis method and system for small-sized submersible electric pump | |
| CN114839466B (en) | EMC electromagnetic compatibility test system for water pump and test method thereof | |
| CN116204821A (en) | Vibration evaluation method and system for rail transit vehicle | |
| CN116017257A (en) | Intelligent production method and system for loudspeaker | |
| CN115456259A (en) | Wind power plant site selection optimization system and method based on mesoscale data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20221129 |
|
| WW01 | Invention patent application withdrawn after publication |