CN117225921B - Automatic control system and method for extrusion of aluminum alloy profile - Google Patents
Automatic control system and method for extrusion of aluminum alloy profile Download PDFInfo
- Publication number
- CN117225921B CN117225921B CN202311250149.0A CN202311250149A CN117225921B CN 117225921 B CN117225921 B CN 117225921B CN 202311250149 A CN202311250149 A CN 202311250149A CN 117225921 B CN117225921 B CN 117225921B
- Authority
- CN
- China
- Prior art keywords
- aluminum alloy
- heat distribution
- image
- semantic
- reconstructed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 229910000838 Al alloy Inorganic materials 0.000 title claims abstract description 205
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000001125 extrusion Methods 0.000 title claims description 96
- 238000010438 heat treatment Methods 0.000 claims abstract description 41
- 238000005520 cutting process Methods 0.000 claims abstract description 19
- 239000000956 alloy Substances 0.000 claims abstract description 17
- 238000001816 cooling Methods 0.000 claims abstract description 7
- 238000007493 shaping process Methods 0.000 claims abstract description 7
- 238000009826 distribution Methods 0.000 claims description 184
- 239000013598 vector Substances 0.000 claims description 32
- 239000011159 matrix material Substances 0.000 claims description 27
- 238000013527 convolutional neural network Methods 0.000 claims description 20
- 238000010586 diagram Methods 0.000 claims description 18
- 238000005457 optimization Methods 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 11
- 238000007781 pre-processing Methods 0.000 claims description 10
- 230000003247 decreasing effect Effects 0.000 claims description 9
- 238000011217 control strategy Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 5
- 238000000465 moulding Methods 0.000 claims description 2
- 238000003062 neural network model Methods 0.000 claims 1
- 230000001902 propagating effect Effects 0.000 claims 1
- 230000008569 process Effects 0.000 abstract description 19
- 230000006870 function Effects 0.000 description 15
- 238000012545 processing Methods 0.000 description 10
- 238000004140 cleaning Methods 0.000 description 7
- 230000007246 mechanism Effects 0.000 description 6
- 238000011176 pooling Methods 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 4
- 239000000284 extract Substances 0.000 description 4
- 238000003825 pressing Methods 0.000 description 4
- 238000013136 deep learning model Methods 0.000 description 3
- 238000010030 laminating Methods 0.000 description 3
- 238000007790 scraping Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000010485 coping Effects 0.000 description 2
- 238000007723 die pressing method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000002441 reversible effect Effects 0.000 description 2
- 238000001931 thermography Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000001192 hot extrusion Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000003678 scratch resistant effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000003466 welding Methods 0.000 description 1
Abstract
An automatic control system for extruding aluminium alloy section and its method are disclosed. Firstly cutting an aluminum alloy blank to obtain a cut aluminum alloy blank, then heating the cut aluminum alloy blank to obtain a heated aluminum alloy blank, then extruding the heated aluminum alloy blank through an extruder to obtain an extruded aluminum alloy material conforming to a preset cross-sectional shape, and finally cooling, solidifying, shaping and cutting the extruded aluminum alloy material to obtain the aluminum alloy profile. In this way, the heating control can be intelligently performed to cope with rapidly changing process demands.
Description
Technical Field
The present disclosure relates to the field of automatic control, and more particularly, to an automatic control system for aluminum alloy profile extrusion and a method thereof.
Background
The aluminum alloy section bar is a light high-strength material widely applied to the fields of construction, traffic, aviation and the like, and the performance and the quality of the aluminum alloy section bar are affected by an extrusion process. The extrusion process is a process of extruding and molding an aluminum alloy blank through a die at high temperature and high pressure, and a plurality of parameters and links are involved in the process.
In order to reduce direct contact of operators and reduce occurrence probability of accidents, production efficiency and product stability can be improved, and automatic control of extrusion of aluminum alloy profiles is a technical problem to be solved urgently.
Disclosure of Invention
In view of this, the present disclosure proposes an automatic control system for aluminum alloy profile extrusion and a method thereof, which can intelligently perform heating control, thereby coping with rapidly changing process requirements.
According to an aspect of the present disclosure, there is provided an automatic control method of aluminum alloy profile extrusion, including:
cutting the aluminum alloy blank to obtain a cut aluminum alloy blank;
heating the cut aluminum alloy blank to obtain a heated aluminum alloy blank;
extruding the heated aluminum alloy blank through an extruder to obtain an extruded aluminum alloy material conforming to a preset cross-sectional shape; and
and cooling, solidifying, shaping and cutting the extruded aluminum alloy material to obtain the aluminum alloy profile.
According to another aspect of the present disclosure, there is provided an automatic control system for extrusion of an aluminum alloy profile, including:
the cutting module is used for cutting the aluminum alloy blank to obtain a cut aluminum alloy blank;
the heating treatment module is used for carrying out heating treatment on the cut aluminum alloy blank to obtain a heated aluminum alloy blank;
the extrusion treatment module is used for extruding the heated aluminum alloy blank through an extruder to obtain an extruded aluminum alloy material conforming to the preset cross-sectional shape; and
and the forming module is used for cooling, solidifying, shaping and cutting the extruded aluminum alloy material to obtain the aluminum alloy profile.
According to the embodiment of the disclosure, firstly, an aluminum alloy blank is cut to obtain a cut aluminum alloy blank, then, the cut aluminum alloy blank is subjected to heating treatment to obtain a heated aluminum alloy blank, then, the heated aluminum alloy blank is extruded by an extruder to obtain an extruded aluminum alloy material conforming to a preset cross-sectional shape, and finally, the extruded aluminum alloy material is cooled, solidified, shaped and cut to obtain an aluminum alloy profile. In this way, the heating control can be intelligently performed to cope with rapidly changing process demands.
Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments, which proceeds with reference to the accompanying drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate exemplary embodiments, features and aspects of the present disclosure and together with the description, serve to explain the principles of the disclosure.
Fig. 1 shows a flowchart of an automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure.
Fig. 2 shows a flowchart of substep S120 of an automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure.
Fig. 3 shows a schematic diagram of the architecture of substep S120 of the automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure.
Fig. 4 shows a flowchart of sub-step S122 of the automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure.
Fig. 5 shows a flowchart of sub-step S1221 of an automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure.
Fig. 6 shows a flowchart of substep S124 of the automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure.
Fig. 7 shows a block diagram of an automatic control system for aluminum alloy profile extrusion in accordance with an embodiment of the present disclosure.
Fig. 8 illustrates an application scenario diagram of an automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure.
Detailed Description
The following description of the embodiments of the present disclosure will be made clearly and fully with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some, but not all embodiments of the disclosure. All other embodiments, which can be made by one of ordinary skill in the art without undue burden based on the embodiments of the present disclosure, are also within the scope of the present disclosure.
As used in this disclosure and in the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
Various exemplary embodiments, features and aspects of the disclosure will be described in detail below with reference to the drawings. In the drawings, like reference numbers indicate identical or functionally similar elements. Although various aspects of the embodiments are illustrated in the accompanying drawings, the drawings are not necessarily drawn to scale unless specifically indicated.
In addition, numerous specific details are set forth in the following detailed description in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art have not been described in detail in order not to obscure the present disclosure.
The present disclosure provides an automatic control method of aluminum alloy profile extrusion, and fig. 1 shows a flowchart of an automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure. As shown in fig. 1, the automatic control method for extrusion of an aluminum alloy profile according to an embodiment of the present disclosure includes the steps of: s110, cutting the aluminum alloy blank to obtain a cut aluminum alloy blank; s120, heating the cut aluminum alloy blank to obtain a heated aluminum alloy blank; s130, extruding the heated aluminum alloy blank through an extruder to obtain an extruded aluminum alloy material conforming to a preset cross-sectional shape; and S140, cooling, solidifying, shaping and cutting the extruded aluminum alloy material to obtain the aluminum alloy profile.
In step S120, the aluminum alloy blank after being cut is subjected to heat treatment, so that grains of the aluminum alloy blank can be thinned, plasticity and extrusion performance can be improved, smooth extrusion process is facilitated, and stress concentration and non-uniform deformation in the extrusion process can be reduced. In order to ensure that the cut aluminum alloy billet reaches the proper extrusion temperature, the heating temperature is typically controlled. Currently, it is most common to use PID (proportional, integral, derivative) controllers to monitor the temperature of the heating device and to adjust it according to a preset temperature profile. However, since the temperature profile is set by man, it may have a large limitation in coping with complex nonlinear systems and rapidly changing process requirements. Thus, an optimized solution is desired.
Based on the above, the technical concept of the present disclosure is to collect the heat distribution image of the cut aluminum alloy blank in real time, reconstruct the resolution of the heat distribution image, and then extract and classify the features of the reconstructed image to intelligently control the heating. In this way, the limitation caused by temperature regulation according to a manually preset temperature curve is broken through, so as to intelligently cope with the rapidly-changing process requirement.
Specifically, the technical scheme can be more completely constructed in a processing mode of extracting and classifying image features. However, in the process of actually performing acquisition of the thermal distribution image, it is considered that the thermal imaging apparatus may be affected by noise, and artifacts (such as artifacts caused by reflection or transmission of radiation) occur when acquiring the image, or the thermal imaging apparatus itself may have a problem of non-uniform response, that is, different temperature responses of different regions. Therefore, in the technical scheme of the disclosure, it is expected to reduce noise level and remove artifacts by an image reconstruction method, and meanwhile, perform non-uniformity compensation on a thermal distribution image, eliminate temperature deviation between different areas, and obtain more accurate temperature distribution information.
Specifically, fig. 2 shows a flowchart of substep S120 of the automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure. Fig. 3 shows a schematic diagram of the architecture of substep S120 of the automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure. As shown in fig. 2 and 3, according to an automatic control method for extrusion of an aluminum alloy profile according to an embodiment of the present disclosure, heating the cut aluminum alloy billet to obtain a heated aluminum alloy billet, including: s121, acquiring a heat distribution image of the cut aluminum alloy blank; s122, performing image preprocessing on the heat distribution image of the cut aluminum alloy blank to obtain a reconstructed heat distribution image; s123, extracting image features of the reconstructed heat distribution image to obtain a reconstructed heat distribution feature matrix; and S124, determining a control strategy of the heating temperature value based on the reconstructed heat distribution characteristic matrix.
Accordingly, in the technical scheme of the present disclosure, first, a heat distribution image of a cut aluminum alloy billet is acquired. Then, the heat distribution image of the cut aluminum alloy blank is subjected to image preprocessing to obtain a reconstructed heat distribution image.
In a specific example of the present disclosure, the encoding process for performing image preprocessing on the heat distribution image of the cut aluminum alloy blank to obtain a reconstructed heat distribution image includes: firstly, passing the heat distribution image of the cut aluminum alloy blank through a heat distribution shallow feature extractor based on a first convolutional neural network model to obtain a heat distribution shallow feature map; then the thermal distribution shallow feature map passes through a thermal distribution deep feature extractor based on a second convolutional neural network model to obtain a thermal distribution deep feature map; then, the heat distribution shallow layer feature map and the heat distribution deep layer feature map pass through a joint semantic propagation module to obtain a semantic joint aluminum alloy profile heat distribution feature map; further, the semantic united aluminum alloy section heat distribution characteristic diagram is passed through a resolution reconstructor based on a decoder to obtain a reconstructed heat distribution image.
The method comprises the steps of firstly constructing a thermal distribution shallow feature extractor and a thermal distribution deep feature extractor by using a convolutional neural network model to capture shallow feature information and deep feature information of a thermal distribution image respectively, then carrying out semantic propagation on the thermal distribution shallow feature image and the thermal distribution deep feature image so that the thermal distribution shallow feature image can retain the shallow feature information and can fuse more abstract and semantic expression in the deep feature information, and then constructing a resolution reconstructor through a decoder to generate the reconstructed thermal distribution image. The joint semantic propagation module can extract global semantic information and local semantic information of deep features and propagate the semantic information into shallow features so as to reduce semantic difference between the deep features and the local semantic information.
More specifically, in an embodiment of the present disclosure, a coding process of passing the thermal distribution shallow layer feature map and the thermal distribution deep layer feature map through a joint semantic propagation module to obtain a semantic joint aluminum alloy profile thermal distribution feature map includes: firstly, upsampling the thermal distribution deep feature map to obtain a resolution reconstruction feature map; performing point convolution, batch normalization operation and non-activated function operation based on ReLU on the global average feature vector obtained after global average pooling of the resolution reconstruction feature map to obtain a global semantic vector; meanwhile, performing point convolution, batch normalization operation and non-activated function operation based on ReLU on the resolution reconfiguration feature map to obtain local semantic vectors; then, carrying out point processing on the global semantic vector and the local semantic vector to obtain a semantic weight vector; then, the semantic weight vector is used as a weight vector, and the thermal distribution shallow feature map is weighted to obtain a semantic joint feature map; and then, fusing the heat distribution shallow feature map and the semantic joint feature map to obtain the semantic joint aluminum alloy profile heat distribution feature map.
Accordingly, as shown in fig. 4, performing image preprocessing on the heat distribution image of the cut aluminum alloy blank to obtain a reconstructed heat distribution image, including: s1221, extracting shallow characteristic information and deep characteristic information of a heat distribution image of the cut aluminum alloy blank to obtain a heat distribution shallow characteristic image and a heat distribution deep characteristic image; s1222, carrying out semantic propagation on the heat distribution shallow feature map and the heat distribution deep feature map to obtain a semantic united aluminum alloy profile heat distribution feature map; and S1223, reconstructing the image resolution based on the semantic united aluminum alloy section heat distribution characteristic diagram to obtain the reconstructed heat distribution image.
More specifically, in step S1221, as shown in fig. 5, shallow feature information and deep feature information of the heat distribution image of the cut aluminum alloy blank are extracted to obtain a heat distribution shallow feature map and a heat distribution deep feature map, including: s12211, enabling the heat distribution image of the cut aluminum alloy blank to pass through a heat distribution shallow feature extractor based on a first convolutional neural network model to obtain a heat distribution shallow feature map; and S12212, passing the thermal profile shallow layer feature map through a thermal profile deep layer feature extractor based on a second convolutional neural network model to obtain the thermal profile deep layer feature map.
It should be noted that the convolutional neural network (Convolutional Neural Network, abbreviated as CNN) is a deep learning model, and is mainly used for processing data with a grid structure, such as images and videos. The core idea of the convolutional neural network is to extract the characteristics of input data by using a convolutional layer, reduce the size of a characteristic diagram by using a pooling layer, and perform tasks such as classification or regression by using a full connection layer. The main characteristics of convolutional neural network are parameter sharing and local perception field, which makes it have good processing ability to the spatial structure of input data. Convolutional neural network models are typically composed of multiple convolutional layers, an activation function, a pooling layer, and a fully-connected layer. The convolution layer extracts features on the input data by convolution operations and introduces nonlinearities using an activation function. The pooling layer is used for reducing the size of the feature map, reducing the computational complexity and extracting main features. The fully connected layer maps the characteristics of the pooling layer output to final output categories or regression values. Training of convolutional neural network models is typically performed by a back-propagation algorithm, i.e., calculating gradients from the output layer to the input layer and updating network parameters. In the training process, the convolutional neural network optimizes model parameters by minimizing a loss function so that the convolutional neural network can better fit training data and has good generalization capability on test data. Namely, the convolutional neural network model is a deep learning model specially used for processing data with a grid structure, extracts characteristics through convolutional operation and parameter sharing, and performs tasks such as classification or regression through pooling and full-connection layers.
More specifically, in step S1222, the semantic propagation is performed on the thermal distribution shallow feature map and the thermal distribution deep feature map to obtain a semantic joint aluminum alloy profile thermal distribution feature map, including: and passing the heat distribution shallow layer feature map and the heat distribution deep layer feature map through a joint semantic propagation module to obtain the semantic joint aluminum alloy profile heat distribution feature map. It should be understood that the joint semantic propagation module is used for performing semantic propagation on the heat distribution shallow feature map and the heat distribution deep feature map, so as to obtain a part of the semantic joint aluminum alloy profile heat distribution feature map, and the function of the module is to fuse the feature maps of different layers so as to obtain more comprehensive and accurate semantic information. Typically, deep feature maps contain more abstract and high-level semantic information, while shallow feature maps contain more detailed and local information. The feature graphs of the two layers can be fused through the joint semantic propagation module so as to obtain the feature graph with detail information and advanced semantic information. The particular implementation of the federated semantic propagation module may depend on the particular network architecture and task requirements. One common implementation is to use a skip connection or attention mechanism (attention mechanism). The jump connection can directly connect the shallow feature map and the deep feature map, and the information of the shallow feature map and the deep feature map is fused. The attention mechanism can adaptively adjust the contribution degree of the feature map according to the importance weight of the feature map, so that important features are focused more. Feature information of different layers can be comprehensively utilized through the combined semantic propagation module, so that understanding and predicting capability of the model on heat distribution of the aluminum alloy section bar are improved. This helps to more accurately analyze and predict the heat distribution of the aluminum alloy profile, providing an important reference for subsequent heat management and optimization.
More specifically, in step S1223, performing image resolution reconstruction based on the semantic united aluminum alloy profile heat distribution feature map to obtain the reconstructed heat distribution image, including: and passing the semantic united aluminum alloy section heat distribution characteristic diagram through a resolution reconstructor based on a decoder to obtain the reconstructed heat distribution image. It should be appreciated that the decoder is a component in the convolutional neural network model for converting the advanced feature representation back into the form of the original input data. In the decoder-based resolution reconstructor mentioned in the step, the decoder functions to remap the optimized semantic joint aluminum alloy profile heat distribution feature map onto the resolution of the original input image, thereby obtaining a reconstructed heat distribution image. The decoder is typically composed of a series of deconvolution layers (also called transpose convolution layers) and other operations, the goal of which is to recover the detail and spatial structure of the original input data by a reverse convolution operation. The structure of the decoder corresponds to the encoder (for feature extraction and abstraction), which are usually symmetrical. The optimized semantic united aluminum alloy profile heat distribution characteristic diagram can be converted back to the resolution of the original input image through a resolution reconstructor based on a decoder. This has the advantage that a reconstructed heat distribution image similar to the original input image can be generated, so that the heat distribution situation of the aluminum alloy profile can be understood and analyzed more intuitively.
By means of the decoder, the model can learn the representation and structure of the input data and convert it into a visual form, providing a more intuitive and interpretable result. In thermal distribution image reconstruction, the decoder functions to recover the detail and spatial information of the original image, making the reconstructed image more accurate and visual.
In the technical scheme of the application, when the heat distribution shallow feature map and the heat distribution deep feature map are subjected to joint semantic propagation module to obtain a semantic joint aluminum alloy section heat distribution feature map, the heat distribution shallow feature map is weighted based on global semantic information and local semantic information of the heat distribution deep feature map, so that the heat distribution shallow feature map contains deep image semantic features of a heat distribution image of the extruded aluminum alloy section to a certain extent, and therefore, synchronous representation of shallow image semantic features and deep image semantic features contained in the semantic joint aluminum alloy section heat distribution feature map is ensured, and when the semantic joint aluminum alloy section heat distribution feature map is generated through decoding regression by a decoder, a weight matrix of the decoder also has difficulty in converging labels relative to image semantic features under a preset depth dimension, and the training effect of the decoder is affected. Therefore, when the semantic united aluminum alloy section heat distribution feature map is generated through the decoder in a decoding regression mode, the semantic united aluminum alloy section heat distribution feature vector obtained after the semantic united aluminum alloy section heat distribution feature map is unfolded is subjected to weight space iteration recursion orientation proposal optimization during each iteration.
Specifically, the automatic control method for extrusion of the aluminum alloy section bar further comprises the following steps: training, namely when the semantic united aluminum alloy section heat distribution feature map is generated by decoding regression through a decoder, performing weight space iteration recursion directional proposal optimization on the semantic united aluminum alloy section heat distribution feature vector obtained after the semantic united aluminum alloy section heat distribution feature map is unfolded at each iteration to obtain an optimized semantic united aluminum alloy section heat distribution feature vector; wherein, training step is used for: carrying out weight space iteration recursion directed proposal optimization on the semantic united aluminum alloy section heat distribution feature vector obtained after the semantic united aluminum alloy section heat distribution feature map is unfolded by using the following optimization formula so as to obtain an optimized semantic united aluminum alloy section heat distribution feature vector;
wherein, the optimization formula is:wherein,and->The weight matrix of the last iteration and the current iteration are respectively adopted, wherein, during the first iteration, different initialization strategies are adopted to set +.>And->(e.g.)>Set as a unitary matrix->Set as the diagonal matrix of the mean value of the feature vector to be classified),>is the heat distribution characteristic vector of the semantic united aluminum alloy section bar, < > of the semantic united aluminum alloy section bar>Is the first feature vector, ">Is the second feature vector, ">Is the optimized semantic combined aluminum alloy section bar heat distribution characteristic vector +.>Representing matrix multiplication +.>、/>Representing addition by location and multiplication by location, respectively.
Here, the weighted spatial iterative recursive directional proposed optimization may be performed by combining the initial semantics of the aluminum alloy profile heat distribution feature vectors to be classifiedAs anchor points, to iterate the heat distribution characteristic vector corresponding to the semantic united aluminum alloy section based on the weight matrix in the weight space>Is of different depth dimensionsThe anchor point footprints (anchor points) under different depth image semantic feature distribution dimensions are obtained in the directions to serve as directional proposals (oriented proposal) for iterative recursion in a weight space, so that the class confidence and the local accuracy of the weight matrix convergence are improved based on prediction proposals, and the training effect of the semantic joint aluminum alloy profile heat distribution feature map through the decoder-based resolution reconstructor is improved.
The reconstructed thermal profile image is then passed through a thermal profile feature extractor based on a convolutional neural network model containing a spatial attention module to obtain a reconstructed thermal profile feature matrix. That is, a spatial attention module is utilized to enable the thermal profile extractor to be more focused on spatial location information among the reconstructed thermal profile images.
Accordingly, in step S123, image feature extraction is performed on the reconstructed heat distribution image to obtain a reconstructed heat distribution feature matrix, including: the reconstructed thermal distribution image is passed through a thermal distribution feature extractor based on a convolutional neural network model comprising a spatial attention module to obtain the reconstructed thermal distribution feature matrix. It should be noted that the spatial attention module is a widely used mechanism in deep learning models to enhance the attention of the model to different locations in the input data, and may help the model adaptively and selectively focus on important spatial regions when processing data (e.g., images, video) with spatial structures. The spatial attention module calculates a weight or attention score for each spatial location based on the content of the input data and the context information, typically by learning adjustable parameters. These weights may be used to weight the input data to emphasize or suppress the characteristic representation of different locations. Specifically, the spatial attention module generally includes the steps of: 1. feature extraction: firstly, extracting feature representations from input data through a convolution layer or other feature extraction methods, wherein the feature representations can be feature graphs of different layers; 2. spatial attention calculation: in this step, the attention weight of each spatial location is calculated according to a specific mechanism, and common methods include the use of convolution operations, full connection layers, or self-attention mechanisms (self-attention), etc.; 3. feature weighting: applying the calculated attention weights to the feature representation, weighting the different spatial locations, which can be achieved by a simple element multiplication operation; 4. feature fusion: the weighted features are fused to obtain a final feature representation, which may be achieved by adding, stitching or other operations to the weighted feature maps. By introducing a spatial attention module, the model can automatically focus on important spatial locations when processing input data, and can flexibly adjust the distribution of attention according to task demands. This helps to improve the perceptibility and expressivity of the model, thus achieving better performance in visual tasks.
Further, the reconstructed heat distribution feature matrix is passed through a classifier to obtain a classification result, which is used to indicate that the heating temperature value should be increased, decreased or kept unchanged.
Accordingly, as shown in fig. 6, based on the reconstructed heat distribution feature matrix, a control strategy for determining a heating temperature value includes: s1241, passing the reconstructed heat distribution feature matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the heating temperature value should be increased, decreased or kept unchanged; and S1242, using the classification result as a control strategy of the heating temperature value.
More specifically, in step S1241, the reconstructed heat distribution feature matrix is passed through a classifier to obtain a classification result, which is used to indicate that the heating temperature value should be increased, decreased, or kept unchanged, including: expanding the reconstructed heat distribution feature matrix into classification feature vectors according to row vectors or column vectors; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
That is, in the technical solution of the present disclosure, the labels of the classifier include a heating temperature value should be increased (first label), a heating temperature value should be decreased (second label), and a heating temperature value should remain unchanged (third label), wherein the classifier determines to which classification label the reconstructed heat distribution feature matrix belongs through a soft maximum function. It is noted that the first tag p1, the second tag p2, the third tag p3, and the third tag do not include the concept set by human, and in fact, the computer model does not have the concept of "the heating temperature value should be increased, should be decreased, or should remain unchanged" during the training process, which is only three kinds of classification tags and outputs the probability that the feature is under the two classification tags, that is, the sum of p1, p2, and p3 is one. Therefore, the classification result that the heating temperature value should be increased, decreased or kept unchanged is actually that the classification label is converted into the classification probability distribution conforming to the two classification of the natural law, and the physical meaning of the natural probability distribution of the label is essentially used instead of the language text meaning that the heating temperature value should be increased, decreased or kept unchanged.
It should be appreciated that the role of the classifier is to learn the classification rules and classifier using a given class, known training data, and then classify (or predict) the unknown data. Logistic regression (logistics), SVM, etc. are commonly used to solve the classification problem, and for multi-classification problems (multi-class classification), logistic regression or SVM can be used as well, but multiple bi-classifications are required to compose multiple classifications, but this is error-prone and inefficient, and the commonly used multi-classification method is the Softmax classification function.
In summary, according to the automatic control method for extrusion of the aluminum alloy profile disclosed by the embodiment of the disclosure, heating control can be intelligently performed, so that rapidly-changing process requirements can be met.
Fig. 7 shows a block diagram of an automated control system 100 for aluminum alloy profile extrusion in accordance with an embodiment of the present disclosure. As shown in fig. 7, an automatic control system 100 for aluminum alloy profile extrusion according to an embodiment of the present disclosure includes: a cutting module 110 for cutting the aluminum alloy billet to obtain a cut aluminum alloy billet; a heat treatment module 120, configured to heat-treat the cut aluminum alloy billet to obtain a heated aluminum alloy billet; an extrusion processing module 130, configured to extrude the heated aluminum alloy billet through an extruder to obtain an extruded aluminum alloy material conforming to a predetermined cross-sectional shape; and the forming module 140 is used for cooling, solidifying, shaping and cutting the extruded aluminum alloy material to obtain the aluminum alloy profile.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described automatic control system 100 for aluminum alloy profile extrusion have been described in detail in the above description of the automatic control method for aluminum alloy profile extrusion with reference to fig. 1 to 6, and thus, repetitive descriptions thereof will be omitted.
As described above, the automatic control system 100 for aluminum alloy profile extrusion according to the embodiment of the present disclosure may be implemented in various wireless terminals, such as a server having an automatic control algorithm for aluminum alloy profile extrusion, and the like. In one possible implementation, the automated control system 100 for aluminum alloy profile extrusion according to embodiments of the present disclosure may be integrated into a wireless terminal as one software module and/or hardware module. For example, the automated control system 100 for aluminum alloy profile extrusion may be a software module in the operating system of the wireless terminal, or may be an application developed for the wireless terminal; of course, the automatic control system 100 for extrusion of aluminum alloy profiles can also be one of a plurality of hardware modules of the wireless terminal.
Alternatively, in another example, the automatic control system 100 for aluminum alloy profile extrusion and the wireless terminal may be separate devices, and the automatic control system 100 for aluminum alloy profile extrusion may be connected to the wireless terminal through a wired and/or wireless network and transmit interactive information in a agreed data format.
Fig. 8 illustrates an application scenario diagram of an automatic control method of aluminum alloy profile extrusion according to an embodiment of the present disclosure. As shown in fig. 8, in this application scenario, first, a heat distribution image of the cut aluminum alloy billet (e.g., D illustrated in fig. 8) is acquired, and then, the heat distribution image of the cut aluminum alloy billet is input to a server (e.g., S illustrated in fig. 8) in which an automatic control algorithm of aluminum alloy profile extrusion is deployed, wherein the server can process the heat distribution image of the cut aluminum alloy billet using the automatic control algorithm of aluminum alloy profile extrusion to obtain a classification result for indicating that a heating temperature value should be increased, decreased, or kept unchanged.
Further, the embodiment of the disclosure also provides a variable cross-section aluminum alloy section bar hot extrusion device, including the mould pressing base, the extrusion case, the locating rack, extrusion subassembly, the feeding apron, power pack, adjusting part and clean subassembly, the top terminal surface central authorities fixed mounting of mould pressing base has the extrusion case, the terminal surface both sides of extrusion case all fixed mounting have the locating rack and be provided with the extrusion subassembly between the locating rack, adjusting part is all installed all around to one side of extrusion subassembly, top terminal surface one side fixed mounting of extrusion case has the feeding apron, one side of mould pressing base is provided with power pack, and power pack is connected with the extrusion subassembly, the opposite side of mould pressing base is provided with clean subassembly. The extrusion subassembly includes extrusion die holder, extrusion chamber, prevents scraping baffle, discharge gate, heat insulating board and electric plate, and the extrusion die holder is pegged graft and is installed in the inside of extrusion case, and the electric plate is installed to the inboard of extrusion die holder, and the inboard laminating of electric plate is installed and is prevented scraping the baffle, and the extrusion chamber has been seted up to the inboard of preventing scraping the baffle, and equal fixed mounting has the heat insulating board all around the outer wall of extrusion die holder, and the heat insulating board is connected with the inner wall laminating of extrusion case, and the discharge gate has been seted up to one side of extrusion chamber, and the one end of discharge gate passes the extrusion die holder and is located between four adjusting part 7.
In a specific example, the fastening studs are inserted and installed on the end faces of two sides of the locating frame, the top of the extrusion die holder is located on the inner side of the locating frame, one end of each fastening stud penetrates through the locating frame to be connected with the outer wall of the extrusion die holder in a fit mode, and the extrusion die holder is installed in the extrusion box through the locating frame, so that connection tightness of the extrusion die holder can be improved. The feed inlet has been seted up at the terminal surface top of extrusion die holder, and the top of feed inlet passes extrusion case and feed apron intercommunication, and the bottom of feed inlet passes heat insulating board, electric plate and scratch-resistant baffle and extrusion chamber intercommunication, opens the back with the feed apron, places the aluminum alloy in the inside of extrusion chamber through the feed inlet, and it is closed with the feed apron, conveniently places the aluminum alloy material. The power component comprises a first support, an extrusion hydraulic cylinder, a hydraulic push rod and an extrusion seat, wherein the first support is fixedly arranged on one side of a die pressing base through a screw, the extrusion hydraulic cylinder is fixedly arranged on the top end face of the first support, the hydraulic push rod is inserted and arranged on one side of the extrusion hydraulic cylinder, one end of the hydraulic push rod is inserted and arranged in an extrusion cavity, the extrusion seat is fixedly arranged at one end of the hydraulic push rod, the outer wall of the extrusion seat is connected with the inner wall of the scraping-proof partition in a fitting manner, the extrusion hydraulic cylinder at the top of the first support works, the extrusion hydraulic cylinder enables the hydraulic push rod to push the extrusion seat, the extrusion seat moves in the extrusion cavity, and the extrusion seat pushes aluminum alloy to slide between the scraping-proof partition in the moving process. The adjusting component comprises a beam frame, a connecting support plate, a fixed bottom plate, a first electric telescopic rod, a laminating base plate and a guide lug, wherein the beam frame is fixedly arranged on the other side of the end face of the extrusion die holder through welding; the fixed bottom plate can be supported through the cross beam frame and the connecting support plate, so that a first electric telescopic rod arranged on one side of the fixed bottom plate works, and the first electric telescopic rod pushes the bonding base plate, so that the bonding base plate pushes the guide lug to perform variable cross section treatment on aluminum alloy extruded from the discharge hole; the guide convex blocks with different shapes can be arranged on the bonding base plate, so that different processing requirements can be met. The cleaning assembly comprises a second bracket, a cleaning sleeve, a mounting groove, a second electric telescopic rod, a cleaning frame and a cleaning brush, wherein the second bracket is fixedly arranged on the other side of the die pressing base through a screw; one end of the second electric telescopic rod is connected with a cleaning rod, one end of the cleaning rod penetrates through the mounting groove and is fixedly connected with the end face of the cleaning frame, the fixed bottom plate can be supported through the cross beam frame and the connecting support plate, the first electric telescopic rod arranged on one side of the fixed bottom plate works, the first electric telescopic rod pushes the bonding base plate, and the bonding base plate pushes the guide protruding block to perform variable cross section treatment on aluminum alloy extruded from the discharge hole; the guide convex blocks with different shapes can be arranged on the bonding base plate, so that different processing requirements can be met.
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The foregoing description of the embodiments of the present disclosure has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described. The terminology used herein was chosen in order to best explain the principles of the embodiments, the practical application, or the improvement of technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.
Claims (8)
1. An automatic control method for extrusion of aluminum alloy profiles is characterized by comprising the following steps:
cutting the aluminum alloy blank to obtain a cut aluminum alloy blank;
heating the cut aluminum alloy blank to obtain a heated aluminum alloy blank;
extruding the heated aluminum alloy blank through an extruder to obtain an extruded aluminum alloy material conforming to a preset cross-sectional shape; and
cooling, solidifying, shaping and cutting the extruded aluminum alloy material to obtain an aluminum alloy profile;
wherein, carry out the heating treatment to the aluminium alloy blank after cutting in order to obtain the aluminium alloy blank after heating, include:
acquiring a heat distribution image of the cut aluminum alloy blank;
performing image preprocessing on the heat distribution image of the cut aluminum alloy blank to obtain a reconstructed heat distribution image;
extracting image features of the reconstructed heat distribution image to obtain a reconstructed heat distribution feature matrix; and
determining a control strategy of a heating temperature value based on the reconstructed heat distribution feature matrix;
the image preprocessing is carried out on the heat distribution image of the cut aluminum alloy blank to obtain a reconstructed heat distribution image, and the image preprocessing comprises the following steps:
extracting shallow characteristic information and deep characteristic information of the heat distribution image of the cut aluminum alloy blank to obtain a heat distribution shallow characteristic image and a heat distribution deep characteristic image;
carrying out semantic propagation on the heat distribution shallow feature map and the heat distribution deep feature map to obtain a semantic united aluminum alloy profile heat distribution feature map; and
and carrying out image resolution reconstruction based on the semantic united aluminum alloy section heat distribution characteristic diagram to obtain the reconstructed heat distribution image.
2. The automatic control method for aluminum alloy profile extrusion according to claim 1, wherein extracting shallow characteristic information and deep characteristic information of the heat distribution image of the cut aluminum alloy billet to obtain a heat distribution shallow characteristic map and a heat distribution deep characteristic map comprises:
passing the heat distribution image of the cut aluminum alloy blank through a heat distribution shallow feature extractor based on a first convolutional neural network model to obtain a heat distribution shallow feature map; and
and the thermal distribution shallow layer characteristic map is passed through a thermal distribution deep layer characteristic extractor based on a second convolution neural network model to obtain the thermal distribution deep layer characteristic map.
3. The automatic control method for extrusion of aluminum alloy profiles according to claim 2, wherein semantically propagating the thermal profile shallow layer feature map and the thermal profile deep layer feature map to obtain semantically combined aluminum alloy profile thermal profile feature map comprises:
and passing the heat distribution shallow layer feature map and the heat distribution deep layer feature map through a joint semantic propagation module to obtain the semantic joint aluminum alloy profile heat distribution feature map.
4. An automatic control method for extrusion of aluminum alloy profiles according to claim 3, characterized in that the image resolution reconstruction based on the semantically combined aluminum alloy profile heat distribution profile to obtain the reconstructed heat distribution image comprises:
and passing the semantic united aluminum alloy section heat distribution characteristic diagram through a resolution reconstructor based on a decoder to obtain the reconstructed heat distribution image.
5. The automatic control method for extrusion of aluminum alloy profiles according to claim 4, wherein the image feature extraction of the reconstructed heat distribution image to obtain a reconstructed heat distribution feature matrix comprises:
the reconstructed thermal distribution image is passed through a thermal distribution feature extractor based on a convolutional neural network model comprising a spatial attention module to obtain the reconstructed thermal distribution feature matrix.
6. The automatic control method for aluminum alloy profile extrusion according to claim 5, wherein determining a control strategy for heating temperature values based on the reconstructed heat distribution feature matrix comprises:
passing the reconstructed heat distribution feature matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the heating temperature value should be increased, decreased or kept unchanged; and
and taking the classification result as a control strategy of the heating temperature value.
7. The automatic control method for aluminum alloy profile extrusion as recited in claim 6, further comprising: training, namely when the semantic united aluminum alloy section heat distribution feature map is generated by decoding regression through a decoder, performing weight space iteration recursion directional proposal optimization on the semantic united aluminum alloy section heat distribution feature vector obtained after the semantic united aluminum alloy section heat distribution feature map is unfolded at each iteration to obtain an optimized semantic united aluminum alloy section heat distribution feature vector;
wherein, training step is used for: carrying out weight space iteration recursion directed proposal optimization on the semantic united aluminum alloy section heat distribution feature vector obtained after the semantic united aluminum alloy section heat distribution feature map is unfolded by using the following optimization formula so as to obtain an optimized semantic united aluminum alloy section heat distribution feature vector;
wherein, the optimization formula is:
wherein M is 1 And M 2 The weight matrix of the last iteration and the current iteration are respectively V c Is the heat distribution characteristic vector of the semantic united aluminum alloy section, V 1 Is a first eigenvector, V 2 Is the second eigenvector, V' c Is the optimized semantic united aluminum alloy section bar heat distribution characteristic vector,representing matrix multiplication +.>As indicated by the letter, ", is added by position and multiplied by position, respectively.
8. An automatic control system for extrusion of aluminum alloy profiles, comprising:
the cutting module is used for cutting the aluminum alloy blank to obtain a cut aluminum alloy blank;
the heating treatment module is used for carrying out heating treatment on the cut aluminum alloy blank to obtain a heated aluminum alloy blank;
the extrusion treatment module is used for extruding the heated aluminum alloy blank through an extruder to obtain an extruded aluminum alloy material conforming to the preset cross-sectional shape; and
the molding module is used for cooling, solidifying, shaping and cutting the extruded aluminum alloy material to obtain an aluminum alloy section;
wherein, carry out the heating treatment to the aluminium alloy blank after cutting in order to obtain the aluminium alloy blank after heating, include:
acquiring a heat distribution image of the cut aluminum alloy blank;
performing image preprocessing on the heat distribution image of the cut aluminum alloy blank to obtain a reconstructed heat distribution image;
extracting image features of the reconstructed heat distribution image to obtain a reconstructed heat distribution feature matrix; and
determining a control strategy of a heating temperature value based on the reconstructed heat distribution feature matrix;
the image preprocessing is carried out on the heat distribution image of the cut aluminum alloy blank to obtain a reconstructed heat distribution image, and the image preprocessing comprises the following steps:
extracting shallow characteristic information and deep characteristic information of the heat distribution image of the cut aluminum alloy blank to obtain a heat distribution shallow characteristic image and a heat distribution deep characteristic image;
carrying out semantic propagation on the heat distribution shallow feature map and the heat distribution deep feature map to obtain a semantic united aluminum alloy profile heat distribution feature map; and
and carrying out image resolution reconstruction based on the semantic united aluminum alloy section heat distribution characteristic diagram to obtain the reconstructed heat distribution image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311250149.0A CN117225921B (en) | 2023-09-26 | 2023-09-26 | Automatic control system and method for extrusion of aluminum alloy profile |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311250149.0A CN117225921B (en) | 2023-09-26 | 2023-09-26 | Automatic control system and method for extrusion of aluminum alloy profile |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117225921A CN117225921A (en) | 2023-12-15 |
| CN117225921B true CN117225921B (en) | 2024-03-12 |
Family
ID=89082347
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311250149.0A Active CN117225921B (en) | 2023-09-26 | 2023-09-26 | Automatic control system and method for extrusion of aluminum alloy profile |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117225921B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101957331A (en) * | 2010-03-16 | 2011-01-26 | 浙江工业大学 | Phase change thermography display system based on image processing algorithm |
| CN105312347A (en) * | 2014-07-18 | 2016-02-10 | 赵敏 | Moulding method of aluminum alloy profile |
| EP3171297A1 (en) * | 2015-11-18 | 2017-05-24 | CentraleSupélec | Joint boundary detection image segmentation and object recognition using deep learning |
| CN110263756A (en) * | 2019-06-28 | 2019-09-20 | 东北大学 | A kind of human face super-resolution reconstructing system based on joint multi-task learning |
| CN111283006A (en) * | 2020-02-26 | 2020-06-16 | 凯维思轻量化智能制造研究院(菏泽)有限公司 | Bidirectional extrusion forming process and equipment |
| CN111438212A (en) * | 2020-04-03 | 2020-07-24 | 广西南南铝加工有限公司 | Preparation method of rapid extrusion hard aluminum alloy section |
| CN113102534A (en) * | 2021-04-14 | 2021-07-13 | 烟台大学 | Hot extrusion device for variable-cross-section aluminum alloy section |
| CN116680987A (en) * | 2023-06-15 | 2023-09-01 | 宁波同耀新材料科技有限公司 | Forming method and system of graphite crucible |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11084225B2 (en) * | 2018-04-02 | 2021-08-10 | Nanotronics Imaging, Inc. | Systems, methods, and media for artificial intelligence process control in additive manufacturing |
-
2023
- 2023-09-26 CN CN202311250149.0A patent/CN117225921B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101957331A (en) * | 2010-03-16 | 2011-01-26 | 浙江工业大学 | Phase change thermography display system based on image processing algorithm |
| CN105312347A (en) * | 2014-07-18 | 2016-02-10 | 赵敏 | Moulding method of aluminum alloy profile |
| EP3171297A1 (en) * | 2015-11-18 | 2017-05-24 | CentraleSupélec | Joint boundary detection image segmentation and object recognition using deep learning |
| CN110263756A (en) * | 2019-06-28 | 2019-09-20 | 东北大学 | A kind of human face super-resolution reconstructing system based on joint multi-task learning |
| CN111283006A (en) * | 2020-02-26 | 2020-06-16 | 凯维思轻量化智能制造研究院(菏泽)有限公司 | Bidirectional extrusion forming process and equipment |
| CN111438212A (en) * | 2020-04-03 | 2020-07-24 | 广西南南铝加工有限公司 | Preparation method of rapid extrusion hard aluminum alloy section |
| CN113102534A (en) * | 2021-04-14 | 2021-07-13 | 烟台大学 | Hot extrusion device for variable-cross-section aluminum alloy section |
| CN116680987A (en) * | 2023-06-15 | 2023-09-01 | 宁波同耀新材料科技有限公司 | Forming method and system of graphite crucible |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117225921A (en) | 2023-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107766794B (en) | Image semantic segmentation method with learnable feature fusion coefficient | |
| Gosala et al. | Bird’s-eye-view panoptic segmentation using monocular frontal view images | |
| CN109671070B (en) | Target detection method based on feature weighting and feature correlation fusion | |
| CN114008663A (en) | Real-time video super-resolution | |
| CN110826596A (en) | Semantic segmentation method based on multi-scale deformable convolution | |
| CN111696110B (en) | Scene segmentation method and system | |
| CN110533022B (en) | Target detection method, system, device and storage medium | |
| CN113610144A (en) | Vehicle classification method based on multi-branch local attention network | |
| CN105243385B (en) | A kind of image quality evaluating method based on unsupervised learning | |
| CN109859166B (en) | Multi-column convolutional neural network-based parameter-free 3D image quality evaluation method | |
| CN113313173A (en) | Human body analysis method based on graph representation and improved Transformer | |
| CN104809706A (en) | Single lens computational imaging method based on gentle image color change priori | |
| CN116872089A (en) | Grinding self-adaptive control system and method of numerically controlled grinder | |
| CN115861246A (en) | Product quality abnormity detection method and system applied to industrial Internet | |
| CN117225921B (en) | Automatic control system and method for extrusion of aluminum alloy profile | |
| CN114842542A (en) | Facial action unit identification method and device based on self-adaptive attention and space-time correlation | |
| CN114338416B (en) | Space-time multi-index prediction method and device and storage medium | |
| CN113781461A (en) | Intelligent patient monitoring and sequencing method | |
| CN116959252A (en) | Space-time convolution traffic flow prediction method and system based on R-C-R structure | |
| CN114494284B (en) | Scene analysis model and method based on explicit supervision area relation | |
| CN111242044B (en) | Night unmanned vehicle scene prediction method based on ConvLSTM dual-channel coding network | |
| CN113223038A (en) | Discrete cosine transform-based mask representation instance segmentation method | |
| CN113822954A (en) | Deep learning image coding method for man-machine cooperation scene under resource constraint | |
| CN117066313B (en) | Automatic control system and method for frame aluminum product production line | |
| Li et al. | Towards real-time segmentation on the edge |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |