CN117021434A - PPR pipe cooling forming control system and method thereof - Google Patents
PPR pipe cooling forming control system and method thereof Download PDFInfo
- Publication number
- CN117021434A CN117021434A CN202311070778.5A CN202311070778A CN117021434A CN 117021434 A CN117021434 A CN 117021434A CN 202311070778 A CN202311070778 A CN 202311070778A CN 117021434 A CN117021434 A CN 117021434A
- Authority
- CN
- China
- Prior art keywords
- matrix
- flow
- feature
- cooling
- cooling state
- 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- Withdrawn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
- 238000001816 cooling Methods 0.000 title claims abstract description 278
- 238000000034 method Methods 0.000 title claims abstract description 33
- 239000000498 cooling water Substances 0.000 claims abstract description 66
- 238000000465 moulding Methods 0.000 claims abstract description 51
- 239000013598 vector Substances 0.000 claims description 317
- 239000011159 matrix material Substances 0.000 claims description 193
- 230000008859 change Effects 0.000 claims description 127
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 54
- 230000009466 transformation Effects 0.000 claims description 44
- 238000012544 monitoring process Methods 0.000 claims description 33
- 238000013527 convolutional neural network Methods 0.000 claims description 30
- 238000012545 processing Methods 0.000 claims description 29
- 230000006870 function Effects 0.000 claims description 28
- 238000000605 extraction Methods 0.000 claims description 19
- 230000004927 fusion Effects 0.000 claims description 17
- 230000000903 blocking effect Effects 0.000 claims description 16
- 238000010606 normalization Methods 0.000 claims description 16
- 238000013507 mapping Methods 0.000 claims description 13
- 230000003247 decreasing effect Effects 0.000 claims description 11
- 239000000463 material Substances 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 10
- 238000012546 transfer Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 230000004044 response Effects 0.000 claims description 8
- 238000003062 neural network model Methods 0.000 claims description 7
- 230000004043 responsiveness Effects 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 6
- 230000015572 biosynthetic process Effects 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000001105 regulatory effect Effects 0.000 abstract description 8
- 230000007704 transition Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 10
- 238000013135 deep learning Methods 0.000 description 8
- 238000013528 artificial neural network Methods 0.000 description 7
- 238000013145 classification model Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 239000002994 raw material Substances 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000009826 distribution Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 241001391944 Commicarpus scandens Species 0.000 description 3
- 239000004743 Polypropylene Substances 0.000 description 3
- 230000001276 controlling effect Effects 0.000 description 3
- -1 polypropylene Polymers 0.000 description 3
- 229920001155 polypropylene Polymers 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000002411 adverse Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000001514 detection method Methods 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 239000004033 plastic Substances 0.000 description 1
- 229920005604 random copolymer Polymers 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C35/00—Heating, cooling or curing, e.g. crosslinking or vulcanising; Apparatus therefor
- B29C35/16—Cooling
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C35/00—Heating, cooling or curing, e.g. crosslinking or vulcanising; Apparatus therefor
- B29C35/16—Cooling
- B29C2035/1616—Cooling using liquids
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29L—INDEXING SCHEME ASSOCIATED WITH SUBCLASS B29C, RELATING TO PARTICULAR ARTICLES
- B29L2023/00—Tubular articles
- B29L2023/22—Tubes or pipes, i.e. rigid
Landscapes
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Thermal Sciences (AREA)
- Control Of Metal Rolling (AREA)
Abstract
The application relates to the technical field of intelligent control, and particularly discloses a PPR (point-to-point) pipe cooling forming control system and a PPR pipe cooling forming control method. Therefore, the flow rate of cooling water can be monitored and regulated in real time so as to ensure the molding quality of the PPR pipe.
Description
Technical Field
The application relates to the technical field of intelligent control, in particular to a PPR (polypropylene random) pipe cooling and forming control system and a PPR pipe cooling and forming control method.
Background
PPR (Poly Propylene Random) pipe, namely the random copolymer polypropylene pipe, is a novel water pipe material commonly adopted in developed countries. Due to its excellent properties, its market share has been far ahead of other plastic water pipe materials in recent years. At present, the production and application of PPR pipes in China also enter the industrialized development stage.
In the process of preparing the PPR pipe, heating treatment is generally required before cooling and forming the PPR pipe, so that raw materials can be softened and adapt to the shape of a forming die, and the PPR pipe is cooled after the pipe is formed, so that the PPR pipe is solidified and stabilized. During cooling, the cooling water temperature is typically below 20 degrees, the first stage temperature may be slightly higher and the later stage lower, creating a temperature gradient. And, cooling water flow is an important factor affecting the quality of the PPR tubing. The excessive flow causes that the cooling is too fast and stress is easy to be generated in the pipe, so that the surface of the pipe is rough and spot pits are generated; the flow is too small, so that bright spots are generated on the surface of the pipe and are easy to break, and the wall thickness of the pipe is uneven.
Accordingly, a PPR pipe cooling molding control system and method thereof are desired.
Disclosure of Invention
The present application has been made to solve the above-mentioned technical problems. The embodiment of the application provides a PPR pipe cooling forming control system and a method thereof, wherein a monitoring video of a preset time period of the PPR pipe in a cooling process and flow values of cooling water at a plurality of preset time points in the preset time period are firstly obtained, time sequence characteristic information of the flow values of the cooling water and state change characteristic information of the PPR pipe are extracted based on a deep learning technology, and real-time control of the flow of the cooling water is carried out based on relevance characteristic representation of the time sequence characteristic information and the state change characteristic information of the PPR pipe. Therefore, the flow rate of cooling water can be monitored and regulated in real time so as to ensure the molding quality of the PPR pipe.
Accordingly, according to one aspect of the present application, there is provided a PPR tubing cooling molding control system comprising:
the data acquisition module is used for acquiring flow values of cooling water at a plurality of preset time points in a preset time period and PPR pipe cooling state monitoring videos of the preset time period acquired by the camera;
the key frame extraction module is used for extracting a plurality of cooling state key frames from the PPR pipe cooling state monitoring video;
the image semantic coding module is used for obtaining a plurality of cooling state feature vectors through a ViT model containing an embedded layer after performing image blocking processing on the plurality of material state key frames respectively;
the cooling state change feature extraction module is used for two-dimensionally arranging the plurality of cooling state feature vectors into a two-dimensional input matrix and then obtaining the cooling state change feature vectors through a convolutional neural network model serving as a filter;
the flow characteristic time sequence coding module is used for arranging the flow values of the cooling water at a plurality of preset time points into flow input vectors according to time dimensions and then obtaining flow characteristic vectors through a time sequence coder comprising a full-connection layer and a one-dimensional convolution layer;
The responsiveness estimation module is used for calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as a classification feature matrix;
the optimizing module is used for fusing the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, and calculating the position-based point multiplication between the fused feature matrix and the classification feature matrix to obtain an optimized classification feature matrix;
and the control result generation module is used for enabling the optimized classification characteristic matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow value of the cooling water at the current time point is increased or decreased.
In the above PPR pipe cooling molding control system, the key frame extraction module is configured to: a plurality of cooling state key frames are extracted from the PPR tube cooling state monitoring video at a predetermined sampling frequency.
In the above PPR pipe cooling molding control system, the image semantic coding module includes: the blocking unit is used for carrying out image blocking processing on the cooling state key frame to obtain a sequence of image blocks; the embedded coding unit is used for respectively carrying out embedded coding on each image block in the sequence of the image blocks by using an embedded layer of the ViT model so as to obtain a sequence of embedded vectors of the image blocks; a conversion coding unit, configured to input the sequence of image block embedded vectors into a converter module of the ViT model to obtain a sequence of image block feature vectors; and the cascading unit is used for cascading the sequence of the image block feature vectors to obtain the cooling state feature vector.
In the above PPR pipe cooling molding control system, the conversion coding unit includes: an arrangement subunit, configured to arrange the sequence of the image block embedding vectors into an input vector; the conversion subunit is used for converting the input vector into a query vector and a key vector respectively through a learning embedding matrix; a self-attention subunit, configured to calculate a product between the query vector and a transpose vector of the key vector to obtain a self-attention correlation matrix; a normalization subunit, configured to perform normalization processing on the self-attention association matrix to obtain a normalized self-attention association matrix; an attention calculating subunit, configured to activate the normalized self-attention association matrix input Softmax activation function to obtain a self-attention feature matrix; and the attention applying subunit is used for multiplying the self-attention characteristic matrix with each image block embedding vector in the sequence of the image block embedding vectors respectively to obtain the sequence of the image block characteristic vectors.
In the above PPR pipe cooling molding control system, the cooling state change feature extraction module is configured to: processing the two-dimensional input matrix by using the convolutional neural network model serving as a filter according to the following convolution formula to obtain the cooling state change feature vector;
Wherein, the convolution formula is:
f i ＝GP{Sigmoid(N i ×f i-1 +B i )}
wherein f i-1 For input of the ith layer convolutional neural network model, f i For the output of the ith layer convolution neural network model, N i Filter for layer i convolutional neural network model, and B i For the bias matrix of the i-th layer convolutional neural network model, sigmoid represents a nonlinear activation function, and GP represents global feature pooling operation on each feature matrix of the feature map along the channel dimension.
In the above-mentioned PPR tubular product cooling molding control system, the flow characteristic time sequence coding module includes: the normalization mapping unit is used for carrying out normalization processing based on the maximum value on the flow input vector so as to obtain a normalized flow input vector; the full-connection coding unit is used for carrying out full-connection coding on the normalized flow input vector by using a full-connection layer of the time sequence coder so as to extract high-dimensional implicit characteristics of characteristic values of all positions in the normalized flow input vector; and the one-dimensional convolution coding unit is used for carrying out one-dimensional coding on the normalized flow input vector by using a one-dimensional convolution layer of the time sequence coder so as to extract associated high-dimensional implicit association features among feature values of all positions in the normalized flow input vector.
In the above PPR pipe cooling molding control system, the responsiveness estimation module is configured to: calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as the classification feature matrix according to the following response formula; wherein, the response formula is:
wherein V is 1 Representing the cooling state change feature vector, V 2 The flow characteristic vector is represented as such,representing matrix multiplication, M representing the classification feature matrix.
In the above PPR pipe cooling molding control system, the optimizing module includes: the normalization unit is used for normalizing the cooling state change feature vector and the flow feature vector to obtain a normalized cooling state change feature vector and a normalized flow feature vector; an included angle calculating unit, configured to calculate an included angle value between the cooling state change feature vector and the flow feature vector; a rotation matrix construction unit configured to construct a rotation matrix in a rotation formula based on the angle value such that the rotation matrix multiplied by the normalized cooling state change feature vector is equal to the normalized flow feature vector, wherein the rotation formula is:
Wherein R is the rotation matrix, theta is the included angle value between the cooling state change feature vector and the flow feature vector, sin theta is a sine function, and cos theta is a cosine function; an affine transformation function constructing unit for controlling linear interpolation between the rotation matrix and the identity matrix in an affine transformation formula based on a predetermined affine transformation superparameter, wherein the affine transformation formula is:
f(α)＝(1-α)I+αR
wherein alpha is a preset affine transformation super parameter, f (alpha) is the affine transformation function, I is the identity matrix, and R is the rotation matrix; an affine transformation unit for applying the affine transformation function to the cooling state change feature vector and the flow feature vector, respectively, to obtain an affine transformed cooling state change feature vector and an affine transformed flow feature vector; and the fusion unit is used for calculating the product between the affine transformation cooling state change feature vector and the transpose vector of the affine transformation flow feature vector to obtain the fusion feature matrix.
According to another aspect of the present application, there is provided a PPR pipe cooling molding control method, comprising:
acquiring flow values of cooling water at a plurality of preset time points in a preset time period and a PPR pipe cooling state monitoring video of the preset time period acquired by a camera;
Extracting a plurality of cooling state key frames from the PPR pipe cooling state monitoring video;
image blocking processing is carried out on the material state key frames respectively, and then a ViT model containing an embedded layer is used for obtaining a plurality of cooling state feature vectors;
the plurality of cooling state feature vectors are two-dimensionally arranged into a two-dimensional input matrix and then pass through a convolutional neural network model serving as a filter to obtain a cooling state change feature vector;
arranging the flow values of the cooling water at a plurality of preset time points into flow input vectors according to a time dimension, and then obtaining flow characteristic vectors through a time sequence encoder comprising a full-connection layer and a one-dimensional convolution layer;
calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as a classification feature matrix;
fusing the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, and calculating the position-based point multiplication between the fused feature matrix and the classification feature matrix to obtain an optimized classification feature matrix;
and the optimized classification characteristic matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow value of cooling water at the current time point is increased or decreased.
In the above-mentioned PPR pipe cooling molding control method, extracting a plurality of cooling state key frames from the PPR pipe cooling state monitoring video includes: a plurality of cooling state key frames are extracted from the PPR tube cooling state monitoring video at a predetermined sampling frequency.
Compared with the prior art, the PPR pipe cooling forming control system and the PPR pipe cooling forming control method provided by the application have the advantages that firstly, the monitoring video of the PPR pipe in a preset time period in the cooling process and the flow values of cooling water at a plurality of preset time points in the preset time period are obtained, the time sequence characteristic information of the cooling water flow value and the state change characteristic information of the PPR pipe are extracted based on a deep learning technology, and the real-time control of the cooling water flow is performed based on the correlation characteristic representation of the time sequence characteristic information and the state change characteristic information of the PPR pipe. Therefore, the flow rate of cooling water can be monitored and regulated in real time so as to ensure the molding quality of the PPR pipe.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing embodiments of the present application in more detail with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and together with the embodiments of the application, and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.
FIG. 1 is a block diagram of a PPR pipe cooling molding control system according to an embodiment of the present application.
Fig. 2 is a schematic diagram of a PPR pipe cooling molding control system according to an embodiment of the present application.
Fig. 3 is a block diagram of an image semantic coding module in a PPR tubing cooling formation control system according to an embodiment of the present application.
Fig. 4 is a block diagram of a conversion coding unit in a PPR tubing cooling formation control system according to an embodiment of the present application.
FIG. 5 is a block diagram of a flow feature timing encoding module in a PPR pipe cooling molding control system according to an embodiment of the present application.
Fig. 6 is a flowchart of a PPR tubing cooling molding control method according to an embodiment of the present application.
Detailed Description
Hereinafter, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Summary of the application
As described above, in the PPR pipe manufacturing process, the raw material is first heated to soften the raw material and adapt to the shape of the forming mold, and the pipe is waited for to be formed, and then cooled to be solidified and stabilized. During cooling, the flow rate of cooling water is an important factor affecting the quality of the PPR tubing. The excessive flow causes that the cooling is too fast and stress is easy to be generated in the pipe, so that the surface of the pipe is rough and spot pits are generated; the flow is too small, so that bright spots are generated on the surface of the pipe and are easy to break, and the wall thickness of the pipe is uneven. Accordingly, a PPR pipe cooling molding control system and method thereof are desired.
At present, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, speech signal processing, and the like. In addition, deep learning and neural networks have also shown levels approaching and even exceeding humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
In recent years, deep learning and development of a neural network provide new solutions and schemes for controlling cooling water flow in the cooling process of PPR pipes.
Accordingly, the flow value of the cooling water is adaptively regulated and controlled according to the state change of the PPR pipe in consideration of the fact that the flow of the cooling water is actually controlled. That is, it is necessary to extract the correlation characteristic distribution information between the state change characteristic of the PPR pipe and the flow dynamic characteristic of the cooling water, and control the flow rate of the cooling water. Based on the above, in the technical scheme of the application, firstly, a monitoring video of a preset time period of the PPR pipe in the cooling process and flow values of cooling water at a plurality of preset time points in the preset time period are obtained, time sequence characteristic information of the flow values of the cooling water and state change characteristic information of the PPR pipe are extracted based on a deep learning technology, and real-time control of the flow of the cooling water is performed based on relevance characteristic representation of the time sequence characteristic information and the state change characteristic information of the PPR pipe. Therefore, the flow rate of cooling water can be monitored and regulated in real time so as to ensure the molding quality of the PPR pipe.
Specifically, in the technical scheme of the application, firstly, flow values of cooling water at a plurality of preset time points in a preset time period are acquired through a flow sensor, and PPR pipe cooling state monitoring videos in the preset time period are acquired through a camera. Then, considering that a large number of similar frames exist in the PPR pipe cooling state monitoring video, in order to reduce the calculation amount and avoid adverse effects of data redundancy on subsequent feature extraction, a plurality of cooling state key frames are extracted from the PPR pipe cooling state monitoring video at a preset sampling frequency so as to capture and represent important states in the cooling process.
Further, feature mining of the cooling state key frames is performed using a ViT model containing an embedded layer having excellent performance in implicit feature extraction of images to extract implicit context semantic association feature distribution information of the respective cooling state key frames, respectively, thereby obtaining cooling state feature vectors corresponding to the respective cooling state key frames. It should be appreciated that the ViT (Vision Transformer) model is a transducer-based neural network architecture that performs well in image processing tasks. Specifically, firstly, the cooling state key frame is subjected to block processing to obtain a sequence of image blocks, so that local features in an image are captured better, and meanwhile, the problem that global features are excessively smoothed is avoided; then, the image block is converted into an embedded vector through the embedded layer, and the embedded vector is encoded by using a converter module of the ViT model so as to capture semantic feature information of the image block, thereby better describing and representing cooling state features of different positions and areas and providing a more detailed and comprehensive information basis for subsequent classification and decision.
Then, considering that the plurality of cooling state feature vectors represent cooling state feature information of the PPR pipe at different time points, in order to capture time sequence dynamic change features of the cooling state of the PPR pipe, the plurality of cooling state feature vectors are two-dimensionally arranged into a two-dimensional input matrix and processed through a Convolutional Neural Network (CNN) model serving as a filter to capture spatial correlation between features by utilizing a two-dimensional convolution operation, thereby obtaining the cooling state change feature vector. It should be appreciated that convolutional neural networks are a widely used deep learning model in computer vision tasks that can effectively extract local features in an image, which can be slid over an input matrix using convolutional kernels to obtain local features and patterns of variation. And by introducing proper convolution layers, pooling layers and activation functions into the convolution neural network, effective feature extraction and change analysis can be performed on the input matrix.
Meanwhile, in order to extract dynamic change characteristics of the flow value of the cooling water in a time sequence, the cooling water flow values at a plurality of preset time points are further arranged into flow input vectors according to a time dimension, and characteristic extraction is carried out through a time sequence encoder comprising a full-connection layer and a one-dimensional convolution layer, so that flow characteristic vectors are obtained. Those of ordinary skill in the art will appreciate that a time series encoder is a special neural network model for time series data that captures the time series patterns and dynamic changes in the time series. The one-dimensional convolution layer of the time sequence encoder can capture the local relevance and the change mode of the flow characteristics; the fully connected layer can perform dimension transformation and mapping on the flow values, so that the model can understand the change relation between the flow values.
Then, in order to capture the correlation between the cooling state change and the flow characteristic, a transfer matrix of the cooling state change characteristic vector relative to the flow characteristic vector is further calculated as a classification characteristic matrix so as to combine the change mode of the cooling state with the flow characteristic, thereby providing more comprehensive classification information. It should be appreciated that the transition matrix represents a matrix of transition probabilities from one state to another, and by calculating the transition matrix, it is possible to analyze how the change in cooling state is affected by the flow characteristics.
Further, the state and flow rate changes of the cooling system are described from different angles, respectively, taking into account the cooling state change feature vector and the flow rate feature vector. Fusing the two feature vectors can combine their information to provide a richer, more comprehensive representation of the feature. This helps capture more relevant features, improving the model's understanding and expressive power of the data. The cooling state change feature vector and the flow feature vector typically contain different information. Fusing them together may provide complementary information that more fully describes the state and characteristics of the cooling system. Fusing different types of feature vectors can provide more information for classification model learning. This helps to improve the performance of the classification model, enabling it to better distinguish between different classes or states. By fusing the cooling state change feature vector and the flow feature vector, the relativity and interaction between the cooling state change feature vector and the flow feature vector can be utilized, and the discrimination capability of the classification model on the cooling system state can be improved.
In particular, considering that the cooling state change feature vector and the flow feature vector are obtained by differently encoding data of a non-use source, the two have different dimensions, and if directly fused, information is lost due to dimension mismatch. Vectors of different dimensions typically represent different features or attributes. When they are fused together, some important characteristic information may be lost, thereby affecting subsequent analysis or application. At the same time, vectors of different dimensions tend to have different semantic representations. Fusing them directly may lead to problems of semantic inconsistency. Fusing them directly may result in confusing or erroneous semantic representations. Therefore, the cooling state change feature vector and the flow feature vector with different dimensions are further mapped into the subspace with the same dimension to be fused by utilizing the property of affine subspace mapping.
Specifically, fusing the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, including: normalizing the cooling state change feature vector and the flow feature vector to obtain a normalized cooling state change feature vector and a normalized flow feature vector; calculating an included angle value between the cooling state change feature vector and the flow feature vector; constructing a rotation matrix based on the included angle values such that the rotation matrix multiplied by the normalized cooling state change feature vector equals the normalized flow feature vector, wherein the rotation matrix is formulated as:
Defining an affine transformation function based on a predetermined affine transformation superparameter, wherein the affine transformation function controls linear interpolation between the rotation matrix and the identity matrix by the predetermined affine transformation superparameter, expressed by a formula:
f(α)＝(1-α)I+αR
wherein alpha is a preset affine transformation super parameter, and I is the identity matrix; applying the affine transformation function to the cooling state change feature vector and the flow feature vector to obtain an affine transformation cooling state change feature vector and an affine transformation flow feature vector; and calculating the product between the affine transformation cooling state change feature vector and the transpose vector of the affine transformation flow feature vector to obtain the fusion feature matrix.
By utilizing the property of affine subspace mapping, the cooling state change feature vectors and the flow feature vectors with different dimensions are mapped into subspaces with the same dimension, namely, the affine transformation subspaces are used as a common medium space to fuse the cooling state change feature vectors and the flow feature vectors with different dimensions, so that the fused feature matrix can reduce redundancy of features, reduce the dimension of data and simultaneously can contain the information of two original feature vectors. And, through carrying out feature fusion among feature vectors based on affine subspace mapping, the robustness of a fusion feature matrix can be enhanced, so that the fusion feature matrix can resist noise and abnormal value interference, and the quality of data is improved.
And after the fusion feature matrix is obtained, multiplying the position points between the fusion feature matrix and the classification feature matrix to obtain an optimized classification feature matrix so as to improve the classification judgment accuracy of the optimized classification feature matrix.
Finally, the optimized classification feature matrix is passed through a classifier to obtain a classification result for indicating whether the cooling water flow value at the current time point should be increased or decreased. The classifier is a machine learning model capable of learning the mapping relationship from input features to output labels. The classifier after training can predict whether the cooling water flow value should be increased or decreased according to the input classification characteristic matrix, and output a corresponding classification result. Therefore, the cooling system can be automatically controlled based on the classification result, so that the purposes of optimizing the cooling effect and saving energy are achieved.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.
Exemplary System
FIG. 1 is a block diagram of a PPR pipe cooling molding control system according to an embodiment of the present application. As shown in fig. 1, a PPR pipe cooling molding control system 100 according to an embodiment of the present application includes: the data acquisition module 110 is used for acquiring flow values of cooling water at a plurality of preset time points in a preset time period and PPR pipe cooling state monitoring videos of the preset time period acquired by the camera; a key frame extraction module 120, configured to extract a plurality of cooling state key frames from the PPR pipe cooling state monitoring video; the image semantic coding module 130 is configured to obtain a plurality of cooling state feature vectors by performing image blocking processing on the plurality of material state key frames respectively and then passing through a ViT model including an embedded layer; the cooling state change feature extraction module 140 is configured to two-dimensionally arrange the plurality of cooling state feature vectors into a two-dimensional input matrix, and obtain a cooling state change feature vector through a convolutional neural network model serving as a filter; the flow characteristic time sequence encoding module 150 is configured to arrange flow values of the cooling water at the plurality of predetermined time points into flow input vectors according to a time dimension, and then obtain flow characteristic vectors through a time sequence encoder including a full-connection layer and a one-dimensional convolution layer; a responsiveness estimation module 160 for calculating a transition matrix of the cooling state change feature vector with respect to the flow feature vector as a classification feature matrix; the optimizing module 170 is configured to fuse the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, and calculate a position-wise point multiplication between the fused feature matrix and the classification feature matrix to obtain an optimized classification feature matrix; the control result generating module 180 is configured to pass the optimized classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate that the flow value of the cooling water at the current time point should be increased or decreased.
Fig. 2 is a schematic diagram of a PPR pipe cooling molding control system according to an embodiment of the present application. As shown in fig. 2, first, flow values of cooling water at a plurality of predetermined time points in a predetermined time period and PPR pipe cooling state monitoring videos of the predetermined time period acquired by a camera are acquired. Next, a plurality of cooling status key frames are extracted from the PPR tube cooling status monitor video. And then, respectively carrying out image blocking processing on the plurality of material state key frames, and then obtaining a plurality of cooling state feature vectors through a ViT model containing an embedded layer. Secondly, the plurality of cooling state characteristic vectors are arranged in two dimensions to form a two-dimensional input matrix, and then the two-dimensional input matrix is used as a convolution neural network model of a filter to obtain the cooling state change characteristic vector. Meanwhile, the flow values of the cooling water at a plurality of preset time points are arranged into flow input vectors according to the time dimension, and then the flow characteristic vectors are obtained through a time sequence encoder comprising a full-connection layer and a one-dimensional convolution layer. Then, a transition matrix of the cooling state change feature vector with respect to the flow feature vector is calculated as a classification feature matrix. And then fusing the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, and calculating the position-based point multiplication between the fused feature matrix and the classification feature matrix to obtain an optimized classification feature matrix. And finally, the optimized classification feature matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow value of cooling water at the current time point is increased or decreased.
In the above-mentioned PPR pipe cooling molding control system 100, the data acquisition module 110 is configured to acquire flow values of cooling water at a plurality of predetermined time points within a predetermined time period and PPR pipe cooling state monitoring videos acquired by a camera for the predetermined time period. As described in the background art above, in the process of preparing the PPR pipe, the raw material is first heated to soften the raw material and adapt to the shape of the forming mold, and then cooled to solidify and stabilize the pipe. During cooling, the flow rate of cooling water is an important factor affecting the quality of the PPR tubing. The excessive flow causes that the cooling is too fast and stress is easy to be generated in the pipe, so that the surface of the pipe is rough and spot pits are generated; the flow is too small, so that bright spots are generated on the surface of the pipe and are easy to break, and the wall thickness of the pipe is uneven.
Accordingly, the flow value of the cooling water is adaptively regulated and controlled according to the state change of the PPR pipe in consideration of the fact that the flow of the cooling water is actually controlled. That is, it is necessary to extract the correlation characteristic distribution information between the state change characteristic of the PPR pipe and the flow dynamic characteristic of the cooling water, and control the flow rate of the cooling water. Based on the above, in the technical scheme of the application, firstly, a monitoring video of a preset time period of the PPR pipe in the cooling process and flow values of cooling water at a plurality of preset time points in the preset time period are obtained, time sequence characteristic information of the flow values of the cooling water and state change characteristic information of the PPR pipe are extracted based on a deep learning technology, and real-time control of the flow of the cooling water is performed based on relevance characteristic representation of the time sequence characteristic information and the state change characteristic information of the PPR pipe. Therefore, the flow rate of cooling water can be monitored and regulated in real time so as to ensure the molding quality of the PPR pipe.
In the above-mentioned PPR pipe cooling molding control system 100, the key frame extracting module 120 is configured to extract a plurality of cooling state key frames from the PPR pipe cooling state monitoring video. Considering that a large number of similar frames exist in the PPR pipe cooling state monitoring video, in order to reduce the calculation amount and avoid adverse effects of data redundancy on subsequent feature extraction, a plurality of cooling state key frames are extracted from the PPR pipe cooling state monitoring video at a preset sampling frequency so as to capture and represent important states in the cooling process.
In the above PPR pipe cooling molding control system 100, the image semantic encoding module 130 is configured to perform image blocking processing on the plurality of material state key frames respectively, and obtain a plurality of cooling state feature vectors through a ViT model including an embedded layer. It should be understood that the ViT (Vision Transformer) model is a transducer-based neural network architecture, and has excellent performance in terms of implicit feature extraction of images. Specifically, firstly, the cooling state key frame is subjected to block processing to obtain a sequence of image blocks, so that local features in an image are captured better, and meanwhile, the problem that global features are excessively smoothed is avoided; then, the image block is converted into an embedded vector through an embedded layer, and the embedded vector is encoded by using a converter module of the ViT model to extract the implicit context semantic association feature distribution information of each cooling state key frame, so that the cooling state features of different positions and areas are better described and represented, and a more detailed and comprehensive information basis is provided for subsequent classification and decision.
Fig. 3 is a block diagram of an image semantic coding module in a PPR tubing cooling formation control system according to an embodiment of the present application. As shown in fig. 3, the image semantic coding module 130 includes: a blocking unit 131, configured to perform image blocking processing on the cooling state key frame to obtain a sequence of image blocks; an embedded coding unit 132, configured to perform embedded coding on each image block in the sequence of image blocks by using an embedded layer of the ViT model to obtain a sequence of image block embedded vectors; a transform coding unit 133 for inputting the sequence of image block embedded vectors into the converter module of the ViT model to obtain a sequence of image block feature vectors; a cascade unit 134, configured to cascade the sequence of image block feature vectors to obtain the cooling state feature vector.
Fig. 4 is a block diagram of a conversion coding unit in a PPR tubing cooling formation control system according to an embodiment of the present application. As shown in fig. 4, the transcoding unit 133 includes: an arrangement subunit 1331, configured to arrange the sequence of the image block embedding vectors into an input vector; a conversion subunit 1332, configured to convert the input vector into a query vector and a key vector through a learning embedding matrix; a self-attention subunit 1333 for calculating a product between the query vector and the transpose vector of the key vectors to obtain a self-attention correlation matrix; a normalization subunit 1334, configured to perform normalization processing on the self-attention correlation matrix to obtain a normalized self-attention correlation matrix; an attention computation subunit 1335, configured to activate the normalized self-attention correlation matrix input Softmax activation function to obtain a self-attention feature matrix; an attention applying subunit 1336 is configured to multiply the self-attention feature matrix with each of the image block embedding vectors in the sequence of image block embedding vectors to obtain the sequence of image block feature vectors.
In the above PPR pipe cooling molding control system 100, the cooling state change feature extraction module 140 is configured to two-dimensionally arrange the plurality of cooling state feature vectors into a two-dimensional input matrix, and obtain the cooling state change feature vector by using a convolutional neural network model as a filter. In order to capture time sequence dynamic change characteristics of the cooling state of the PPR pipe, the cooling state characteristic vectors are two-dimensionally arranged into a two-dimensional input matrix and processed through a Convolutional Neural Network (CNN) model serving as a filter so as to capture spatial correlation among the characteristics by utilizing a two-dimensional convolution operation, thereby obtaining the cooling state change characteristic vector. It should be appreciated that convolutional neural networks are a widely used deep learning model in computer vision tasks that can effectively extract local features in an image, which can be slid over an input matrix using convolutional kernels to obtain local features and patterns of variation. And by introducing proper convolution layers, pooling layers and activation functions into the convolution neural network, effective feature extraction and change analysis can be performed on the input matrix.
Accordingly, in one specific example, the cooling state change feature extraction module 140 is configured to: processing the two-dimensional input matrix by using the convolutional neural network model serving as a filter according to the following convolution formula to obtain the cooling state change feature vector;
wherein, the convolution formula is:
f i ＝GP{Sigmoid(N i ×f i-1 +B i )}
wherein f i-1 For input of the ith layer convolutional neural network model, f i For the output of the ith layer convolution neural network model, N i Filter for layer i convolutional neural network model, and B i For the bias matrix of the i-th layer convolutional neural network model, sigmoid represents a nonlinear activation function, and GP represents global feature pooling operation on each feature matrix of the feature map along the channel dimension.
In the above PPR pipe cooling molding control system 100, the flow characteristic time sequence encoding module 150 is configured to arrange the flow values of the cooling water at the plurality of predetermined time points into flow input vectors according to a time dimension, and then obtain the flow characteristic vectors by a time sequence encoder including a full connection layer and a one-dimensional convolution layer. As will be appreciated by those of ordinary skill in the art, a time series encoder is a special neural network model for time series data that captures the time series patterns and dynamic changes in the time series. The one-dimensional convolution layer of the time sequence encoder can capture the local relevance and the change mode of the flow characteristics; the full connection layer can conduct dimension transformation and mapping on the flow values so as to extract dynamic change characteristics of the flow values of the cooling water in a time sequence, and the model can understand the change relation between the flow values.
FIG. 5 is a block diagram of a flow feature timing encoding module in a PPR pipe cooling molding control system according to an embodiment of the present application. As shown in fig. 5, the traffic feature timing encoding module 150 includes: a normalization mapping unit 151, configured to perform normalization processing based on a maximum value on the flow input vector to obtain a normalized flow input vector; a full-connection encoding unit 152, configured to perform full-connection encoding on the normalized traffic input vector by using a full-connection layer of the timing encoder, so as to extract high-dimensional implicit features of feature values of each position in the normalized traffic input vector; a one-dimensional convolution encoding unit 153, configured to perform one-dimensional encoding on the normalized traffic input vector using a one-dimensional convolution layer of the timing encoder, so as to extract high-dimensional implicit correlation features related to feature values of each position in the normalized traffic input vector.
In the PPR pipe cooling molding control system 100, the responsiveness estimation module 160 is configured to calculate a transition matrix of the cooling state change feature vector with respect to the flow feature vector as a classification feature matrix. In order to capture the correlation between the cooling state change and the flow characteristics, a transition matrix of the cooling state change characteristic vector relative to the flow characteristic vector is further calculated to combine the change pattern of the cooling state with the flow characteristics, thereby providing more comprehensive classification information. It should be appreciated that the transition matrix represents a matrix of transition probabilities from one state to another, and by calculating the transition matrix, it is possible to analyze how the change in cooling state is affected by the flow characteristics.
Accordingly, in one specific example, the responsiveness estimation module 160 is configured to: calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as the classification feature matrix according to the following response formula;
wherein, the response formula is:
wherein V is 1 Representing the cooling state change feature vector, V 2 The flow characteristic vector is represented as such,representing matrix multiplication, M representing the classification feature matrix.
In the above-mentioned PPR pipe cooling molding control system 100, the optimization module 170 is configured to fuse the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, and calculate a position-wise multiplication between the fused feature matrix and the classification feature matrix to obtain an optimized classification feature matrix. Further, the state and flow rate changes of the cooling system are described from different angles, respectively, taking into account the cooling state change feature vector and the flow rate feature vector. Fusing the two feature vectors can combine their information to provide a richer, more comprehensive representation of the feature. This helps capture more relevant features, improving the model's understanding and expressive power of the data. The cooling state change feature vector and the flow feature vector typically contain different information. Fusing them together may provide complementary information that more fully describes the state and characteristics of the cooling system. Fusing different types of feature vectors can provide more information for classification model learning. This helps to improve the performance of the classification model, enabling it to better distinguish between different classes or states. By fusing the cooling state change feature vector and the flow feature vector, the relativity and interaction between the cooling state change feature vector and the flow feature vector can be utilized, and the discrimination capability of the classification model on the cooling system state can be improved.
In particular, considering that the cooling state change feature vector and the flow feature vector are obtained by differently encoding data of a non-use source, the two have different dimensions, and if directly fused, information is lost due to dimension mismatch. Vectors of different dimensions typically represent different features or attributes. When they are fused together, some important characteristic information may be lost, thereby affecting subsequent analysis or application. At the same time, vectors of different dimensions tend to have different semantic representations. Fusing them directly may lead to problems of semantic inconsistency. Fusing them directly may result in confusing or erroneous semantic representations. Therefore, the cooling state change feature vector and the flow feature vector with different dimensions are further mapped into the subspace with the same dimension to be fused by utilizing the property of affine subspace mapping.
Specifically, the optimization module 170 includes: the normalization unit is used for normalizing the cooling state change feature vector and the flow feature vector to obtain a normalized cooling state change feature vector and a normalized flow feature vector; an included angle calculating unit, configured to calculate an included angle value between the cooling state change feature vector and the flow feature vector; a rotation matrix construction unit configured to construct a rotation matrix in a rotation formula based on the angle value such that the rotation matrix multiplied by the normalized cooling state change feature vector is equal to the normalized flow feature vector, wherein the rotation formula is:
Wherein R is the rotation matrix, theta is the included angle value between the cooling state change feature vector and the flow feature vector, sin theta is a sine function, and cos theta is a cosine function; an affine transformation function constructing unit for controlling linear interpolation between the rotation matrix and the identity matrix in an affine transformation formula based on a predetermined affine transformation superparameter, wherein the affine transformation formula is:
f(α)＝(1-α)I+αR
wherein alpha is a preset affine transformation super parameter, f (alpha) is the affine transformation function, I is the identity matrix, and R is the rotation matrix; an affine transformation unit for applying the affine transformation function to the cooling state change feature vector and the flow feature vector, respectively, to obtain an affine transformed cooling state change feature vector and an affine transformed flow feature vector; and the fusion unit is used for calculating the product between the affine transformation cooling state change feature vector and the transpose vector of the affine transformation flow feature vector to obtain the fusion feature matrix.
In this way, by utilizing the property of affine subspace mapping, the cooling state change feature vectors and the flow feature vectors with different dimensions are mapped into subspaces with the same dimension, namely, the affine transformation subspaces are used as a common medium space to fuse the cooling state change feature vectors and the flow feature vectors with different dimensions, so that the fused feature matrix can reduce redundancy of features, reduce the dimension of data and simultaneously can contain the information of two original feature vectors. And, through carrying out feature fusion among feature vectors based on affine subspace mapping, the robustness of a fusion feature matrix can be enhanced, so that the fusion feature matrix can resist noise and abnormal value interference, and the quality of data is improved.
And after the fusion feature matrix is obtained, multiplying the position points between the fusion feature matrix and the classification feature matrix to obtain an optimized classification feature matrix so as to improve the classification judgment accuracy of the optimized classification feature matrix.
In the above PPR pipe cooling molding control system 100, the control result generating module 180 is configured to pass the optimized classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate that the flow value of the cooling water at the current time point should be increased or decreased. The classifier is a machine learning model capable of learning the mapping relationship from input features to output labels. The classifier after training can predict whether the cooling water flow value should be increased or decreased according to the input classification characteristic matrix, and output a corresponding classification result. Therefore, the cooling system can be automatically controlled based on the classification result, so that the purposes of optimizing the cooling effect and saving energy are achieved.
In summary, the PPR pipe cooling forming control system according to the embodiment of the present application is explained, firstly, a monitoring video of a predetermined time period of the PPR pipe in a cooling process and flow values of cooling water at a plurality of predetermined time points in the predetermined time period are obtained, time sequence characteristic information of the cooling water flow value and state change characteristic information of the PPR pipe are extracted based on a deep learning technology, and real-time control of the cooling water flow is performed based on correlation characteristic representation of the time sequence characteristic information and the state change characteristic information of the PPR pipe. Therefore, the flow rate of cooling water can be monitored and regulated in real time so as to ensure the molding quality of the PPR pipe.
Exemplary method
Fig. 6 is a flowchart of a PPR tubing cooling molding control method according to an embodiment of the present application. As shown in fig. 6, the PPR pipe cooling molding control method according to the embodiment of the application includes the steps of: s110, acquiring flow values of cooling water at a plurality of preset time points in a preset time period and PPR pipe cooling state monitoring videos of the preset time period acquired by a camera; s120, extracting a plurality of cooling state key frames from the PPR pipe cooling state monitoring video; s130, respectively performing image blocking processing on the plurality of material state key frames, and then obtaining a plurality of cooling state feature vectors through a ViT model containing an embedded layer; s140, two-dimensionally arranging the plurality of cooling state feature vectors into a two-dimensional input matrix, and then obtaining a cooling state change feature vector through a convolutional neural network model serving as a filter; s150, arranging the flow values of the cooling water at a plurality of preset time points into flow input vectors according to a time dimension, and then obtaining flow characteristic vectors through a time sequence encoder comprising a full-connection layer and a one-dimensional convolution layer;
s160, calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as a classification feature matrix; s170, fusing the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, and calculating the position-based point multiplication between the fused feature matrix and the classification feature matrix to obtain an optimized classification feature matrix; and S180, the optimized classification feature matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow value of cooling water at the current time point is increased or decreased.
In a specific example, in the above PPR pipe cooling molding control method, the step S120 of extracting a plurality of cooling state key frames from the PPR pipe cooling state monitoring video includes: a plurality of cooling state key frames are extracted from the PPR tube cooling state monitoring video at a predetermined sampling frequency.
In a specific example, in the above PPR pipe cooling molding control method, the step S130, after performing image blocking processing on the plurality of material state key frames respectively, obtains a plurality of cooling state feature vectors through a ViT model including an embedded layer, includes: performing image blocking processing on the cooling state key frame to obtain a sequence of image blocks; using the embedding layer of the ViT model to respectively carry out embedded coding on each image block in the sequence of the image blocks so as to obtain a sequence of image block embedded vectors; inputting the sequence of image block embedded vectors into a converter module of the ViT model to obtain a sequence of image block feature vectors; and cascading the sequence of image block feature vectors to obtain the cooling state feature vector.
In a specific example, in the above PPR pipe cooling molding control method, inputting the sequence of image block embedded vectors into the converter module of the ViT model to obtain the sequence of image block feature vectors includes: arranging the sequence of the image block embedded vectors into an input vector; converting the input vector into a query vector and a key vector through a learning embedding matrix; calculating the product between the query vector and the transpose vector of the key vector to obtain a self-attention correlation matrix; carrying out standardization processing on the self-attention association matrix to obtain a standardized self-attention association matrix; inputting the standardized self-attention association matrix into a Softmax activation function to activate so as to obtain a self-attention feature matrix; and multiplying the self-attention characteristic matrix with each image block embedded vector in the sequence of image block embedded vectors respectively to obtain the sequence of image block characteristic vectors.
In a specific example, in the above PPR pipe cooling molding control method, the step S140 of two-dimensionally arranging the plurality of cooling state feature vectors into a two-dimensional input matrix and obtaining the cooling state change feature vector by using a convolutional neural network model as a filter includes: processing the two-dimensional input matrix by using the convolutional neural network model serving as a filter according to the following convolution formula to obtain the cooling state change feature vector;
wherein, the convolution formula is:
f i ＝GP{Sigmoid(N i ×f i-1 +B i )}
wherein f i-1 For input of the ith layer convolutional neural network model, f i For the output of the ith layer convolution neural network model, N i Filter for layer i convolutional neural network model, and B i For the bias matrix of the i-th layer convolutional neural network model, sigmoid represents a nonlinear activation function, and GP represents global feature pooling operation on each feature matrix of the feature map along the channel dimension.
In a specific example, in the above PPR pipe cooling molding control method, the step S150 is to arrange the flow values of the cooling water at the plurality of predetermined time points into flow input vectors according to a time dimension, and then obtain flow feature vectors by a time sequence encoder including a full connection layer and a one-dimensional convolution layer, and includes: carrying out maximum value-based normalization processing on the flow input vector to obtain a normalized flow input vector; performing full-connection coding on the normalized flow input vector by using a full-connection layer of the time sequence coder to extract high-dimensional implicit features of feature values of all positions in the normalized flow input vector; and carrying out one-dimensional coding on the normalized flow input vector by using a one-dimensional convolution layer of the time sequence coder so as to extract associated high-dimensional implicit association features among feature values of all positions in the normalized flow input vector.
In a specific example, in the above PPR pipe cooling molding control method, the step S160 of calculating a transition matrix of the cooling state change feature vector with respect to the flow feature vector as a classification feature matrix includes: calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as the classification feature matrix according to the following response formula;
wherein, the response formula is:
wherein V is 1 Representing the cooling state change feature vector, V 2 The flow characteristic vector is represented as such,representing matrix multiplication, M representing the classification feature matrix.
Here, it will be understood by those skilled in the art that the specific operations of the respective steps in the above-described PPR pipe cooling molding control method have been described in detail in the above description of the PPR pipe cooling molding control system with reference to fig. 1 to 5, and thus, repetitive descriptions thereof will be omitted.
Claims (10)
1. A PPR tubing cooling formation control system, comprising:
the data acquisition module is used for acquiring flow values of cooling water at a plurality of preset time points in a preset time period and PPR pipe cooling state monitoring videos of the preset time period acquired by the camera;
The key frame extraction module is used for extracting a plurality of cooling state key frames from the PPR pipe cooling state monitoring video;
the image semantic coding module is used for obtaining a plurality of cooling state feature vectors through a ViT model containing an embedded layer after performing image blocking processing on the plurality of material state key frames respectively;
the cooling state change feature extraction module is used for two-dimensionally arranging the plurality of cooling state feature vectors into a two-dimensional input matrix and then obtaining the cooling state change feature vectors through a convolutional neural network model serving as a filter;
the flow characteristic time sequence coding module is used for arranging the flow values of the cooling water at a plurality of preset time points into flow input vectors according to time dimensions and then obtaining flow characteristic vectors through a time sequence coder comprising a full-connection layer and a one-dimensional convolution layer;
the responsiveness estimation module is used for calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as a classification feature matrix;
the optimizing module is used for fusing the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, and calculating the position-based point multiplication between the fused feature matrix and the classification feature matrix to obtain an optimized classification feature matrix;
And the control result generation module is used for enabling the optimized classification characteristic matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow value of the cooling water at the current time point is increased or decreased.
2. The PPR tube cooling molding control system according to claim 1, wherein the key frame extraction module is configured to: a plurality of cooling state key frames are extracted from the PPR tube cooling state monitoring video at a predetermined sampling frequency.
3. The PPR tube cooling molding control system according to claim 2, wherein said image semantic coding module comprises:
the blocking unit is used for carrying out image blocking processing on the cooling state key frame to obtain a sequence of image blocks;
the embedded coding unit is used for respectively carrying out embedded coding on each image block in the sequence of the image blocks by using an embedded layer of the ViT model so as to obtain a sequence of embedded vectors of the image blocks;
a conversion coding unit, configured to input the sequence of image block embedded vectors into a converter module of the ViT model to obtain a sequence of image block feature vectors;
and the cascading unit is used for cascading the sequence of the image block feature vectors to obtain the cooling state feature vector.
4. The PPR tube cooling forming control system according to claim 3, wherein said conversion coding unit comprises:
an arrangement subunit, configured to arrange the sequence of the image block embedding vectors into an input vector;
the conversion subunit is used for converting the input vector into a query vector and a key vector respectively through a learning embedding matrix;
a self-attention subunit, configured to calculate a product between the query vector and a transpose vector of the key vector to obtain a self-attention correlation matrix;
a normalization subunit, configured to perform normalization processing on the self-attention association matrix to obtain a normalized self-attention association matrix;
an attention calculating subunit, configured to activate the normalized self-attention association matrix input Softmax activation function to obtain a self-attention feature matrix;
and the attention applying subunit is used for multiplying the self-attention characteristic matrix with each image block embedding vector in the sequence of the image block embedding vectors respectively to obtain the sequence of the image block characteristic vectors.
5. The PPR tube cooling molding control system according to claim 4, wherein said cooling state change feature extraction module is configured to: processing the two-dimensional input matrix by using the convolutional neural network model serving as a filter according to the following convolution formula to obtain the cooling state change feature vector;
Wherein, the convolution formula is:
f i ＝GP{Sigmoid(N i ×f i-1 +B i )}
wherein f i-1 For input of the ith layer convolutional neural network model, f i For the output of the ith layer convolution neural network model, N i Filter for layer i convolutional neural network model, and B i For the bias matrix of the i-th layer convolutional neural network model, sigmoid represents a nonlinear activation function, and GP represents global feature pooling operation on each feature matrix of the feature map along the channel dimension.
6. The PPR tubing cooling molding control system according to claim 5, wherein said flow signature timing encoding module comprises:
the normalization mapping unit is used for carrying out normalization processing based on the maximum value on the flow input vector so as to obtain a normalized flow input vector;
the full-connection coding unit is used for carrying out full-connection coding on the normalized flow input vector by using a full-connection layer of the time sequence coder so as to extract high-dimensional implicit characteristics of characteristic values of all positions in the normalized flow input vector;
and the one-dimensional convolution coding unit is used for carrying out one-dimensional coding on the normalized flow input vector by using a one-dimensional convolution layer of the time sequence coder so as to extract associated high-dimensional implicit association features among feature values of all positions in the normalized flow input vector.
7. The PPR tube cooling molding control system of claim 6, wherein the responsiveness estimation module is to: calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as the classification feature matrix according to the following response formula;
wherein, the response formula is:
wherein V is 1 Representing the cooling state change feature vector, V 2 The flow characteristic vector is represented as such,representing matrix multiplication, M representing the classification feature matrix.
8. The PPR tubing cooling molding control system according to claim 7, wherein said optimization module comprises:
the normalization unit is used for normalizing the cooling state change feature vector and the flow feature vector to obtain a normalized cooling state change feature vector and a normalized flow feature vector;
an included angle calculating unit, configured to calculate an included angle value between the cooling state change feature vector and the flow feature vector;
a rotation matrix construction unit configured to construct a rotation matrix in a rotation formula based on the angle value such that the rotation matrix multiplied by the normalized cooling state change feature vector is equal to the normalized flow feature vector, wherein the rotation formula is:
Wherein R is the rotation matrix, theta is the included angle value between the cooling state change feature vector and the flow feature vector, sin theta is a sine function, and cos theta is a cosine function;
an affine transformation function constructing unit for controlling linear interpolation between the rotation matrix and the identity matrix in an affine transformation formula based on a predetermined affine transformation superparameter, wherein the affine transformation formula is:
f(α)＝(1-α)I+αR
wherein alpha is a preset affine transformation super parameter, f (alpha) is the affine transformation function, I is the identity matrix, and R is the rotation matrix;
an affine transformation unit for applying the affine transformation function to the cooling state change feature vector and the flow feature vector, respectively, to obtain an affine transformed cooling state change feature vector and an affine transformed flow feature vector;
and the fusion unit is used for calculating the product between the affine transformation cooling state change feature vector and the transpose vector of the affine transformation flow feature vector to obtain the fusion feature matrix.
9. The PPR pipe cooling forming control method is characterized by comprising the following steps of:
acquiring flow values of cooling water at a plurality of preset time points in a preset time period and a PPR pipe cooling state monitoring video of the preset time period acquired by a camera;
Extracting a plurality of cooling state key frames from the PPR pipe cooling state monitoring video;
image blocking processing is carried out on the material state key frames respectively, and then a ViT model containing an embedded layer is used for obtaining a plurality of cooling state feature vectors;
the plurality of cooling state feature vectors are two-dimensionally arranged into a two-dimensional input matrix and then pass through a convolutional neural network model serving as a filter to obtain a cooling state change feature vector;
arranging the flow values of the cooling water at a plurality of preset time points into flow input vectors according to a time dimension, and then obtaining flow characteristic vectors through a time sequence encoder comprising a full-connection layer and a one-dimensional convolution layer;
calculating a transfer matrix of the cooling state change feature vector relative to the flow feature vector as a classification feature matrix;
fusing the cooling state change feature vector and the flow feature vector to obtain a fused feature matrix, and calculating the position-based point multiplication between the fused feature matrix and the classification feature matrix to obtain an optimized classification feature matrix;
and the optimized classification characteristic matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow value of cooling water at the current time point is increased or decreased.
10. The PPR tube cooling molding control method according to claim 9, wherein extracting a plurality of cooling state key frames from the PPR tube cooling state monitoring video comprises: a plurality of cooling state key frames are extracted from the PPR tube cooling state monitoring video at a predetermined sampling frequency.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311070778.5A CN117021434A (en) | 2023-08-23 | 2023-08-23 | PPR pipe cooling forming control system and method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311070778.5A CN117021434A (en) | 2023-08-23 | 2023-08-23 | PPR pipe cooling forming control system and method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117021434A true CN117021434A (en) | 2023-11-10 |
Family
ID=88638998
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311070778.5A Withdrawn CN117021434A (en) | 2023-08-23 | 2023-08-23 | PPR pipe cooling forming control system and method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117021434A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117404853A (en) * | 2023-12-14 | 2024-01-16 | 山西省水利建筑工程局集团有限公司 | External circulating water cooling system and method for tunnel boring machine |
-
2023
- 2023-08-23 CN CN202311070778.5A patent/CN117021434A/en not_active Withdrawn
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117404853A (en) * | 2023-12-14 | 2024-01-16 | 山西省水利建筑工程局集团有限公司 | External circulating water cooling system and method for tunnel boring machine |
| CN117404853B (en) * | 2023-12-14 | 2024-03-08 | 山西省水利建筑工程局集团有限公司 | External circulating water cooling system and method for tunnel boring machine |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lin et al. | EAPT: efficient attention pyramid transformer for image processing | |
| CN117021434A (en) | PPR pipe cooling forming control system and method thereof | |
| CN114202740A (en) | Pedestrian re-identification method based on multi-scale feature fusion | |
| CN113591770A (en) | Multimode fusion obstacle detection method and device based on artificial intelligence blind guiding | |
| CN111695523B (en) | Double-flow convolutional neural network action recognition method based on skeleton space-time and dynamic information | |
| An et al. | LPViT: a transformer based model for PCB image classification and defect detection | |
| Jeon et al. | ABCD: Attentive bilateral convolutional network for robust depth completion | |
| CN115719298A (en) | Spinning elasticizer | |
| CN113297972A (en) | Transformer substation equipment defect intelligent analysis method based on data fusion deep learning | |
| CN117058456A (en) | Visual target tracking method based on multiphase attention mechanism | |
| CN116051948A (en) | Fine granularity image recognition method based on attention interaction and anti-facts attention | |
| Fu et al. | Pt-flownet: Scene flow estimation on point clouds with point transformer | |
| CN116452862A (en) | Image classification method based on domain generalization learning | |
| Min et al. | Geometry guided network for point cloud registration | |
| CN117227005A (en) | Production control system and method for concrete raw material processing | |
| CN116890405A (en) | Control system and method for plastic particle production | |
| CN116533529A (en) | Intelligent control method and system for ultrasonic welding PC (polycarbonate) sheet | |
| CN112365456B (en) | Transformer substation equipment classification method based on three-dimensional point cloud data | |
| CN114998904A (en) | Sketch semantic segmentation method based on class encoding and decoding structure | |
| He et al. | SCIR-Net: Structured color image representation based 3D object detection network from point clouds | |
| CN117140775A (en) | Automatic production equipment for waterproof material | |
| CN117105251A (en) | Intelligent preparation method of nanoscale aluminum oxide | |
| Chen et al. | Dual Attention Network for Unsupervised Domain Adaptive Person Re-identification | |
| Lin et al. | Deep Fusion for Multi-Modal 6D Pose Estimation | |
| CN116228825B (en) | Point cloud registration method based on significant anchor point geometric embedding |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20231110 |
|
| WW01 | Invention patent application withdrawn after publication |