CN115291646B - Energy management control system for lithium fluoride preparation and control method thereof - Google Patents

Energy management control system for lithium fluoride preparation and control method thereof Download PDF

Info

Publication number
CN115291646B
CN115291646B CN202210800838.3A CN202210800838A CN115291646B CN 115291646 B CN115291646 B CN 115291646B CN 202210800838 A CN202210800838 A CN 202210800838A CN 115291646 B CN115291646 B CN 115291646B
Authority
CN
China
Prior art keywords
vector
feature
characteristic
temperature
product
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
Application number
CN202210800838.3A
Other languages
Chinese (zh)
Other versions
CN115291646A (en
Inventor
华博文
雷炎芳
廖育能
黄吉华
陈三凤
吴仕显
胡新
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Fujian Longfu New Material Co ltd
Original Assignee
Fujian Longfu New Material Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Fujian Longfu New Material Co ltd filed Critical Fujian Longfu New Material Co ltd
Priority to CN202210800838.3A priority Critical patent/CN115291646B/en
Priority to PCT/CN2022/117768 priority patent/WO2024007445A1/en
Publication of CN115291646A publication Critical patent/CN115291646A/en
Application granted granted Critical
Publication of CN115291646B publication Critical patent/CN115291646B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D23/00—Control of temperature
    • G05D23/19—Control of temperature characterised by the use of electric means
    • G05D23/20—Control of temperature characterised by the use of electric means with sensing elements having variation of electric or magnetic properties with change of temperature

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Automation & Control Theory (AREA)
  • Analysing Materials By The Use Of Radiation (AREA)
  • Battery Electrode And Active Subsutance (AREA)

Abstract

The application relates to the field of intelligent control of preparation, and particularly discloses an energy management control system for lithium fluoride preparation and a control method thereof.

Description

Energy management control system for lithium fluoride preparation and control method thereof
Technical Field
The invention relates to the field of intelligent control of preparation, in particular to an energy management control system for lithium fluoride preparation and a control method thereof.
Background
Lithium fluoride is an important lithium-based material and is a white, non-hygroscopic cubic crystal at normal temperature. As an important inorganic fluoride, high-purity lithium fluoride is widely used for the preparation of fluorinated glass and optical fibers. Meanwhile, the high-purity lithium fluoride is also an important raw material of an electrolyte material for a lithium ion battery.
The commonly used preparation methods of lithium fluoride mainly comprise a direct preparation method, an ion exchange preparation method and an extraction preparation method. Regardless of the preparation scheme, energy management during the preparation process is a very important concern, especially in the current large background of energy conservation and environmental protection.
In particular, in the process of preparing lithium fluoride, temperature control such as calcination, cooling and the like is mostly involved, and it should be understood that the temperature control not only relates to energy management, but also relates to the quality of the final lithium fluoride product. In the existing lithium fluoride preparation schemes, a technical scheme of preset temperature is mostly adopted, which is not only unfavorable for energy optimization, but also unfavorable for controlling the quality of the final lithium fluoride finished product.
Therefore, an energy management control system for lithium fluoride preparation is desired to intelligently control the temperature of the calcinator so as to ensure the quality of the finished lithium fluoride product while optimizing energy.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides an energy management control system for lithium fluoride preparation and a control method thereof, which dynamically extract the temperature correlation characteristics inside a calcinator in real time through a convolutional neural network model based on deep learning, deeply excavate the structural change characteristics and the internal heat distribution characteristics of a calcinated product, and then intelligently adjust the temperature of the calcinator by combining the characteristic information of the three on the time sequence, so as to ensure the finished product quality of final lithium fluoride while optimizing energy.
According to one aspect of the present application, there is provided an energy management control system for lithium fluoride production, comprising: the calcining temperature acquisition module is used for acquiring calcining temperature values of the calciner at a plurality of preset time points through a temperature sensor arranged in the calciner, wherein the calciner is used for calcining crystals forming lithium fluoride finished products; the product data acquisition module is used for acquiring thermal infrared images and X-Ray scanning charts of the calcined products at the plurality of preset time points through a thermal infrared camera and an X-Ray scanner which are arranged in the calciner; a thermal infrared coding module for passing the thermal infrared images of the calcinated products at the plurality of predetermined time points through a first convolution neural network using a spatial attention mechanism to obtain a thermal infrared characteristic map; a perspective view coding module for passing the X-Ray scans of the calcination products at the plurality of predetermined time points through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray feature map; the correlation coding module is used for cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram and then obtaining a product characteristic vector through a third convolutional neural network serving as a filter; the energy data coding module is used for enabling the calcinations temperature values of the calcinator at a plurality of preset time points to pass through a time sequence coder comprising a one-dimensional convolution layer and a full connection layer so as to obtain a temperature characteristic vector; the characteristic fusion module is used for fusing the product characteristic vector and the temperature characteristic vector to obtain a classification characteristic vector; and the energy management result generation module is used for enabling the classification characteristic vectors to pass through a classifier to obtain a classification result, and the classification result is used for indicating that the temperature of the calcinator at the current time point should be increased or decreased.
In the above energy management control system for lithium fluoride preparation, the thermal infrared coding module is further configured to perform, in layer forward direction, input data using each layer of the first convolutional neural network: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram; pooling the convolution feature map to generate a pooled feature map; performing activation processing on the pooled feature map to generate an activated feature map; performing global average pooling along a channel dimension on the activation feature map to obtain a spatial feature matrix; performing convolution processing and activation processing on the spatial feature matrix to generate a weight vector; weighting each feature matrix of the activation feature map by using the weight value of each position in the weight vector to obtain a generated feature map; wherein the generated feature map output by the last layer of the first convolutional neural network model is the thermal infrared feature map.
In the above energy management control system for lithium fluoride production, the perspective view coding module is further configured to perform the following operations on the input data in the forward direction transfer of layers by the second convolutional neural network using the three-dimensional convolutional kernel: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram; pooling the convolution characteristic map to obtain a pooled characteristic map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the second convolutional neural network is the X-Ray characteristic diagram, and the input of the first layer of the second convolutional neural network is the X-Ray scanning diagram of the calcination products at the plurality of preset time points.
In the energy management control system for preparing lithium fluoride, the correlation coding module and the cascade unit are used for cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram to obtain a cascade characteristic diagram; and (d). A feature extraction unit configured to perform convolution processing, pooling processing along a feature matrix, and activation processing on input data in forward pass of layers using layers of the third convolutional neural network as a filter to generate the product feature vector from a last layer of the third convolutional neural network, wherein an input of the first layer of the third convolutional neural network is the concatenated feature map.
In the above energy management control system for lithium fluoride production, the energy data encoding module includes: the input vector construction unit is used for arranging the calcination temperature values of the calciner at a plurality of preset time points into a one-dimensional input vector according to the time dimension; a full-concatenation coding unit, configured to perform full-concatenation coding on the input vector using a full-concatenation layer of the time sequence encoder according to the following formula to extract a high-dimensional implicit feature of a feature value at each position in the input vector, where the formula is:
Figure 674808DEST_PATH_IMAGE001
wherein
Figure 363147DEST_PATH_IMAGE002
Is the input vector of the said one or more input vectors,
Figure 842670DEST_PATH_IMAGE003
is the output vector of the output vector,
Figure 593981DEST_PATH_IMAGE004
is a matrix of the weights that is,
Figure 831059DEST_PATH_IMAGE005
is a vector of the offset to be used,
Figure 73821DEST_PATH_IMAGE006
represents a matrix multiplication; a one-dimensional convolution encoding unit for using the one-dimensional convolution layer of the time-series encoderPerforming one-dimensional convolution coding on the input vector by using the following formula to extract high-dimensional implicit correlation characteristics among characteristic values of all positions in the input vector, wherein the formula is as follows:
Figure 578490DEST_PATH_IMAGE007
wherein, the first and the second end of the pipe are connected with each other,ais a convolution kernelxWidth in the direction,FIs a convolution kernel parameter vector,GIs a local vector matrix that operates with a convolution kernel,wthe size of the convolution kernel.
In the above energy management control system for lithium fluoride production, the feature fusion module includes: a vector difference unit for calculating a difference feature vector between the product feature vector and the temperature feature vector; the logarithm operation unit is used for calculating the logarithm function value of the characteristic value of each position in the difference characteristic vector to obtain a logarithm difference characteristic vector; the depth information characteristic value calculation unit is used for calculating a norm of the differential feature vector; the full-scene homography incidence matrix constructing unit is used for calculating the product between the product characteristic vector and the transposed vector of the temperature characteristic vector to obtain a full-field Jing Shanying incidence matrix; the depth perception unit is used for calculating the Frobenius norm of the full-scene homography incidence matrix; an alignment unit, configured to process the log-difference eigenvector by using a norm of the difference eigenvector as a weighting coefficient and using a Frobenius norm of the full-field Jing Shanying correlation matrix as a bias term to obtain a corrected temperature eigenvector that is homologically aligned with the product eigenvector in depth; and the fusion unit is used for calculating the weighted sum of the corrected temperature characteristic vector and the product characteristic vector according to the position to obtain the classification characteristic vector.
In the above energy management control system for lithium fluoride production, the alignment unit is configured to apply the logarithm of the log with the one norm of the differential eigenvector as a weighting coefficient and the Frobenius norm of the full-field Jing Shanying correlation matrix as a bias term according to the following formulaProcessing the differential characteristic vector to obtain the corrected temperature characteristic vector; wherein the formula is:
Figure 858292DEST_PATH_IMAGE008
wherein, the first and the second end of the pipe are connected with each other,
Figure 806833DEST_PATH_IMAGE009
represents a norm of the differential feature vector, and
Figure 463073DEST_PATH_IMAGE010
a Frobenius norm representing the full scene homographic incidence matrix,
Figure 963194DEST_PATH_IMAGE011
showing dot-by-dot multiplication according to position,
Figure 476215DEST_PATH_IMAGE012
Indicating that the addition is made by position,
Figure 658191DEST_PATH_IMAGE013
indicating subtraction by position.
In the above energy management control system for lithium fluoride preparation, the energy management result generation module is further configured to: processing the classification feature vector using the classifier to obtain the classification result with the following formula:
Figure 586963DEST_PATH_IMAGE014
wherein, in the process,
Figure 535066DEST_PATH_IMAGE015
to
Figure 891092DEST_PATH_IMAGE016
In order to be a weight matrix, the weight matrix,
Figure 825943DEST_PATH_IMAGE017
to
Figure 558407DEST_PATH_IMAGE018
In order to be a vector of the offset,
Figure 236382DEST_PATH_IMAGE019
the classified feature vector is obtained.
In the above energy management control system for lithium fluoride production, the system further comprises an energy control module for adjusting the temperature of the calciner based on the classification result.
According to another aspect of the present application, a control method of an energy management control system for lithium fluoride production, includes: acquiring calcination temperature values of a calciner at a plurality of preset time points through a temperature sensor arranged in the calciner, wherein the calciner is used for calcining crystals forming lithium fluoride finished products; acquiring thermal infrared images and X-Ray scanning diagrams of the calcined products at the plurality of preset time points through a thermal infrared camera and an X-Ray scanner which are arranged in the calciner; passing the thermal infrared images of the calcined product at the plurality of predetermined time points through a first convolutional neural network using a spatial attention mechanism to obtain a thermal infrared signature; passing the X-Ray scans of the calcined product at the plurality of predetermined time points through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray profile; cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram, and then passing through a third convolutional neural network serving as a filter to obtain a product characteristic vector; passing the calcination temperature values of the calciner at a plurality of preset time points through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer to obtain a temperature characteristic vector; fusing the product feature vector and the temperature feature vector to obtain a classification feature vector; and passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the calciner at the current time point should be increased or decreased.
In the above control method of the energy management control system for lithium fluoride production, passing the thermal infrared images of the calcinated product at the plurality of predetermined time points through a first convolution neural network using a spatial attention mechanism to obtain a thermal infrared profile, comprising: performing, using layers of the first convolutional neural network, in forward pass of layers, input data: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram; pooling the convolved feature map to generate a pooled feature map; performing activation processing on the pooled feature map to generate an activated feature map; performing global average pooling of the activation signature along a channel dimension to obtain a spatial signature matrix; performing convolution processing and activation processing on the spatial feature matrix to generate a weight vector; weighting each feature matrix of the activation feature map by using the weight value of each position in the weight vector to obtain a generated feature map; wherein the generated feature map output by the last layer of the first convolutional neural network model is the thermal infrared feature map.
In the above control method of the energy management control system for lithium fluoride production, passing the X-Ray scans of the calcined product at the plurality of predetermined time points through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray profile, comprising: the second convolutional neural network using the three-dimensional convolutional kernel performs, in forward transfer of layers, respectively: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram; pooling the convolution characteristic map to obtain a pooled characteristic map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the second convolutional neural network is the X-Ray characteristic diagram, and the input of the first layer of the second convolutional neural network is the X-Ray scanning diagram of the calcination products at the plurality of preset time points.
In the above control method of the energy management control system for lithium fluoride production, cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram, and then passing through a third convolutional neural network as a filter to obtain a product characteristic vector, includes: cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram to obtain a cascading characteristic diagram; and performing convolution processing, pooling processing along a feature matrix, and activation processing on input data in forward pass of layers using layers of the third convolutional neural network as a filter to generate the product feature vector from a last layer of the third convolutional neural network, wherein an input of a first layer of the third convolutional neural network is the concatenated feature map.
In the above control method of the energy management control system for lithium fluoride production, passing the calcination temperature values of the calciner at a plurality of predetermined time points through a time-sequence encoder comprising a one-dimensional convolution layer and a full link layer to obtain a temperature characteristic vector, the method comprises: arranging the calcination temperature values of the calciner at a plurality of preset time points into a one-dimensional input vector according to the time dimension; using a full-connection layer of the time sequence encoder to perform full-connection encoding on the input vector by using the following formula to extract high-dimensional implicit features of feature values of all positions in the input vector, wherein the formula is as follows:
Figure 153522DEST_PATH_IMAGE020
wherein
Figure 575670DEST_PATH_IMAGE021
Is the input vector of the said one or more input vectors,
Figure 111824DEST_PATH_IMAGE022
is the output vector of the digital video signal,
Figure 565677DEST_PATH_IMAGE023
is a matrix of weights that is a function of,
Figure 997927DEST_PATH_IMAGE024
is a vector of the offset to be used,
Figure 60034DEST_PATH_IMAGE025
represents a matrix multiplication; performing one-dimensional convolutional coding on the input vector by using a one-dimensional convolutional layer of the time sequence encoder according to the following formula to extract high-dimensional implicit correlation characteristics among characteristic values of all positions in the input vector, wherein the formula is as follows:
Figure 993355DEST_PATH_IMAGE026
wherein the content of the first and second substances,ais a convolution kernelxWidth in the direction,FIs a convolution kernel parameter vector,GIs a matrix of local vectors operating with a convolution kernel,wthe size of the convolution kernel.
In the above control method of an energy management control system for lithium fluoride production, fusing the product eigenvector and the temperature eigenvector to obtain a classification eigenvector, including: calculating a difference feature vector between the product feature vector and the temperature feature vector; calculating a logarithmic function value of the characteristic value of each position in the differential characteristic vector to obtain a logarithmic differential characteristic vector; calculating a norm of the difference feature vector; calculating the product between the product eigenvector and the transposed vector of the temperature eigenvector to obtain a full-field Jing Shanying correlation matrix; calculating the Frobenius norm of the full-scene homographic incidence matrix; processing the logarithmic difference characteristic vector by taking a norm of the difference characteristic vector as a weighting coefficient and taking a Frobenius norm of the full-field Jing Shanying incidence matrix as a bias term to obtain a corrected temperature characteristic vector which is homologically aligned with the product characteristic vector in depth; and calculating the weighted sum of the corrected temperature characteristic vector and the product characteristic vector according to the position to obtain the classification characteristic vector.
In the above control method of an energy management control system for lithium fluoride production, fusing the product eigenvector and the temperature eigenvector to obtain a classification eigenvector, including: processing the logarithmic difference eigenvector by taking a norm of the difference eigenvector as a weighting coefficient and a Frobenius norm of the full-field Jing Shanying incidence matrix as a bias term according to the following formula to obtain the corrected temperature eigenvector; wherein the formula is:
Figure 835803DEST_PATH_IMAGE027
wherein the content of the first and second substances,
Figure 530964DEST_PATH_IMAGE028
represents a norm of the differential feature vector, and
Figure 629501DEST_PATH_IMAGE029
a Frobenius norm representing the full scene homographic incidence matrix,
Figure 274503DEST_PATH_IMAGE030
indicating dot-by-dot,
Figure 79779DEST_PATH_IMAGE031
Indicating that the sums are added by location,
Figure 883524DEST_PATH_IMAGE032
indicating subtraction by position.
In the above control method of the energy management control system for lithium fluoride production, passing the classification feature vector through a classifier to obtain a classification result includes: processing the classification feature vector using the classifier to obtain the classification result with the following formula:
Figure 328412DEST_PATH_IMAGE033
wherein, in the step (A),
Figure 511525DEST_PATH_IMAGE034
to
Figure 905729DEST_PATH_IMAGE035
In order to be a weight matrix, the weight matrix,
Figure 145955DEST_PATH_IMAGE036
to
Figure 78139DEST_PATH_IMAGE037
In order to be a vector of the offset,
Figure 928590DEST_PATH_IMAGE038
is that theAnd classifying the feature vectors.
In the above control method of the energy management control system for lithium fluoride production, passing the classification feature vector through a classifier to obtain a classification result, the method further includes: adjusting the temperature of the calciner based on the classification result.
Compared with the prior art, the energy management control system and the control method for preparing the lithium fluoride provided by the application dynamically extract the temperature correlation characteristics in the calcinator in real time through the convolutional neural network model based on deep learning, deeply excavate the structural change characteristics of the calcinated product and the internal heat distribution characteristics, and then intelligently adjust the temperature of the calcinator by combining the characteristic information of the three on the time sequence, so that the energy is optimized and the quality of the final lithium fluoride product is guaranteed.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
Fig. 1 is a diagram of an application scenario of an energy management control system for lithium fluoride production according to an embodiment of the present application.
Fig. 2 is a block diagram of an energy management control system for lithium fluoride production according to an embodiment of the present application.
Fig. 3 is a block diagram of a feature fusion module in an energy management control system for lithium fluoride production according to an embodiment of the present application.
Fig. 4 is a flowchart of a control method of an energy management control system for lithium fluoride production according to an embodiment of the present application.
Fig. 5 is a schematic block diagram of a control method of an energy management control system for lithium fluoride production according to an embodiment of the present disclosure.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of a scene
As mentioned above, lithium fluoride is an important lithium-based material and is a white non-hygroscopic cubic crystal at normal temperature. As an important inorganic fluoride, high-purity lithium fluoride is widely used in the production of fluorinated glasses and optical fibers. Meanwhile, the high-purity lithium fluoride is also an important raw material of an electrolyte material for a lithium ion battery.
The commonly used preparation methods of lithium fluoride mainly comprise a direct preparation method, an ion exchange preparation method and an extraction preparation method. Regardless of the preparation scheme, energy management during the preparation process is a very important concern, especially in the current large background of energy conservation and environmental protection.
In particular, in the process of preparing lithium fluoride, temperature control such as calcination, cooling and the like is mostly involved, and it should be understood that the temperature control not only relates to energy management, but also relates to the quality of the final lithium fluoride product. In the existing lithium fluoride preparation schemes, the technical scheme of preset temperature is mostly adopted, which is not only unfavorable for energy optimization, but also unfavorable for controlling the quality of the final lithium fluoride finished product. Therefore, an energy management control system for lithium fluoride preparation is desired to intelligently control the temperature of the calcinator so as to ensure the quality of the finished lithium fluoride product while optimizing energy.
At present, deep learning and neural networks have been widely applied in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks also exhibit a level close to or even exceeding that of humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
In recent years, deep learning and the development of neural networks provide solutions for controlling the temperature of the calcinator.
Based on this, the present inventors considered that if the temperature of the calciner is controlled intelligently and accurately to improve the quality of the lithium fluoride finished product, the temperature inside the calciner needs to be measured dynamically in real time, and the structural characteristics and the heat distribution of the calcined product also need to be analyzed. Therefore, in the application, a temperature sensor is adopted to collect heat information in the calcinator, the internal heat distribution and the internal structure information of the calcinated product are collected through a thermal infrared camera and an X-Ray scanner, the implicit correlation characteristic distribution of the calcinated product is further deeply mined through a convolutional neural network model based on deep learning, and therefore the calcinated product temperature regulation at the current time point can be guaranteed to be more suitable for the preparation of lithium fluoride during classification.
Specifically, in the technical solution of the present application, first, calcination temperature values of a calciner at a plurality of predetermined time points are obtained through a temperature sensor disposed in the calciner, wherein the calciner is used for calcining crystals forming lithium fluoride finished products. And acquiring thermal infrared images and X-Ray scans of the calcined product at the plurality of predetermined time points by a thermal infrared camera and an X-Ray scanner which are arranged in the calciner, wherein the thermal infrared camera can acquire the internal heat distribution condition of the calcined product so as to more accurately control the temperature in the calciner, and the X-Ray scanner can acquire the change information of the internal structure and the shape of the calcined product so as to better monitor the formation preparation of the lithium fluoride product.
Then, feature mining is performed on the thermal infrared image of the calcined product by using a convolutional neural network model having an excellent performance in local implicit feature extraction of an image, but considering that when feature mining is performed on the thermal infrared image of the calcined product, attention should be paid more to a heat change feature of the calcined product and an interference heat feature around the heat change feature needs to be removed, in the technical solution of the present application, a first convolutional neural network of a spatial attention mechanism is used to perform feature extraction on the thermal infrared images of the calcined product at a plurality of predetermined time points so as to extract local implicit heat feature distributions of the thermal infrared images of the calcined product at the plurality of predetermined time points, thereby obtaining a thermal infrared feature map.
Meanwhile, considering that when an X-Ray scanner is used for mining the internal space structure characteristics of the calcined product, attention needs to be paid to the dynamic change characteristics of the calcined product to prevent the lithium fluoride finished product from failing to meet the required quality due to too fast reaction, in the technical scheme of the application, a second convolution neural network of a three-dimensional convolution kernel is used for processing the X-Ray scanning maps of the calcined product at the plurality of preset time points to obtain an X-Ray characteristic map with the dynamic change characteristics of the calcined product. And then, further cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram to fuse the heat characteristic distribution in the calcined product and the characteristic information of the internal structure change, and then performing characteristic extraction on the obtained cascading characteristic diagram in a third convolutional neural network serving as a filter to obtain a product characteristic vector.
Regarding the calcination temperature values of the calciner at a plurality of preset time points, considering the correlation among the calcination temperature values at each time point, a context encoder comprising an embedded layer is used for encoding the calcination temperature values of the calciner at the plurality of preset time points so as to extract the global high-dimensional semantic features among the calcination temperature values at the plurality of preset time points to be more suitable for characterizing the essential features of the temperature correlation inside the calciner, thereby obtaining a temperature feature vector.
It should be understood that, considering that the product feature vector is obtained by the first and third convolutional neural networks and the three-dimensional convolutional neural network and the third convolutional neural network in cascade, it may be larger in feature depth than the temperature feature vector obtained by the time-series encoder, and thus, the product feature vector is, for example, described as
Figure 301934DEST_PATH_IMAGE039
With the temperature feature vector, e.g.As is
Figure 447482DEST_PATH_IMAGE040
Before fusion, it is first depth-homography aligned, expressed as:
Figure 601383DEST_PATH_IMAGE041
wherein the content of the first and second substances,
Figure 126298DEST_PATH_IMAGE042
represents a norm of a vector, and
Figure 495094DEST_PATH_IMAGE043
the Frobenius norm of the matrix is represented.
Here, the depth homography alignment performs homography alignment based on scene depth stream between vectors by characteristic depth information characteristics characterized by vector difference, and performs depth perception by full field Jing Shanying incidence matrix between vectors, thereby performing dense depth fitting between vectors on the basis of depth difference of characteristic distribution between vectors to obtain the characteristic vector of the product and the feature vector
Figure 77123DEST_PATH_IMAGE044
The corrected temperature eigenvectors homographically aligned in depth
Figure 124844DEST_PATH_IMAGE045
And further improve the accuracy of subsequent classification.
Further, a position-weighted sum of the corrected temperature feature vector and the product feature vector is calculated to fuse feature information of the two, thereby obtaining the classification feature vector. And then, carrying out classification processing on the classification characteristic vector through a classifier to obtain a classification result which is used for indicating that the temperature of the calcinator at the current time point should be increased or decreased.
Based on this, the present application proposes an energy management control system for lithium fluoride production, comprising: the calcining temperature acquisition module is used for acquiring calcining temperature values of the calciner at a plurality of preset time points through a temperature sensor arranged in the calciner, wherein the calciner is used for calcining to form crystals of lithium fluoride finished products; the product data acquisition module is used for acquiring thermal infrared images and X-Ray scanning charts of the calcined products at the plurality of preset time points through a thermal infrared camera and an X-Ray scanner which are arranged in the calciner; a thermal infrared coding module for passing the thermal infrared images of the calcinated products at the plurality of predetermined time points through a first convolution neural network using a spatial attention mechanism to obtain a thermal infrared characteristic map; a perspective view coding module for passing the X-Ray scans of the calcination products at the plurality of predetermined time points through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray feature map; the correlation coding module is used for cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram and then obtaining a product characteristic vector through a third convolutional neural network serving as a filter; the energy data coding module is used for enabling the calcinations temperature values of the calcinator at a plurality of preset time points to pass through a time sequence coder comprising a one-dimensional convolution layer and a full connection layer so as to obtain a temperature characteristic vector; the characteristic fusion module is used for fusing the product characteristic vector and the temperature characteristic vector to obtain a classification characteristic vector; and the energy management result generation module is used for enabling the classification characteristic vectors to pass through the classifier to obtain a classification result, and the classification result is used for indicating that the temperature of the calcinator at the current time point should be increased or decreased.
Fig. 1 illustrates an application scenario of an energy management control system for lithium fluoride production according to an embodiment of the present application. As shown in fig. 1, in this application scenario, first, calcination temperature values of a calciner at a plurality of predetermined time points are obtained through a temperature sensor (e.g., T as illustrated in fig. 1) disposed within the calciner (e.g., R as illustrated in fig. 1) for calcining crystals forming a lithium fluoride finished product, and a thermal infrared image and an X-Ray scan of the calcined product (e.g., P as illustrated in fig. 1) at the plurality of predetermined time points are obtained through a thermal infrared camera (e.g., C as illustrated in fig. 1) and an X-Ray scanner (E as illustrated in fig. 1) disposed within the calciner. Then, the obtained calcination temperature values of the calciner at a plurality of predetermined time points and the obtained thermal infrared image and X-Ray scan of the calcination product are input into a server (for example, a server S as illustrated in fig. 1) deployed with an energy management control algorithm for lithium fluoride preparation, wherein the server can process the calcination temperature values of the calciner at a plurality of predetermined time points and the thermal infrared image and X-Ray scan of the calcination product with the energy management control algorithm for lithium fluoride preparation to generate a classification result indicating that the temperature of the calciner at the current time point should be increased or decreased. Further, the temperature of the calciner is adjusted based on the classification result.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary System
Fig. 2 illustrates a block diagram of an energy management control system for lithium fluoride production according to an embodiment of the application. As shown in fig. 2, an energy management control system 200 for lithium fluoride preparation according to an embodiment of the present application includes: the calcining temperature acquisition module 210 is configured to acquire calcining temperature values of a calciner at a plurality of predetermined time points through a temperature sensor disposed in the calciner, where the calciner is configured to calcine crystals forming a lithium fluoride finished product; a product data acquisition module 220, configured to acquire thermal infrared images and X-Ray scanning charts of the calcinated products at the multiple predetermined time points through a thermal infrared camera and an X-Ray scanner deployed in the calcinator; a thermal infrared encoding module 230, configured to pass the thermal infrared images of the calcinated product at the plurality of predetermined time points through a first convolution neural network using a spatial attention mechanism to obtain a thermal infrared feature map; a perspective view encoding module 240 for passing the X-Ray scans of the calcine at the plurality of predetermined time points through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray profile; the correlation coding module 250 is used for cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram and then obtaining a product characteristic vector through a third convolutional neural network serving as a filter; the energy data encoding module 260 is used for enabling the calcining temperature values of the calciner at a plurality of preset time points to pass through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer so as to obtain a temperature characteristic vector; a feature fusion module 270, configured to fuse the product feature vector and the temperature feature vector to obtain a classification feature vector; and an energy management result generation module 280 for passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the calciner at the current time point should be increased or decreased.
Specifically, in the embodiment of the present application, the calcination temperature acquisition module 210 and the product data acquisition module 220 are configured to acquire calcination temperature values of a calciner at a plurality of predetermined time points through a temperature sensor disposed in the calciner, wherein the calciner is configured to calcine crystals forming a lithium fluoride finished product, and acquire thermal infrared images and X-Ray scans of the calcined product at the plurality of predetermined time points through a thermal infrared camera and an X-Ray scanner disposed in the calciner. As described above, it can be understood that, considering that the temperature of the calcinator needs to be intelligently and accurately controlled to improve the quality of the lithium fluoride finished product, the temperature inside the calcinator needs to be dynamically measured in real time, and the structural characteristics and heat distribution of the calcinated product also need to be analyzed. Therefore, in the technical scheme of the application, the temperature sensor is adopted to collect heat information in the calcinator, the thermal infrared camera and the X-Ray scanner are used to collect internal heat distribution and internal structure information of the calcinated product, and further deep mining is performed on implicit associated feature distribution of the calcinated product through a convolutional neural network model based on deep learning, so that the calcinated product at the current time point can be guaranteed to be more suitable for the preparation of lithium fluoride during classification.
That is, specifically, in the technical solution of the present application, first, calcination temperature values of a calciner at a plurality of predetermined time points are obtained by a temperature sensor disposed in the calciner, wherein the calciner is used for calcining crystals forming lithium fluoride finished products. And acquiring thermal infrared images and X-Ray scanning diagrams of the calcined product at a plurality of preset time points through a thermal infrared camera and an X-Ray scanner which are arranged in the calciner, wherein the thermal infrared camera can acquire the internal heat distribution condition of the calcined product so as to more accurately control the temperature inside the calciner, and the X-Ray scanner can acquire the change information of the internal structure and the shape of the calcined product so as to better monitor the formation preparation of the lithium fluoride product.
Specifically, in this embodiment, the thermal infrared encoding module 230 is configured to pass the thermal infrared images of the calcinated product at the multiple predetermined time points through a first convolution neural network using a spatial attention mechanism to obtain a thermal infrared feature map. That is, in the technical solution of the present application, a convolutional neural network model having an excellent performance in local implicit feature extraction of an image is used to perform feature mining on a thermal infrared image of the calcined product, but considering that in the feature mining on the thermal infrared image of the calcined product, attention should be paid more to a heat variation feature of the calcined product, and an interference heat feature around the heat variation feature needs to be removed. Therefore, in the technical solution of the present application, a first convolution neural network of a spatial attention mechanism is used to perform feature extraction on the thermal infrared images of the calcinated products at the multiple predetermined time points, so as to extract a local implicit thermal feature distribution of the thermal infrared images of the calcinated products at the multiple predetermined time points, thereby obtaining a thermal infrared feature map.
More specifically, in this embodiment of the present application, the thermal infrared encoding module is further configured to perform, in layer forward direction transfer, using the layers of the first convolutional neural network, on input data: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram; pooling the convolved feature map to generate a pooled feature map; performing activation processing on the pooled feature map to generate an activated feature map; performing global average pooling along a channel dimension on the activation feature map to obtain a spatial feature matrix; performing convolution processing and activation processing on the spatial feature matrix to generate a weight vector; weighting each feature matrix of the activation feature map by using the weight value of each position in the weight vector to obtain a generated feature map; wherein the generated feature map output by the last layer of the first convolutional neural network model is the thermal infrared feature map.
Specifically, in the embodiment of the present application, the perspective view encoding module 240 and the association encoding module 250 are configured to pass the X-Ray scans of the calcination products at the plurality of predetermined time points through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray feature map, and cascade the thermal infrared feature map and the X-Ray feature map and then pass through a third convolutional neural network as a filter to obtain a product feature vector. It should be understood that the dynamic change characteristics of the calcined product need to be considered when the X-Ray scanner is used to mine the internal spatial structural characteristics of the calcined product, so as to prevent the lithium fluoride product from reacting too fast to meet the required quality. Therefore, in the technical scheme of the application, the X-Ray scanning images of the calcined products at the plurality of preset time points are processed by using a second convolutional neural network of a three-dimensional convolution kernel so as to obtain an X-Ray characteristic diagram with the dynamic change characteristic of the calcined products. And then, further cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram to fuse the heat characteristic distribution in the calcined product and the characteristic information of the internal structure change to obtain a cascading characteristic diagram, and then performing characteristic extraction on the obtained cascading characteristic diagram through a third convolutional neural network serving as a filter to obtain a product characteristic vector. Accordingly, in one particular example, the layers of the third convolutional neural network as filters are used to perform convolution processing, pooling processing along a feature matrix, and activation processing on input data in forward pass of layers to generate the product feature vector from a last layer of the third convolutional neural network, wherein an input of the first layer of the third convolutional neural network is the concatenated feature map.
More specifically, in this embodiment of the present application, the perspective view coding module is further configured to perform, in the forward direction passing through the layers, the following respectively on the input data by the second convolutional neural network using the three-dimensional convolution kernel: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram; pooling the convolution characteristic map to obtain a pooled characteristic map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the second convolutional neural network is the X-Ray characteristic diagram, and the input of the first layer of the second convolutional neural network is the X-Ray scanning diagram of the calcination products at the plurality of preset time points.
Specifically, in the embodiment of the present application, the energy data encoding module 260 is configured to pass the calcination temperature values of the calciner at a plurality of predetermined time points through a time-sequence encoder including a one-dimensional convolution layer and a full link layer to obtain a temperature feature vector. It should be understood that, for the calcination temperature values of the calciner at a plurality of predetermined time points, considering the correlation between the calcination temperature values at the respective time points, in the technical solution of the present application, a context encoder comprising an embedded layer is used to encode the calcination temperature values of the calciner at the plurality of predetermined time points to extract the global-based high-dimensional semantic features between the calcination temperature values at the plurality of predetermined time points so as to be more suitable for characterizing the essential features of the temperature correlation inside the calciner, thereby obtaining a temperature feature vector.
More specifically, in this embodiment of the present application, the energy data encoding module includes: the input vector construction unit is used for arranging the calcination temperature values of the calciner at a plurality of preset time points into a one-dimensional input vector according to the time dimension; a full-concatenation coding unit, configured to perform full-concatenation coding on the input vector using a full-concatenation layer of the time sequence encoder according to the following formula to extract a high-dimensional implicit feature of a feature value at each position in the input vector, where the formula is:
Figure 46926DEST_PATH_IMAGE046
in which
Figure 129283DEST_PATH_IMAGE047
Is the input vector of the said one or more input vectors,
Figure 85475DEST_PATH_IMAGE048
is the output vector of the digital video signal,
Figure 620493DEST_PATH_IMAGE049
is a matrix of weights that is a function of,
Figure 487211DEST_PATH_IMAGE050
is a vector of the offset to be used,
Figure 220812DEST_PATH_IMAGE051
represents a matrix multiplication; a one-dimensional convolution coding unit, configured to perform one-dimensional convolution coding on the input vector by using a one-dimensional convolution layer of the time-series encoder according to the following formula to extract a high-dimensional implicit correlation feature between feature values of each position in the input vector, where the formula is:
Figure 754430DEST_PATH_IMAGE026
wherein the content of the first and second substances,ais a convolution kernel inxA width in the direction,FIs a convolution kernel parameter vector,GIs a matrix of local vectors operating with a convolution kernel,wthe size of the convolution kernel.
Specifically, in this embodiment, the feature fusion module 270 is configured to fuse the product feature vector and the temperature feature vector to obtain a classification feature vector. It should be understood that, in the technical solution of the present application, the product feature vector is described as the temperature feature vector obtained by the time-series encoder, for example, considering that the product feature vector is obtained by the first convolutional neural network and the third convolutional neural network and the three-dimensional convolutional neural network and the third convolutional neural network which are cascaded, and is greater in feature depth than the temperature feature vector obtained by the time-series encoder
Figure 901378DEST_PATH_IMAGE052
With said temperature characteristic vector, e.g. as
Figure 317927DEST_PATH_IMAGE053
Before fusion, the temperature feature vectors need to be depth homography aligned first.
That is, specifically, in the present embodiment, first, a differential feature vector between the product feature vector and the temperature feature vector is calculated. Then, a logarithmic function value of the feature value of each position in the difference feature vector is calculated to obtain a logarithmic difference feature vector. Then, a norm of the difference feature vector is calculated. Then, the product between the product eigenvector and the transposed vector of the temperature eigenvector is calculated to obtain a full-field Jing Shanying correlation matrix. Then, the Frobenius norm of the full scene homography incidence matrix is calculated. Then, a norm of the difference feature vector is used as a weighting coefficient, and a Frobenius norm of the full-field Jing Shanying incidence matrix is used as a bias term to process the logarithm difference feature vector so as to obtain a corrected temperature feature vector which is homologically aligned with the product feature vector in depth. Accordingly, in a specific example, the logarithmic difference eigenvector is processed by using a norm of the difference eigenvector as a weighting coefficient and a Frobenius norm of the full-field Jing Shanying incidence matrix as a bias term according to the following formula to obtain the corrected temperature eigenvector; wherein the formula is:
Figure 781400DEST_PATH_IMAGE027
wherein the content of the first and second substances,
Figure 876133DEST_PATH_IMAGE054
represents a norm of the differential feature vector, and
Figure 293732DEST_PATH_IMAGE055
represents the wholeThe Frobenius norm of the scene homography incidence matrix,
Figure 374690DEST_PATH_IMAGE056
indicating dot-by-dot,
Figure 551724DEST_PATH_IMAGE057
Indicating that the sums are added by location,
Figure 554709DEST_PATH_IMAGE058
indicating subtraction by position. And finally, calculating the weighted sum of the corrected temperature characteristic vector and the product characteristic vector according to the position to obtain the classification characteristic vector. It should be understood that the depth homography alignment performs homography alignment based on scene depth stream between vectors according to the characteristic of the feature depth information characterized by vector difference, and performs depth perception through a full-field Jing Shanying incidence matrix between vectors, so that dense depth fitting between vectors is performed on the basis of the depth difference of feature distribution between vectors to obtain the feature vector and the product
Figure 82773DEST_PATH_IMAGE059
The corrected temperature eigenvectors homologically aligned in depth
Figure 826476DEST_PATH_IMAGE060
And further improve the accuracy of subsequent classification.
Fig. 3 illustrates a block diagram of a feature fusion module in an energy management control system for lithium fluoride production according to an embodiment of the application. As shown in fig. 3, the feature fusion module 270 includes: a vector difference unit 271 for calculating a difference feature vector between the product feature vector and the temperature feature vector; a logarithm operation unit 272, configured to calculate a logarithm function value of a feature value of each position in the difference feature vector to obtain a logarithm difference feature vector; a depth information characteristic value calculating unit 273 for calculating a norm of the differential feature vector; a full-scene homography incidence matrix constructing unit 274, configured to calculate a product between the product feature vector and a transposed vector of the temperature feature vector to obtain a full-field Jing Shanying incidence matrix; a depth perception unit 275, configured to calculate a Frobenius norm of the full scene homography incidence matrix; an alignment unit 276, configured to process the log-difference eigenvector with a norm of the difference eigenvector as a weighting coefficient and a Frobenius norm of the full-field Jing Shanying correlation matrix as a bias term to obtain a corrected temperature eigenvector that is homologically aligned with the product eigenvector in depth; a fusion unit 277, configured to calculate a weighted sum, according to location, of the corrected temperature feature vector and the product feature vector to obtain the classification feature vector.
Specifically, in the embodiment of the present application, the energy management result generating module 280 is configured to pass the classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate that the temperature of the calciner at the current time point should be increased or decreased. That is, in the technical solution of the present application, after fusing the corrected temperature feature vector and the product feature vector, the classification feature vector is further subjected to classification processing in a classifier, so as to obtain a classification result indicating that the temperature of the calciner at the current time point should be increased or decreased.
More specifically, in this embodiment, the energy management result generating module is further configured to: processing the classification feature vector using the classifier to obtain the classification result with a formula:
Figure 858017DEST_PATH_IMAGE061
wherein, in the step (A),
Figure 31903DEST_PATH_IMAGE062
to
Figure 922629DEST_PATH_IMAGE063
In the form of a matrix of weights,
Figure 204444DEST_PATH_IMAGE064
to
Figure 683967DEST_PATH_IMAGE065
In order to be a vector of the offset,
Figure 423560DEST_PATH_IMAGE066
the classified feature vector is obtained.
In summary, the energy management control system 200 for lithium fluoride preparation according to the embodiment of the present application is illustrated, which dynamically extracts the temperature-related characteristics inside the calciner in real time through the convolutional neural network model based on deep learning, and deeply mines the structural change characteristics and the internal heat distribution characteristics of the calcined product, and then intelligently adjusts the temperature of the calciner by combining the time-series characteristic information of the three characteristics, so as to ensure the quality of the final lithium fluoride product while optimizing the energy.
As described above, the energy management control system 200 for lithium fluoride production according to the embodiment of the present application may be implemented in various terminal devices, such as a server of an energy management control algorithm for lithium fluoride production, and the like. In one example, the energy management control system 200 for lithium fluoride preparation according to the embodiment of the present application may be integrated into a terminal device as one software module and/or hardware module. For example, the energy management control system 200 for lithium fluoride production may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the energy management control system 200 for lithium fluoride preparation can also be one of the hardware modules of the terminal device.
Alternatively, in another example, the energy management control system 200 for lithium fluoride preparation and the terminal device may be separate devices, and the energy management control system 200 for lithium fluoride preparation may be connected to the terminal device through a wired and/or wireless network and transmit the interactive information according to an agreed data format.
Exemplary method
Fig. 4 illustrates a flow chart of a control method of an energy management control system for lithium fluoride production. As shown in fig. 4, a control method of an energy management control system for lithium fluoride production according to an embodiment of the present application includes the steps of: s110, acquiring calcination temperature values of a calciner at a plurality of preset time points through a temperature sensor arranged in the calciner, wherein the calciner is used for calcining to form crystals of lithium fluoride finished products; s120, acquiring thermal infrared images and X-Ray scanning images of the calcinated products at a plurality of preset time points through a thermal infrared camera and an X-Ray scanner which are arranged in the calcinator; s130, passing the thermal infrared images of the calcinated products at the plurality of preset time points through a first convolution neural network using a spatial attention mechanism to obtain a thermal infrared characteristic diagram; s140, passing the X-Ray scanning images of the calcination products at the plurality of preset time points through a second convolution neural network using a three-dimensional convolution kernel to obtain an X-Ray characteristic diagram; s150, cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram, and then passing through a third convolution neural network serving as a filter to obtain a product characteristic vector; s160, passing the calcination temperature values of the calciner at a plurality of preset time points through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer to obtain a temperature characteristic vector; s170, fusing the product feature vector and the temperature feature vector to obtain a classification feature vector; and S180, passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the calcinator at the current time point should be increased or decreased.
Fig. 5 illustrates an architecture diagram of a control method of an energy management control system for lithium fluoride production according to an embodiment of the present application. As shown in fig. 5, in the network architecture of the control method of the energy management control system for lithium fluoride production, first, the obtained thermal infrared images (e.g., P1 as illustrated in fig. 5) of the calcination products at the plurality of predetermined time points are passed through a first convolutional neural network (e.g., CNN1 as illustrated in fig. 5) using a spatial attention mechanism to obtain a thermal infrared profile (e.g., F1 as illustrated in fig. 5); next, passing the obtained X-Ray scans (e.g., P2 as illustrated in fig. 5) of the calcination products at the plurality of predetermined time points through a second convolutional neural network (e.g., CNN2 as illustrated in fig. 5) using a three-dimensional convolutional kernel to obtain an X-Ray profile (e.g., F2 as illustrated in fig. 5); then, cascading the thermal infrared feature map and the X-Ray feature map and then passing through a third convolutional neural network (e.g., CNN3 as illustrated in fig. 5) as a filter to obtain a product feature vector (e.g., VF1 as illustrated in fig. 5); then, passing the obtained calcination temperature values (e.g., Q as illustrated in fig. 5) of the calciner at a plurality of predetermined time points through a time-sequence encoder (e.g., E as illustrated in fig. 5) comprising a one-dimensional convolutional layer and a fully-connected layer to obtain a temperature characteristic vector (e.g., VF2 as illustrated in fig. 5); then, fusing the product feature vector and the temperature feature vector to obtain a classification feature vector (e.g., VF as illustrated in fig. 5); and, finally, passing the classification feature vector through a classifier (e.g., circle S as illustrated in fig. 5) to obtain a classification result, which is used to indicate that the temperature of the calciner at the current time point should be increased or decreased.
More specifically, in steps S110 and S120, calcination temperature values of a calciner at a plurality of predetermined time points are obtained by a temperature sensor disposed in the calciner for calcining crystals forming a lithium fluoride finished product, and thermal infrared images and X-Ray scans of the calcination products at the plurality of predetermined time points are obtained by a thermal infrared camera and an X-Ray scanner disposed in the calciner. It should be understood that, considering that the temperature of the calciner is required to be intelligently and accurately controlled to improve the quality of the lithium fluoride finished product, the temperature inside the calciner needs to be dynamically measured in real time, and the structural characteristics and the heat distribution of the calcined product also need to be analyzed. Therefore, in the technical scheme of the application, the temperature sensor is adopted to collect heat information in the calcinator, the thermal infrared camera and the X-Ray scanner are used to collect internal heat distribution and internal structure information of the calcinated product, and further deep mining is performed on implicit associated feature distribution of the calcinated product through a convolutional neural network model based on deep learning, so that the calcinated product at the current time point can be guaranteed to be more suitable for the preparation of lithium fluoride during classification.
That is, specifically, in the technical solution of the present application, first, calcination temperature values of a calciner at a plurality of predetermined time points are obtained by a temperature sensor disposed in the calciner, wherein the calciner is used for calcining crystals forming lithium fluoride finished products. And acquiring thermal infrared images and X-Ray scanning diagrams of the calcined product at a plurality of preset time points through a thermal infrared camera and an X-Ray scanner which are arranged in the calciner, wherein the thermal infrared camera can acquire the internal heat distribution condition of the calcined product so as to more accurately control the temperature inside the calciner, and the X-Ray scanner can acquire the change information of the internal structure and the shape of the calcined product so as to better monitor the formation preparation of the lithium fluoride product.
More specifically, in step S130, the thermal infrared images of the calcined product at the plurality of predetermined time points are passed through a first convolutional neural network using a spatial attention mechanism to obtain a thermal infrared profile. That is, in the technical solution of the present application, a convolutional neural network model having an excellent performance in local implicit feature extraction of an image is used to perform feature mining on a thermal infrared image of the calcined product, but considering that in the feature mining on the thermal infrared image of the calcined product, attention should be paid more to a heat variation feature of the calcined product, and an interference heat feature around the heat variation feature needs to be removed. Therefore, in the technical solution of the present application, a first convolution neural network with a spatial attention mechanism is used to perform feature extraction on the thermal infrared images of the calcinated products at the multiple predetermined time points, so as to extract a local latent heat feature distribution of the thermal infrared images of the calcinated products at the multiple predetermined time points, thereby obtaining a thermal infrared feature map.
More specifically, in steps S140 and S150, the X-Ray scans of the calcined product at the plurality of predetermined time points are passed through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray feature map, and the thermal infrared feature map and the X-Ray feature map are cascaded and then passed through a third convolutional neural network as a filter to obtain a product feature vector. It should be understood that the dynamic change characteristics of the calcined product need to be considered when the X-Ray scanner is used to mine the internal spatial structural characteristics of the calcined product, so as to prevent the lithium fluoride product from reacting too fast to meet the required quality. Therefore, in the technical scheme of the application, the X-Ray scanning images of the calcined products at the plurality of preset time points are processed by using a second convolutional neural network of a three-dimensional convolution kernel so as to obtain an X-Ray characteristic diagram with the dynamic change characteristic of the calcined products. And then, further cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram to fuse the heat characteristic distribution in the calcined product and the characteristic information of the internal structure change to obtain a cascading characteristic diagram, and then performing characteristic extraction on the obtained cascading characteristic diagram through a third convolutional neural network serving as a filter to obtain a product characteristic vector. Accordingly, in one particular example, the input data is convolved, pooled along a feature matrix, and activated in forward passes of layers using layers of the third convolutional neural network as filters to generate the product feature vector from a last layer of the third convolutional neural network, wherein an input of the first layer of the third convolutional neural network is the concatenated feature map.
More specifically, in step S160, the calcinations temperature values of the calcinator at a plurality of predetermined time points are passed through a time sequence encoder including a one-dimensional convolution layer and a full connection layer to obtain a temperature characteristic vector. It should be understood that, for the calcination temperature values of the calciner at a plurality of predetermined time points, considering the correlation between the calcination temperature values at the respective time points, in the technical solution of the present application, a context encoder comprising an embedded layer is used to encode the calcination temperature values of the calciner at the plurality of predetermined time points so as to extract a global-based high-dimensional semantic feature between the calcination temperature values at the plurality of predetermined time points, so as to be more suitable for characterizing the essential feature of the temperature correlation inside the calciner, thereby obtaining a temperature feature vector.
More specifically, in step S170, the product feature vector and the temperature feature vector are fused to obtain a classification feature vector. It should be understood that, in the technical solution of the present application, the product feature vector is, for example, denoted as the temperature feature vector obtained by the time sequence encoder, considering that the product feature vector is obtained by the first convolutional neural network and the third convolutional neural network and the three-dimensional convolutional neural network and the third convolutional neural network which are cascaded, and is greater in feature depth than the temperature feature vector obtained by the time sequence encoder
Figure 660637DEST_PATH_IMAGE067
With said temperature characteristic vector, e.g. as
Figure 746143DEST_PATH_IMAGE068
Before fusion, the temperature feature vectors need to be depth homography aligned first.
More specifically, in step S180, the classification feature vector is passed through a classifier to obtain a classification result, which is used to indicate that the temperature of the calciner at the current time point should be increased or decreased. That is, in the technical solution of the present application, after fusing the corrected temperature feature vector and the product feature vector, the classification feature vector is further subjected to classification processing in a classifier to obtain a classification result indicating that the temperature of the calciner at the current time point should be increased or decreased.
In summary, the control method of the energy management control system for lithium fluoride preparation based on the embodiments of the present application is illustrated, which dynamically extracts the temperature-related characteristics inside the calciner in real time through the convolutional neural network model based on deep learning, and deeply excavates the structural change characteristics and the internal heat distribution characteristics of the calcined product, and then intelligently adjusts the temperature of the calciner by combining the time-series characteristic information of the three, so as to ensure the final product quality of lithium fluoride while optimizing the energy.
The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is provided for purposes of illustration and understanding only, and is not intended to limit the application to the details which are set forth in order to provide a thorough understanding of the present application.
The block diagrams of devices, apparatuses, devices, systems referred to in this application are only used as illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, each component or step can be decomposed and/or re-combined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.

Claims (10)

1. An energy management control system for lithium fluoride production, comprising:
the calcining temperature acquisition module is used for acquiring calcining temperature values of the calciner at a plurality of preset time points through a temperature sensor arranged in the calciner, wherein the calciner is used for calcining to form crystals of lithium fluoride finished products;
the product data acquisition module is used for acquiring thermal infrared images and X-Ray scanning charts of the calcined products at the plurality of preset time points through a thermal infrared camera and an X-Ray scanner which are arranged in the calciner;
a thermal infrared coding module for passing the thermal infrared images of the calcinated products at the plurality of predetermined time points through a first convolution neural network using a spatial attention mechanism to obtain a thermal infrared characteristic map;
a perspective view coding module for passing the X-Ray scans of the calcination products at the plurality of predetermined time points through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray feature map;
the correlation coding module is used for cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram and then obtaining a product characteristic vector through a third convolutional neural network serving as a filter;
the energy data coding module is used for enabling the calcinations temperature values of the calcinator at a plurality of preset time points to pass through a time sequence coder comprising a one-dimensional convolution layer and a full connection layer so as to obtain a temperature characteristic vector;
the characteristic fusion module is used for fusing the product characteristic vector and the temperature characteristic vector to obtain a classification characteristic vector; and
and the energy management result generation module is used for enabling the classification characteristic vector to pass through a classifier to obtain a classification result, and the classification result is used for indicating that the temperature of the calcinator at the current time point should be increased or decreased.
2. The energy management control system for lithium fluoride production of claim 1, wherein the thermal infrared encoding module is further configured to perform, in forward pass of layers, using the layers of the first convolutional neural network, input data for:
performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution characteristic diagram;
pooling the convolved feature map to generate a pooled feature map;
performing activation processing on the pooled feature map to generate an activated feature map;
performing global average pooling along a channel dimension on the activation feature map to obtain a spatial feature matrix;
performing convolution processing and activation processing on the spatial feature matrix to generate a weight vector; and
weighting each feature matrix of the activation feature map by the weight value of each position in the weight vector to obtain a generated feature map;
wherein the generated feature map output by the last layer of the first convolutional neural network model is the thermal infrared feature map.
3. The energy management control system for lithium fluoride production of claim 2, wherein the perspective view coding module is further configured to perform the second convolutional neural network using three-dimensional convolutional kernels on the input data in forward direction of the layer by:
performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram;
pooling the convolution characteristic map to obtain a pooled characteristic map; and
performing nonlinear activation on the pooled feature map to obtain an activated feature map;
wherein the output of the last layer of the second convolutional neural network is the X-Ray characteristic diagram, and the input of the first layer of the second convolutional neural network is the X-Ray scanning diagram of the calcinated product at the plurality of preset time points.
4. The energy management control system for lithium fluoride production of claim 3, wherein the correlation encoding module,
the cascade unit is used for cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram to obtain a cascade characteristic diagram; and
a feature extraction unit configured to perform convolution processing, pooling processing along a feature matrix, and activation processing on input data in forward pass of layers using layers of the third convolutional neural network as a filter to generate the product feature vector from a last layer of the third convolutional neural network, wherein an input of the first layer of the third convolutional neural network is the concatenated feature map.
5. The energy management control system for lithium fluoride production of claim 4, wherein the energy data encoding module comprises:
the input vector construction unit is used for arranging the calcination temperature values of the calciner at a plurality of preset time points into a one-dimensional input vector according to the time dimension;
a full-concatenation coding unit, configured to perform full-concatenation coding on the input vector using a full-concatenation layer of the time sequence encoder according to the following formula to extract a high-dimensional implicit feature of a feature value at each position in the input vector, where the formula is:
Figure FDA0003949042100000021
where X is the input vector, Y is the output vector, W is the weight matrix, B is the offset vector,
Figure FDA0003949042100000022
represents a matrix multiplication;
a one-dimensional convolution coding unit, configured to perform one-dimensional convolution coding on the input vector by using a one-dimensional convolution layer of the time-series encoder according to the following formula to extract a high-dimensional implicit correlation feature between feature values of each position in the input vector, where the formula is:
Figure FDA0003949042100000031
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
6. The energy management control system for lithium fluoride production of claim 5, wherein the feature fusion module comprises:
a vector difference unit for calculating a difference feature vector between the product feature vector and the temperature feature vector;
the logarithm operation unit is used for calculating the logarithm function value of the characteristic value of each position in the difference characteristic vector to obtain a logarithm difference characteristic vector;
the depth information characteristic value calculation unit is used for calculating a norm of the differential feature vector;
the full-scene homography incidence matrix constructing unit is used for calculating the product between the product characteristic vector and the transposed vector of the temperature characteristic vector to obtain a full-field Jing Shanying incidence matrix;
the depth perception unit is used for calculating the Frobenius norm of the full-scene homography incidence matrix;
an alignment unit, configured to process the logarithmic difference feature vector by using a norm of the difference feature vector as a weighting coefficient and using a Frobenius norm of the full-field Jing Shanying correlation matrix as a bias term to obtain a corrected temperature feature vector that is homography aligned with the product feature vector in depth;
and the fusion unit is used for calculating the weighted sum of the corrected temperature characteristic vector and the product characteristic vector according to the position to obtain the classification characteristic vector.
7. The energy management control system for lithium fluoride production of claim 6, wherein the alignment unit is configured to process the logarithmic difference eigenvector with a norm of the difference eigenvector as a weighting factor and a Frobenius norm of the full-field Jing Shanying correlation matrix as a bias term to obtain the corrected temperature eigenvector;
wherein the formula is:
Figure FDA0003949042100000032
wherein the content of the first and second substances,
Figure FDA0003949042100000033
a norm representing the difference feature vector, and V 1 T V 2 || F A Frobenius norm indicating the full scene homography incidence matrix, an-indicating being multiplied by a position point,
Figure FDA0003949042100000041
Indicating that the addition is made by position,
Figure FDA0003949042100000042
indicating subtraction by position.
8. The energy management control system for lithium fluoride production of claim 7, wherein the energy management result generation module is further configured to: processing the classification feature vector using the classifier to obtain the classification result with the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) | X }, wherein W 1 To W n As a weight matrix, B 1 To B n Is a bias vector and X is the classification feature vector.
9. The energy management control system for lithium fluoride production of claim 8, wherein the energy management result generation module further comprises an energy control module for adjusting the temperature of the calciner based on the classification result.
10. A control method of an energy management control system for lithium fluoride preparation is characterized by comprising the following steps:
acquiring calcination temperature values of a calciner at a plurality of preset time points through a temperature sensor arranged in the calciner, wherein the calciner is used for calcining crystals forming lithium fluoride finished products;
acquiring thermal infrared images and X-Ray scanning diagrams of the calcined products at the plurality of preset time points through a thermal infrared camera and an X-Ray scanner which are arranged in the calciner;
passing the thermal infrared images of the calcined product at the plurality of predetermined time points through a first convolutional neural network using a spatial attention mechanism to obtain a thermal infrared signature;
passing the X-Ray scans of the calcination product at the plurality of predetermined time points through a second convolutional neural network using a three-dimensional convolutional kernel to obtain an X-Ray profile;
cascading the thermal infrared characteristic diagram and the X-Ray characteristic diagram, and then passing through a third convolutional neural network serving as a filter to obtain a product characteristic vector;
enabling the calcination temperature values of the calciner at a plurality of preset time points to pass through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer so as to obtain a temperature characteristic vector;
fusing the product feature vector and the temperature feature vector to obtain a classification feature vector; and
and passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the calciner at the current time point should be increased or decreased.
CN202210800838.3A 2022-07-08 2022-07-08 Energy management control system for lithium fluoride preparation and control method thereof Active CN115291646B (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202210800838.3A CN115291646B (en) 2022-07-08 2022-07-08 Energy management control system for lithium fluoride preparation and control method thereof
PCT/CN2022/117768 WO2024007445A1 (en) 2022-07-08 2022-09-08 Energy management control system for lithium fluoride preparation and control method therefor

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210800838.3A CN115291646B (en) 2022-07-08 2022-07-08 Energy management control system for lithium fluoride preparation and control method thereof

Publications (2)

Publication Number Publication Date
CN115291646A CN115291646A (en) 2022-11-04
CN115291646B true CN115291646B (en) 2022-12-23

Family

ID=83822429

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210800838.3A Active CN115291646B (en) 2022-07-08 2022-07-08 Energy management control system for lithium fluoride preparation and control method thereof

Country Status (2)

Country Link
CN (1) CN115291646B (en)
WO (1) WO2024007445A1 (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116000297A (en) * 2023-01-03 2023-04-25 赣州市光华有色金属有限公司 Preparation device and method for high-strength tungsten lanthanum wire
CN116003126B (en) * 2023-03-22 2023-06-20 珠海和泽科技有限公司 Preparation method and system of electrostatic chuck surface ceramic material
CN116048031B (en) * 2023-03-31 2023-08-04 克拉玛依市蓝润环保科技有限责任公司 Control system and method for petroleum auxiliary production
CN116130721B (en) * 2023-04-11 2023-06-16 杭州鄂达精密机电科技有限公司 Status diagnostic system and method for hydrogen fuel cell
CN116907214B (en) * 2023-05-09 2024-03-08 广东夏和瓷业有限公司 Preparation process and system of environment-friendly domestic ceramic
CN116652681A (en) * 2023-05-24 2023-08-29 浙江沃沃阀门有限公司 Automatic multi-station machining equipment and method for forged steel valve cover

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105338262B (en) * 2015-10-09 2018-09-21 浙江大华技术股份有限公司 A kind of graphic images processing method and processing device
CN112313067B (en) * 2018-04-23 2023-02-03 物化股份有限公司 Thermal control in laser sintering
CN110544247B (en) * 2019-09-03 2023-11-07 东莞德福得精密五金制品有限公司 Method for checking and monitoring operation defects of artificial intelligent cloud computing multi-path equipment system
CN111782459A (en) * 2020-06-03 2020-10-16 东南大学 Traveling wave tube internal temperature prediction method based on tube external temperature image and CNN
CN114001360B (en) * 2021-12-31 2022-04-26 南京沃谱瑞环境研究院有限公司 Burning rotary kiln intelligent control method and system based on Internet of things

Also Published As

Publication number Publication date
CN115291646A (en) 2022-11-04
WO2024007445A1 (en) 2024-01-11

Similar Documents

Publication Publication Date Title
CN115291646B (en) Energy management control system for lithium fluoride preparation and control method thereof
Eldesokey et al. Propagating confidences through cnns for sparse data regression
CN115079572B (en) Energy management control system for preparing lithium hexafluorophosphate and control method thereof
CN109934282B (en) SAGAN sample expansion and auxiliary information-based SAR target classification method
CN109389043B (en) Crowd density estimation method for aerial picture of unmanned aerial vehicle
US20230316651A1 (en) Three-dimensional reconstruction and angle of view synthesis method for moving human body
Fu et al. Hyperspectral image super-resolution with a mosaic RGB image
WO2019022798A1 (en) Image capture devices featuring intelligent use of lightweight hardware-generated statistics
US9961330B2 (en) Device and method of generating multi-view immersive content
CN110490907A (en) Motion target tracking method based on multiple target feature and improvement correlation filter
CN115294639B (en) Color temperature adjustable lamp strip and control method thereof
CN116003126B (en) Preparation method and system of electrostatic chuck surface ceramic material
CN117058622A (en) Intelligent monitoring system and method for sewage treatment equipment
CN111274988B (en) Multispectral-based vehicle weight identification method and device
CN116091854B (en) Method and system for classifying space targets of HRRP sequence
CN116343329A (en) Infrared-visible light multi-source image fusion integrated management and control system and equipment
CN114170526A (en) Remote sensing image multi-scale target detection and identification method based on lightweight network
CN111898565B (en) Forest smoke and fire real-time monitoring system and method based on robust multi-view
Peng et al. Pixelwise adaptive discretization with uncertainty sampling for depth completion
WO2023226236A1 (en) Energy management control system for electronic grade hydrofluoric acid preparation and control method therefor
CN110633732A (en) Multi-modal image recognition method based on low-rank and joint sparsity
CN115810007A (en) Energy-saving production method and system for aluminum composite plate
CN113343770B (en) Face anti-counterfeiting method based on feature screening
CN113628111A (en) Hyperspectral image super-resolution method based on gradient information constraint
CN112489128A (en) RGB-D indoor unmanned aerial vehicle positioning implementation method based on unsupervised deep learning

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
CB02 Change of applicant information
CB02 Change of applicant information

Address after: 364200 No. 9, Gongye Road, Pingpu village, Jiaoyang Town, Shanghang County, Longyan City, Fujian Province

Applicant after: Fujian Longfu New Material Co.,Ltd.

Address before: 364200 No. 9, Gongye Road, Pingpu village, Jiaoyang Town, Shanghang County, Longyan City, Fujian Province

Applicant before: FUJIAN LONGFU CHEMICAL Co.,Ltd.

SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant