CN116185099A

CN116185099A - Automatic temperature control system for electronic grade hydrogen peroxide preparation

Info

Publication number: CN116185099A
Application number: CN202310450084.8A
Authority: CN
Inventors: 华鹏; 范建平; 赖志林
Original assignee: Fujian Tianfu Electronic Materials Co ltd
Current assignee: Fujian Tianfu Electronic Materials Co ltd
Priority date: 2023-04-25
Filing date: 2023-04-25
Publication date: 2023-05-30

Abstract

The utility model relates to the field of electronic industry, it specifically discloses a temperature automatic control system for electronic grade hydrogen peroxide preparation, and it digs out the time sequence change characteristic information of the temperature value of waiting to purify hydrogen peroxide solution through adopting neural network model based on deep learning, further controls the circulating water velocity value in the cooling tower in a self-adaptation way based on the time sequence change characteristic of the temperature value of hydrogen peroxide solution, like this, can avoid hydrogen peroxide to decompose at purification in-process to improve the preparation quality of electronic grade hydrogen peroxide.

Description

Automatic temperature control system for electronic grade hydrogen peroxide preparation

Technical Field

The present application relates to the field of electronics industry, and more particularly, to an automatic temperature control system for electronic grade hydrogen peroxide preparation.

Background

In the 90 s of the 20 th century, along with the rapid development of the electronic industry, particularly with the high integration of integrated circuits, the demand for electronic grade hydrogen peroxide has been rapidly increasing. Electronic grade hydrogen peroxide is widely used in the field of electronic industry and is one of electronic chemicals necessary for integrated circuit production. Its purity has a very important impact on the yield, electrical performance and reliability of integrated circuits.

The hydrogen peroxide solution obtained by the traditional preparation method contains a plurality of inorganic ions or compound impurities, and cannot meet the use standard of the electronic industry, so that the hydrogen peroxide solution is required to be purified to remove the impurities, and the required high purity is achieved. At present, the hydrogen peroxide is purified mainly by a resin adsorption method, and circulating water in a cooling tower is required to be used for cooling and purifying in the resin adsorption method.

However, in the existing cooling purification scheme, the flow rate of the circulating water in the cooling tower is constant, that is, the heat which can be taken away in unit time is relatively fixed, but the instability of hydrogen peroxide increases geometrically with the increase of temperature, so that the stability of hydrogen peroxide is difficult to ensure, and the decomposition of hydrogen peroxide is easy to be promoted.

Therefore, an automatic temperature control system for electronic grade hydrogen peroxide preparation is desired.

Disclosure of Invention

The present application has been made in order to solve the above technical problems. The embodiment of the application provides a temperature automatic control system for electronic grade hydrogen peroxide preparation, which digs out time sequence change characteristic information of a temperature value of hydrogen peroxide solution to be purified by adopting a neural network model based on deep learning, and further adaptively controls a circulating water flow velocity value in a cooling tower based on the time sequence change characteristic of the temperature value of the hydrogen peroxide solution, so that hydrogen peroxide can be prevented from being decomposed in the purification process, and the preparation quality of the electronic grade hydrogen peroxide is improved.

According to one aspect of the present application, there is provided an automatic temperature control system for electronic grade hydrogen peroxide preparation, comprising: the temperature data acquisition module is used for acquiring temperature values of the hydrogen peroxide solution to be purified at a plurality of preset time points in a preset time period; the temperature relative data construction module is used for arranging the temperature values of the plurality of preset time points into temperature time sequence input vectors according to a time dimension, and then calculating the difference value between the temperature values of every two adjacent positions in the temperature time sequence input vectors to obtain temperature change time sequence input vectors; the temperature correlation coding module is used for performing correlation coding on the temperature time sequence input vector and the temperature change time sequence input vector to obtain a temperature static-dynamic correlation matrix; the matrix dividing module is used for carrying out matrix division on the temperature static-dynamic correlation matrix to obtain a plurality of temperature static-dynamic correlation submatrices; the temperature time sequence local association coding module is used for enabling the plurality of temperature static-dynamic association submatrices to pass through a convolutional neural network model serving as a filter to obtain a plurality of temperature static-dynamic local association feature vectors; a global context-dependent encoding module for passing the plurality of temperature static-dynamic local-dependent feature vectors through a converter-based context encoder to obtain a decoded feature vector; and a flow rate control module for passing the decoded feature vector through a decoder to obtain a decoded value, wherein the decoded value is used for representing a recommended flow rate value of circulating water in the cooling tower at the current time point.

In the above automatic temperature control system for electronic grade hydrogen peroxide preparation, the temperature-related encoding module is configured to: performing association coding on the temperature time sequence input vector and the temperature change time sequence input vector by using the following association formula to obtain a temperature static-dynamic association matrix; wherein, the formula is:

wherein->

Representing the temperature timing input vector, +.>

A transpose vector representing the temperature timing input vector, < >>

Representing the temperature variation time sequence input vector, +.>

Representing said obtained temperature static-dynamic correlation matrix,/->

Representing vector multiplication.

In the above automatic temperature control system for electronic grade hydrogen peroxide preparation, the local association coding module of temperature time sequence is used for: each layer of the convolutional neural network model used as the filter performs the following steps on input data in forward transfer of the layer: carrying out convolution processing on input data to obtain a convolution characteristic diagram; pooling the convolution feature images based on a feature matrix to obtain pooled feature images; performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the convolutional neural network as a filter is the plurality of temperature static-dynamic local correlation eigenvectors, and the input of the first layer of the convolutional neural network as a filter is the plurality of temperature static-dynamic correlation submatrices.

In the above-mentioned automatic temperature control system for electronic grade hydrogen peroxide preparation, the global context-related encoding module includes: the word embedding unit is used for mapping each temperature static-dynamic local association characteristic vector in the plurality of temperature static-dynamic local association characteristic vectors into a temperature static-dynamic local association embedding vector by using an embedding layer of the context encoder based on the converter so as to obtain a sequence of the temperature static-dynamic local association embedding vectors; a context coding unit, configured to perform global context semantic coding based on a converter thought on the sequence of temperature static-dynamic local association embedded vectors by using a converter of the converter-based context encoder to obtain a plurality of global context semantic temperature static-dynamic local association feature vectors; and a concatenation unit, configured to concatenate the plurality of global context semantic temperature static-dynamic local association feature vectors to obtain the decoded feature vector.

In the above-mentioned automatic temperature control system for electronic grade hydrogen peroxide preparation, the context coding unit includes: the query vector construction subunit is used for carrying out one-dimensional arrangement on the sequence of the temperature static-dynamic local association embedded vector to obtain a global temperature static-dynamic local association feature vector; a self-attention subunit, configured to calculate a product between the global feature vector and a transpose vector of each temperature static-dynamic local association embedding vector in the sequence of temperature static-dynamic local association embedding vectors to obtain a plurality of self-attention association matrices; the normalization subunit is used for respectively performing normalization processing on each self-attention correlation matrix in the plurality of self-attention correlation matrices to obtain a plurality of normalized self-attention correlation matrices; the attention calculating subunit is used for obtaining a plurality of probability values through a Softmax classification function by each normalized self-attention correlation matrix in the normalized self-attention correlation matrices; an attention applying subunit, configured to weight each of the temperature static-dynamic local association embedding vectors in the sequence of temperature static-dynamic local association embedding vectors with each of the plurality of probability values as a weight to obtain the plurality of context semantic temperature static-dynamic local association feature vectors; and a cascading subunit, configured to concatenate the plurality of context semantic temperature static-dynamic local association feature vectors to obtain the plurality of global context semantic temperature static-dynamic local association feature vectors.

In the above-mentioned automatic temperature control system for electronic grade hydrogen peroxide preparation, the system further comprises a training module for training the convolutional neural network model used as the filter, the context encoder based on the converter and the decoder.

In the above-mentioned automatic temperature control system for electronic grade hydrogen peroxide preparation, the training module includes: the training data acquisition module is used for acquiring training temperature values of the hydrogen peroxide solution to be purified at a plurality of preset time points in a preset time period and a true value of a recommended flow velocity value of circulating water in the cooling tower at the current time point; the training relative data construction module is used for arranging the training temperature values of the plurality of preset time points into training temperature time sequence input vectors according to the time dimension, and then calculating the difference value between the training temperature values of every two adjacent positions in the training temperature time sequence input vectors to obtain training temperature change time sequence input vectors; the training temperature association coding module is used for carrying out association coding on the training temperature time sequence input vector and the training temperature change time sequence input vector to obtain a training temperature static-dynamic association matrix; the feature optimization module is used for carrying out eigenvoice bitwise displacement association matching optimization on the training temperature static-dynamic association matrix to obtain an optimized training temperature static-dynamic association matrix; the training matrix dividing module is used for dividing the matrix of the optimized training temperature static-dynamic correlation matrix to obtain a plurality of training temperature static-dynamic correlation submatrices; the training temperature time sequence local correlation coding module is used for enabling the training temperature static-dynamic correlation submatrices to pass through the convolutional neural network model serving as a filter to obtain training temperature static-dynamic local correlation feature vectors; a training global context-dependent encoding module for passing the plurality of training temperature static-dynamic local-dependent feature vectors through the converter-based context encoder to obtain a training decoded feature vector; and a decoding loss module for passing the training decoding feature vector through the decoder to obtain a decoding loss function value; and a training module for training the convolutional neural network model as a filter, the converter-based context encoder, and the decoder based on the decoding loss function value and by back propagation of gradient descent.

In the above automatic temperature control system for electronic grade hydrogen peroxide preparation, the feature optimization module is configured to: performing eigen unitized bitwise displacement association matching optimization on the training temperature static-dynamic association matrix by using the following reinforcement formula to obtain the optimized training temperature static-dynamic association matrix; wherein, the formula is:

wherein->

Is the training temperature static-dynamic correlation matrix,

to->

Is obtained by carrying out eigen decomposition on the training temperature static-dynamic correlation matrix>

Intrinsic value->

For said->

The eigenvalues are arranged diagonally to obtain an eigenvoization matrix, and +.>

And->

Are all in the form of a diagonal matrix,

for the distance between the eigen-unitized matrix and the training temperature static-dynamic correlation matrix,/I>

Representing matrix multiplication +.>

Representing matrix addition, ++>

Representing multiplication by location +.>

And training a temperature static-dynamic correlation matrix for the optimization.

In the above automatic temperature control system for electronic grade hydrogen peroxide preparation, the decoding loss module is configured to: performing decoding regression on the training decoding feature vector using the decoder in the following formula to obtain a decoding loss function value; wherein, the formula is:

Wherein->

Representing the training decoded feature vector, +.>

Is the decoded value,/->

Is a weight matrix, < >>

Representing matrix multiplication.

According to another aspect of the present application, there is provided a temperature automatic control method for electronic grade hydrogen peroxide preparation, including: acquiring temperature values of hydrogen peroxide solution to be purified at a plurality of preset time points within a preset time period; after arranging the temperature values of the plurality of preset time points into temperature time sequence input vectors according to a time dimension, calculating the difference value between the temperature values of every two adjacent positions in the temperature time sequence input vectors to obtain temperature change time sequence input vectors; performing association coding on the temperature time sequence input vector and the temperature change time sequence input vector to obtain a temperature static-dynamic association matrix; performing matrix division on the temperature static-dynamic correlation matrix to obtain a plurality of temperature static-dynamic correlation submatrices; the temperature static-dynamic correlation submatrices are processed through a convolutional neural network model serving as a filter to obtain a plurality of temperature static-dynamic local correlation feature vectors; passing the plurality of temperature static-dynamic locally-associated feature vectors through a converter-based context encoder to obtain a decoded feature vector; and passing the decoded feature vector through a decoder to obtain a decoded value, wherein the decoded value is used for representing a recommended flow velocity value of circulating water in the cooling tower at the current time point.

According to still another aspect of the present application, there is provided an electronic apparatus including: a processor; and a memory in which computer program instructions are stored which, when executed by the processor, cause the processor to perform the automatic temperature control method for electronic grade hydrogen peroxide production as described above.

According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the temperature automatic control method for electronic grade hydrogen peroxide production as described above.

Compared with the prior art, the temperature automatic control system for preparing the electronic grade hydrogen peroxide provided by the application has the advantages that the time sequence change characteristic information of the temperature value of the hydrogen peroxide solution to be purified is dug out by adopting the neural network model based on deep learning, the circulating water flow velocity value in the cooling tower is further adaptively controlled based on the time sequence change characteristic of the temperature value of the hydrogen peroxide solution, and therefore decomposition of hydrogen peroxide in the purification process can be avoided, and the preparation quality of the electronic grade hydrogen peroxide is improved.

Drawings

The foregoing and other objects, features and advantages of the present application will become more apparent from the following more particular description of embodiments of the present application, as illustrated in the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.

Fig. 1 is a schematic view of a scenario of an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application.

Fig. 2 is a block diagram of an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application.

Fig. 3 is a block diagram of a training module in the automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application.

Fig. 4 is a system architecture diagram of an inference module in an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application.

Fig. 5 is a system architecture diagram of a training module in an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application.

Fig. 6 is a flowchart of convolutional neural network coding in an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application.

Fig. 7 is a block diagram of a global context-related encoding module in an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application.

Fig. 8 is a flowchart of a method for automatically controlling temperature for electronic grade hydrogen peroxide preparation according to an embodiment of the present application.

Fig. 9 is a block diagram of an electronic device according to an embodiment of the present application.

Detailed Description

Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.

Scene overview: as the background art mentioned above, the hydrogen peroxide solution obtained by the traditional preparation method contains a plurality of inorganic ions or compound impurities, which can not meet the use standard of the electronic industry, so that the hydrogen peroxide solution needs to be purified to remove the impurities so as to achieve the required high purity. At present, the hydrogen peroxide is purified mainly by a resin adsorption method, and circulating water in a cooling tower is required to be used for cooling and purifying in the resin adsorption method. However, in the existing cooling purification scheme, the flow rate of the circulating water in the cooling tower is constant, that is, the heat which can be taken away in unit time is relatively fixed, but the instability of hydrogen peroxide increases geometrically with the increase of temperature, so that the stability of hydrogen peroxide is difficult to ensure, and the decomposition of hydrogen peroxide is easy to be promoted. Therefore, an automatic temperature control system for electronic grade hydrogen peroxide preparation is desired.

Accordingly, in the actual preparation process of the electronic grade hydrogen peroxide, the flow rate control of the circulating water in the cooling tower is adapted to the temperature change condition during the purification of the hydrogen peroxide solution, that is, the flow rate value of the circulating water in the cooling tower is adaptively controlled based on the time sequence change characteristic of the temperature value of the hydrogen peroxide solution, so that the decomposition of the hydrogen peroxide is avoided, and the preparation quality of the electronic grade hydrogen peroxide is improved. However, as the instability of hydrogen peroxide increases geometrically along with the rise of temperature, the time sequence change characteristic of the temperature value of the hydrogen peroxide solution to be purified is small-scale change characteristic information relative to the whole temperature, and the precise capture and extraction are difficult to perform, so that the accuracy of controlling the flow velocity value of circulating water in the cooling tower is lower. Therefore, in the process, the difficulty is how to fully dig out the time sequence change characteristic information of the temperature value of the hydrogen peroxide solution to be purified, so as to enhance the time sequence change characteristic expression of the hydrogen peroxide solution, and to adaptively adjust the flow velocity value of circulating water in the cooling tower based on the time sequence change condition of the hydrogen peroxide solution accurately in real time, thereby avoiding the decomposition of hydrogen peroxide in the purification process and further improving the preparation quality of electronic grade hydrogen peroxide.

In recent years, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks have also shown levels approaching and even exceeding humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.

The development of deep learning and neural networks provides a new solution idea and scheme for mining time sequence change characteristic information of the temperature value of the hydrogen peroxide solution to be purified.

Specifically, in the technical scheme of the application, first, temperature values of hydrogen peroxide solution to be purified at a plurality of preset time points in a preset time period are obtained. Then, taking into consideration that the temperature value of the hydrogen peroxide solution to be purified has time sequence dynamic change characteristic information in a time dimension, and the time sequence dynamic change characteristic is a small-scale fine change characteristic relative to the whole temperature value, capturing and extracting are difficult to perform, so in the technical scheme of the application, the temperature values of a plurality of preset time points are arranged into temperature time sequence input vectors according to the time dimension, after the time sequence distribution information of the temperature value of the hydrogen peroxide solution to be purified is integrated, the difference value between the temperature values of every two adjacent positions in the temperature time sequence input vectors is calculated, so that the temperature relative dynamic change information of two adjacent time points is obtained, and the temperature change time sequence input vectors are obtained.

Then, considering that the absolute static change characteristic and the relative dynamic change characteristic of the temperature value of the hydrogen peroxide to be purified have a correlation relationship in time sequence, in order to improve the expression capability of the time sequence change characteristic of the temperature value of the hydrogen peroxide to be purified, the time sequence change characteristic capturing and extracting of the temperature value are fully performed, in the technical scheme of the application, the temperature time sequence input vector and the temperature change time sequence input vector are further subjected to correlation coding, so that correlation information between the absolute static change and the relative dynamic change of the temperature value of the hydrogen peroxide to be purified in the time dimension is constructed, and the subsequent extraction of fusion characteristics of the absolute static change and the relative dynamic change of the temperature value of the hydrogen peroxide to be purified is facilitated, so that a temperature static-dynamic correlation matrix is obtained.

Further, considering that the change characteristic of the temperature value of the hydrogen peroxide solution to be purified in the time dimension is weak, although the weak change characteristic information can be effectively captured by utilizing the correlation characteristic between the absolute static change and the relative dynamic change of the temperature value, the capturing capability of the change characteristic information can be reduced if only the static-dynamic time sequence correlation characteristic of the temperature value is globally extracted due to different static and dynamic change modes of the temperature value of the hydrogen peroxide solution to be purified under different time period spans. Therefore, in the technical scheme of the application, the temperature static-dynamic correlation matrix is further subjected to matrix division to obtain a plurality of temperature static-dynamic correlation sub-matrices, and the characteristic mining of each temperature static-dynamic correlation sub-matrix is performed by using a convolution neural network model which is used as a filter and has excellent performance in the aspect of local implicit correlation characteristic extraction, so that the high-dimensional implicit correlation characteristic distribution information between the absolute static time sequence change and the relative dynamic time sequence change of the temperature in each temperature static-dynamic correlation sub-matrix is respectively extracted, and a plurality of temperature static-dynamic local correlation characteristic vectors are obtained.

Then, it is considered that there is a correlation between the high-dimensional implicit correlation characteristic distribution information between the absolute static time series variation and the relative dynamic time series variation with respect to the temperature in the respective temperature static-dynamic correlation sub-matrices, that is, there is a time-series global-based correlation characteristic distribution information between the temperature variation characteristics of the respective local time periods with respect to the entirety of the predetermined time period. Therefore, in order to sufficiently capture and extract the dynamic change characteristics of the temperature value of the hydrogen peroxide solution to be purified in the time dimension so as to enhance the expression of the temperature time sequence dynamic characteristics, in the technical scheme of the application, the plurality of temperature static-dynamic local association characteristic vectors are further encoded in a context encoder based on a converter so as to extract the global association characteristic distribution information of the static and dynamic time sequence association change characteristics of the temperature value in each temperature static-dynamic association submatrix, and the global association characteristic distribution information is used as a decoding characteristic vector.

And then, further carrying out decoding regression on the decoding characteristic vector through a decoder to obtain a decoding value for representing the recommended flow velocity value of circulating water in the cooling tower at the current time point. That is, the static and dynamic time sequence correlation change characteristics based on the temperature value are subjected to decoding regression based on global correlation characteristic distribution information, so that the change characteristic of the temperature value of the hydrogen peroxide solution to be purified in the time dimension is obtained, and the flow velocity value of circulating water in an actual cooling tower is regulated and controlled according to the change characteristic, so that the decomposition of hydrogen peroxide in the purification process is avoided, and the preparation quality of electronic grade hydrogen peroxide is improved.

In particular, in the technical solution of the present application, here, regarding the temperature static-dynamic correlation matrix obtained by performing correlation encoding on the temperature time sequence input vector and the temperature change time sequence input vector, considering that an outlier inevitably occurs at the time of temperature measurement, the influence caused by such an outlier is amplified along with a series of processes of calculating dynamic transformation-static dynamic correlation-feature extraction-context encoding, so that an abnormal feature value which may deviate substantially from the overall feature distribution is introduced into a decoded feature vector obtained by directly cascading a plurality of context temperature static-dynamic local correlation feature vectors obtained by a context encoder based on a converter, thereby affecting the training effect of the model.

In the solution according to the present application, therefore, the temperature static-dynamic correlation matrix is preferably pre-determined, for example, as

Performing eigenvoization bitwise displacement associated matching optimization, expressed as:

，/>

to->

Is the temperature static-dynamic correlation matrix +.>

Intrinsic decomposition of the obtained->

Intrinsic value->

For said->

The eigenvalue matrix obtained by arranging the eigenvalues along a diagonal is also a diagonal matrix,/-j- >

For the eigenvoization matrix->

A static-dynamic correlation matrix with said temperature>

Distance between them.

That is, by static-dynamic correlation matrix based on the temperature

Is obtained by eigenvoization of the matrix ≡>

To static-dynamic correlation matrix for said temperature>

Performing bit-by-bit displacement correlation, and using the temperature static-dynamic correlation matrix +.>

Matching of characteristic association relation with respect to projection distance in eigen unitized space can solve the problem that model parameters are reversely propagated through the temperature static-dynamic association matrix>

At this time, due to the temperature static-dynamic correlation matrix +.>

The problem of mismatch of the optimization direction due to weak correlation distribution of local abnormal features of (a) is avoided by avoiding the temperature static-dynamic correlation matrix +.>

The eigenvalues at the edges of the regression target domain are mismatching constrained in the opposite optimization direction, resulting in poor training results. Therefore, the flow velocity value of circulating water in the cooling tower can be adaptively adjusted in real time and accurately based on the temperature time sequence change condition of the hydrogen peroxide solution, so that the hydrogen peroxide is prevented from being decomposed in the purification process, and the preparation quality of the electronic grade hydrogen peroxide is improved.

Based on this, the application provides a temperature automatic control system for electronic grade hydrogen peroxide preparation, it includes: the temperature data acquisition module is used for acquiring temperature values of the hydrogen peroxide solution to be purified at a plurality of preset time points in a preset time period; the temperature relative data construction module is used for arranging the temperature values of the plurality of preset time points into temperature time sequence input vectors according to a time dimension, and then calculating the difference value between the temperature values of every two adjacent positions in the temperature time sequence input vectors to obtain temperature change time sequence input vectors; the temperature correlation coding module is used for performing correlation coding on the temperature time sequence input vector and the temperature change time sequence input vector to obtain a temperature static-dynamic correlation matrix; the matrix dividing module is used for carrying out matrix division on the temperature static-dynamic correlation matrix to obtain a plurality of temperature static-dynamic correlation submatrices; the temperature time sequence local association coding module is used for enabling the plurality of temperature static-dynamic association submatrices to pass through a convolutional neural network model serving as a filter to obtain a plurality of temperature static-dynamic local association feature vectors; a global context-dependent encoding module for passing the plurality of temperature static-dynamic local-dependent feature vectors through a converter-based context encoder to obtain a decoded feature vector; and a flow rate control module for passing the decoded feature vector through a decoder to obtain a decoded value, wherein the decoded value is used for representing a recommended flow rate value of circulating water in the cooling tower at the current time point.

Fig. 1 is a schematic view of a scenario of an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application. As shown in fig. 1, in this application scenario, temperature values of the hydrogen peroxide solution to be purified at a plurality of predetermined time points within a predetermined period of time are acquired by a temperature sensor (e.g., T as illustrated in fig. 1). Then, the data are input into a server (for example, S in fig. 1) deployed with an automatic temperature control algorithm for electronic grade hydrogen peroxide preparation, wherein the server can process the input data by using the automatic temperature control algorithm for electronic grade hydrogen peroxide preparation to generate a decoding value, and the decoding value is used for representing a recommended flow velocity value of circulating water in a cooling tower at a current time point.

Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.

Exemplary System: fig. 2 is a block diagram of an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application. As shown in fig. 2, the automatic temperature control system 300 for preparing electronic grade hydrogen peroxide according to an embodiment of the present application includes an inference module, where the inference module includes: a temperature data acquisition module 310; a temperature versus data construction module 320; a temperature-dependent encoding module 330; a matrix partitioning module 340; a temperature timing local correlation encoding module 350; global context association encoding module 360; and a flow rate control module 370.

The temperature data acquisition module 310 is configured to acquire temperature values of a hydrogen peroxide solution to be purified at a plurality of predetermined time points within a predetermined time period; the temperature relative data construction module 320 is configured to arrange the temperature values at the plurality of predetermined time points into a temperature time sequence input vector according to a time dimension, and then calculate a difference value between the temperature values at each two adjacent positions in the temperature time sequence input vector to obtain a temperature change time sequence input vector; the temperature correlation encoding module 330 is configured to perform correlation encoding on the temperature time sequence input vector and the temperature change time sequence input vector to obtain a temperature static-dynamic correlation matrix; the matrix dividing module 340 is configured to perform matrix division on the temperature static-dynamic correlation matrix to obtain a plurality of temperature static-dynamic correlation sub-matrices; the temperature time sequence local correlation encoding module 350 is configured to pass the plurality of temperature static-dynamic correlation submatrices through a convolutional neural network model serving as a filter to obtain a plurality of temperature static-dynamic local correlation feature vectors; the global context-dependent encoding module 360 is configured to pass the plurality of temperature static-dynamic local-dependent feature vectors through a context encoder based on a converter to obtain a decoded feature vector; and the flow rate control module 370 is configured to pass the decoded feature vector through a decoder to obtain a decoded value, where the decoded value is used to represent a recommended flow rate value of circulating water in the cooling tower at a current point in time.

Fig. 4 is a system architecture diagram of an inference module in an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application. As shown in fig. 4, in the system architecture of the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation, in the process of inference, temperature values of hydrogen peroxide solution to be purified at a plurality of predetermined time points within a predetermined time period are first obtained through the temperature data acquisition module 310; next, the temperature relative data construction module 320 arranges the temperature values at a plurality of predetermined time points acquired by the temperature data acquisition module 310 into a temperature time sequence input vector according to a time dimension, and calculates a difference value between the temperature values at every two adjacent positions in the temperature time sequence input vector to obtain a temperature change time sequence input vector; the temperature correlation encoding module 330 performs correlation encoding on the temperature time sequence input vector obtained by the temperature relative data construction module 320 and the temperature change time sequence input vector to obtain a temperature static-dynamic correlation matrix; then, the matrix dividing module 340 performs matrix division on the temperature static-dynamic correlation matrix obtained by the temperature correlation encoding module 330 to obtain a plurality of temperature static-dynamic correlation sub-matrices; the temperature time sequence local association coding module 350 obtains a plurality of temperature static-dynamic local association feature vectors by passing the plurality of temperature static-dynamic association submatrices obtained by the matrix dividing module 340 through a convolutional neural network model serving as a filter; then, the global context correlation encoding module 360 passes the plurality of temperature static-dynamic local correlation feature vectors obtained by the temperature timing local correlation encoding module 350 through a context encoder based on a converter to obtain a decoding feature vector; further, the flow rate control module 370 passes the decoded feature vector obtained by the global context-dependent encoding module 360 through a decoder to obtain a decoded value representing a recommended flow rate value of the circulating water within the cooling tower at the current point in time.

Specifically, in the operation process of the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation, the temperature data acquisition module 310 is configured to acquire temperature values of the hydrogen peroxide solution to be purified at a plurality of predetermined time points within a predetermined time period. It should be understood that when the preparation process of electronic grade hydrogen peroxide is actually performed, in the process of cooling and purifying, the cooling and purifying are performed through the circulating water in the cooling tower, the flow velocity of the circulating water in the cooling tower is constant, so that the heat quantity which can be taken away in unit time is relatively fixed, but the instability of hydrogen peroxide increases geometrically along with the rise of temperature, therefore, in the technical scheme of the application, in order to avoid the decomposition of hydrogen peroxide caused by the instability of temperature, the flow velocity value of the circulating water in the cooling tower is further adaptively controlled based on the time sequence change characteristic of the temperature value of the hydrogen peroxide solution, so as to improve the preparation quality of electronic grade hydrogen peroxide. More specifically, in one specific example of the present application, first, temperature values of the hydrogen peroxide solution to be purified at a plurality of predetermined time points within a predetermined period of time may be obtained by a temperature sensor.

Specifically, in the operation process of the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation, the temperature relative data construction module 320 is configured to arrange the temperature values of the plurality of predetermined time points into a temperature time sequence input vector according to a time dimension, and then calculate a difference value between the temperature values of every two adjacent positions in the temperature time sequence input vector to obtain a temperature change time sequence input vector. In consideration of the fact that the temperature value of the hydrogen peroxide solution to be purified has time sequence dynamic change characteristic information in a time dimension, and the time sequence dynamic change characteristic is a small-scale fine change characteristic relative to the whole temperature value, capturing and extracting are difficult to achieve, in the technical scheme of the application, the temperature values of the plurality of preset time points are arranged into temperature time sequence input vectors according to the time dimension, after the time sequence distribution information of the temperature value of the hydrogen peroxide solution to be purified is integrated, the difference value between the temperature values of every two adjacent positions in the temperature time sequence input vectors is calculated, so that the temperature relative dynamic change information of two adjacent time points is obtained, and the temperature change time sequence input vectors are obtained.

Specifically, in the operation process of the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation, the temperature association encoding module 330 is configured to perform association encoding on the temperature time sequence input vector and the temperature change time sequence input vector to obtain a temperature static-dynamic association matrix. Taking into consideration the absolute static change characteristic and the relative dynamic change characteristic of the temperature value of the hydrogen peroxide to be purified on time sequenceIn order to improve the expression capability of time sequence change characteristics of the temperature value of the hydrogen peroxide to be purified, the time sequence change characteristic capturing and extracting of the temperature value are fully carried out, in the technical scheme of the application, the temperature time sequence input vector and the temperature change time sequence input vector are further subjected to associated coding, so that the associated information between absolute static change and relative dynamic change of the temperature value of the hydrogen peroxide to be purified in the time dimension is constructed, and the subsequent extraction of fusion characteristics of the absolute static change and the relative dynamic change is convenient, so that a temperature static-dynamic associated matrix is obtained. In a specific example of the application, the temperature time sequence input vector and the temperature change time sequence input vector are subjected to association coding by the following association formula to obtain a temperature static-dynamic association matrix; wherein, the formula is: the automatic temperature control system for preparing electronic grade hydrogen peroxide of 1, wherein the temperature-related encoding module is used for: performing association coding on the temperature time sequence input vector and the temperature change time sequence input vector by using the following association formula to obtain a temperature static-dynamic association matrix; wherein, the formula is:

Wherein->

Representing the temperature timing input vector, +.>

A transpose vector representing the temperature timing input vector, < >>

Representing the temperature variation time sequence input vector, +.>

Representing said obtained temperature static-dynamic correlation matrix,/->

Representing vector multiplication.

Specifically, in the operation process of the automatic temperature control system 300 for preparing electronic grade hydrogen peroxide, the matrix dividing module 340 is configured to perform matrix division on the temperature static-dynamic correlation matrix to obtain a plurality of temperature static-dynamic correlation submatrices. Considering that the change characteristic of the temperature value of the hydrogen peroxide solution to be purified in the time dimension is weak, although the weak change characteristic information can be effectively captured by utilizing the correlation characteristic between the absolute static change and the relative dynamic change of the temperature value, the capturing capability of the change characteristic information can be reduced if only the static-dynamic time sequence correlation characteristic of the temperature value is globally extracted because the temperature value of the hydrogen peroxide solution to be purified has different static and dynamic change modes in different time period spans. Therefore, in the technical scheme of the application, the temperature static-dynamic correlation matrix is further subjected to matrix division to obtain a plurality of temperature static-dynamic correlation submatrices.

Specifically, in the operation process of the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation, the temperature time sequence local association encoding module 350 is configured to pass the plurality of temperature static-dynamic association submatrices through a convolutional neural network model serving as a filter to obtain a plurality of temperature static-dynamic local association feature vectors. That is, the convolutional neural network with excellent performance in the aspect of local implicit correlation feature extraction is used for feature mining on the plurality of temperature static-dynamic correlation submatrices, so as to extract high-dimensional implicit correlation feature distribution information between absolute static time sequence change and relative dynamic time sequence change of the temperature in each temperature static-dynamic correlation submatrix respectively, and thus a plurality of temperature static-dynamic local correlation feature vectors are obtained. In one particular example, the convolutional neural network includes a plurality of neural network layers that are cascaded with one another, wherein each neural network layer includes a convolutional layer, a pooling layer, and an activation layer. In the coding process of the convolutional neural network, each layer of the convolutional neural network carries out convolutional processing based on a convolutional kernel on input data by using the convolutional layer in the forward transmission process of the layer, carries out pooling processing on a convolutional feature map output by the convolutional layer by using the pooling layer and carries out activation processing on the pooling feature map output by the pooling layer by using the activation layer.

Fig. 6 is a flowchart of convolutional neural network coding in an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application. As shown in fig. 6, in the encoding process of the convolutional neural network, the method includes: each layer of the convolutional neural network model used as the filter performs the following steps on input data in forward transfer of the layer: s210, carrying out convolution processing on input data to obtain a convolution characteristic diagram; s220, pooling the convolution feature map based on a feature matrix to obtain a pooled feature map; s230, carrying out nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the convolutional neural network as a filter is the plurality of temperature static-dynamic local correlation eigenvectors, and the input of the first layer of the convolutional neural network as a filter is the plurality of temperature static-dynamic correlation submatrices.

Specifically, during the operation of the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation, the global context-related encoding module 360 is configured to pass the plurality of temperature static-dynamic local-related feature vectors through a context encoder based on a converter to obtain a decoded feature vector. Considering that there is a correlation between the high-dimensional implicit correlation characteristic distribution information between the absolute static time sequence variation and the relative dynamic time sequence variation with respect to the temperature in the respective temperature static-dynamic correlation sub-matrices, that is, there is correlation characteristic distribution information based on the time sequence global with respect to the whole of the predetermined time period between the temperature variation characteristics of the respective local time periods. Therefore, in order to sufficiently capture and extract the dynamic change characteristics of the temperature value of the hydrogen peroxide solution to be purified in the time dimension so as to enhance the expression of the temperature time sequence dynamic characteristics, in the technical scheme of the application, the plurality of temperature static-dynamic local association characteristic vectors are further encoded in a context encoder based on a converter so as to extract the global association characteristic distribution information of the static and dynamic time sequence association change characteristics of the temperature value in each temperature static-dynamic association submatrix, and the global association characteristic distribution information is used as a decoding characteristic vector. In particular, the context encoder is based on a transducer model.

Fig. 7 is a block diagram of a global context-related encoding module in an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application. As shown in fig. 7, the global context-associated encoding module 360 includes: a word embedding unit 361, configured to map each of the plurality of temperature static-dynamic local association feature vectors into a temperature static-dynamic local association embedding vector by using an embedding layer of the context encoder based on the converter to obtain a sequence of temperature static-dynamic local association embedding vectors; a context encoding unit 362, configured to perform global context semantic encoding based on a converter concept on the sequence of temperature static-dynamic local association embedded vectors using a converter of the converter-based context encoder to obtain a plurality of global context semantic temperature static-dynamic local association feature vectors; and a concatenation unit 363, configured to concatenate the plurality of global context semantic temperature static-dynamic local association feature vectors to obtain the decoded feature vector. Wherein the context encoding unit 362 includes: the query vector construction subunit is used for carrying out one-dimensional arrangement on the sequence of the temperature static-dynamic local association embedded vector to obtain a global temperature static-dynamic local association feature vector; a self-attention subunit, configured to calculate a product between the global feature vector and a transpose vector of each temperature static-dynamic local association embedding vector in the sequence of temperature static-dynamic local association embedding vectors to obtain a plurality of self-attention association matrices; the normalization subunit is used for respectively performing normalization processing on each self-attention correlation matrix in the plurality of self-attention correlation matrices to obtain a plurality of normalized self-attention correlation matrices; the attention calculating subunit is used for obtaining a plurality of probability values through a Softmax classification function by each normalized self-attention correlation matrix in the normalized self-attention correlation matrices; an attention applying subunit, configured to weight each of the temperature static-dynamic local association embedding vectors in the sequence of temperature static-dynamic local association embedding vectors with each of the plurality of probability values as a weight to obtain the plurality of context semantic temperature static-dynamic local association feature vectors; and a cascading subunit, configured to concatenate the plurality of context semantic temperature static-dynamic local association feature vectors to obtain the plurality of global context semantic temperature static-dynamic local association feature vectors.

Specifically, during the operation of the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation, the flow rate control module 370 is configured to pass the decoding feature vector through a decoder to obtain a decoding value, where the decoding value is used to represent a recommended flow rate value of circulating water in the cooling tower at the current time point. In other words, in the technical scheme of the application, the decoding feature vector is decoded and regressed through a decoder to obtain a decoding value for representing a recommended flow velocity value of circulating water in a cooling tower at a current time point, that is, the decoding and regressing are performed based on global relevance feature distribution information based on static and dynamic time sequence relevance change features of the temperature value, so that the change characteristic of the temperature value of the hydrogen peroxide solution to be purified in the time dimension is obtained, and then the circulating water flow velocity value in an actual cooling tower is regulated and controlled according to the change characteristic, so that the hydrogen peroxide is prevented from being decomposed in the purification process, and the preparation quality of electronic grade hydrogen peroxide is improved. In one specific example of the present application, the decoder is used to perform a decoding regression on the decoded feature vector to obtain a decoding loss function value with the following formula;

Wherein, the formula is:

wherein->

Representing the training decoded feature vector,/>

is the decoded value,/->

Is a weight matrix, < >>

Representing matrix multiplication.

It should be appreciated that training the convolutional neural network model as a filter, the converter-based context encoder, and the decoder is required before the inference is made using the neural network model described above. That is, in the temperature automatic control system for preparing electronic grade hydrogen peroxide of the present application, the system further comprises a training module for training the convolutional neural network model as a filter, the context encoder based on the converter and the decoder. The training of deep neural networks mostly adopts a back propagation algorithm, and the back propagation algorithm updates the parameters of the current layer through errors transmitted by the later layer by using a chained method, which can suffer from the problem of gradient disappearance or more broadly, the problem of unstable gradient when the network is deep.

Fig. 3 is a block diagram of an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application. As shown in fig. 3, the automatic temperature control system 300 for preparing electronic grade hydrogen peroxide according to an embodiment of the present application further includes a training module 400, where the training module includes: a training data acquisition module 410; training the relative data construction module 420; training the temperature-dependent encoding module 430; a feature optimization module 440; training matrix partitioning module 450; training the temperature timing local correlation encoding module 460; training the global context correlation encoding module 470; and, a decode loss module 480; and a training module 490.

The training data acquisition module 410 is configured to acquire training temperature values of a hydrogen peroxide solution to be purified at a plurality of predetermined time points within a predetermined time period, and a true value of a recommended flow velocity value of circulating water in the cooling tower at the current time point; the training relative data construction module 420 is configured to arrange the training temperature values at the plurality of predetermined time points into training temperature time sequence input vectors according to a time dimension, and then calculate a difference value between the training temperature values at every two adjacent positions in the training temperature time sequence input vectors to obtain a training temperature change time sequence input vector; the training temperature correlation encoding module 430 is configured to perform correlation encoding on the training temperature time sequence input vector and the training temperature change time sequence input vector to obtain a training temperature static-dynamic correlation matrix; the feature optimization module 440 is configured to perform eigen-unitized bitwise displacement association matching optimization on the training temperature static-dynamic association matrix to obtain an optimized training temperature static-dynamic association matrix; the training matrix dividing module 450 is configured to perform matrix division on the optimized training temperature static-dynamic correlation matrix to obtain a plurality of training temperature static-dynamic correlation sub-matrices; the training temperature time sequence local correlation encoding module 460 is configured to pass the training temperature static-dynamic correlation submatrices through the convolutional neural network model as a filter to obtain a plurality of training temperature static-dynamic local correlation feature vectors; the training global context correlation encoding module 470, configured to pass the plurality of training temperature static-dynamic local correlation feature vectors through the converter-based context encoder to obtain training decoding feature vectors; and, the decoding loss module 480 for passing the training decoding feature vector through the decoder to obtain a decoding loss function value; and, the training module 490 for training the convolutional neural network model as a filter, the converter-based context encoder, and the decoder based on the decoding loss function value and by back propagation of gradient descent.

Fig. 5 is a system architecture diagram of a training module in an automatic temperature control system for electronic grade hydrogen peroxide preparation according to an embodiment of the present application. As shown in fig. 5, in the system architecture of the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation, in a training module 400, training temperature values of hydrogen peroxide solution to be purified at a plurality of preset time points in a preset time period and a true value of a recommended flow velocity value of circulating water in a cooling tower at the current time point are firstly obtained through the training data acquisition module 410; next, the training relative data construction module 420 arranges the training temperature values at a plurality of predetermined time points acquired by the training data acquisition module 410 into training temperature time sequence input vectors according to a time dimension, and calculates a difference value between the training temperature values at every two adjacent positions in the training temperature time sequence input vectors to obtain training temperature change time sequence input vectors; the training temperature association coding module 430 performs association coding on the training temperature time sequence input vector and the training temperature change time sequence input vector obtained by the training relative data construction module 420 to obtain a training temperature static-dynamic association matrix; then, the feature optimization module 440 performs eigen-unitized bitwise displacement correlation matching optimization on the training temperature static-dynamic correlation matrix obtained by the training temperature correlation encoding module 430 to obtain an optimized training temperature static-dynamic correlation matrix; the training matrix dividing module 450 performs matrix division on the optimized training temperature static-dynamic correlation matrix obtained by the feature optimizing module 440 to obtain a plurality of training temperature static-dynamic correlation submatrices; then, the training temperature time sequence local association coding module 460 passes the training temperature static-dynamic association submatrices obtained by the training matrix dividing module 450 through the convolutional neural network model as a filter to obtain a plurality of training temperature static-dynamic local association feature vectors; the training global context correlation encoding module 470 passes the plurality of training temperature static-dynamic local correlation feature vectors obtained by the training temperature time sequence local correlation encoding module 460 through the converter-based context encoder to obtain training decoding feature vectors; the decoding loss module 480 passes the training decoding feature vector through the decoder to obtain a decoding loss function value; further, the training module 490 trains the convolutional neural network model as a filter, the converter-based context encoder, and the decoder based on the decoding loss function values and by back propagation of gradient descent.

In the technical solution of the present application, here, regarding the temperature static-dynamic correlation matrix obtained by performing correlation encoding on the temperature time sequence input vector and the temperature variation time sequence input vector, it is considered that an abnormal value inevitably occurs at the time of temperature measurement, and the influence caused by such an abnormal value is amplified along with a series of processes of calculating dynamic transformation-static dynamic correlation-feature extraction-context encoding, so that an abnormal feature value which may deviate substantially from the overall feature distribution is introduced into a decoded feature vector obtained by directly cascading a plurality of context temperature static-dynamic local correlation feature vectors obtained by a context encoder based on a converter, thereby affecting the training effect of the model. In the solution according to the present application, therefore, the temperature static-dynamic correlation matrix is preferably pre-determined, for example, as

wherein->

Is the training temperature static-dynamic correlation matrix,

to->

Is obtained by carrying out eigen decomposition on the training temperature static-dynamic correlation matrix >

Intrinsic value->

For said->

The eigenvalues are along the diagonalArranging the obtained eigenvoization matrix, and +.>

And->

Are all in the form of a diagonal matrix,

Representing matrix multiplication +.>

Representing matrix addition, ++>

Representing multiplication by location +.>

And training a temperature static-dynamic correlation matrix for the optimization. That is, by static-dynamic correlation matrix based on the temperature +.>

Is obtained by eigenvoization of the matrix ≡>

To static-dynamic correlation matrix for said temperature>

At this time, due to the temperature static-dynamic correlation matrix +.>

In summary, the temperature automatic control system 300 for preparing electronic grade hydrogen peroxide according to the embodiment of the application is illustrated, and by adopting a neural network model based on deep learning to dig out time sequence variation characteristic information of a temperature value of hydrogen peroxide solution to be purified, the flow velocity value of circulating water in a cooling tower is further adaptively controlled based on the time sequence variation characteristic of the temperature value of the hydrogen peroxide solution, so that decomposition of hydrogen peroxide in the purification process can be avoided, and the preparation quality of electronic grade hydrogen peroxide is improved.

As described above, the automatic temperature control system for electronic grade hydrogen peroxide preparation according to the embodiments of the present application may be implemented in various terminal devices. In one example, the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation according to embodiments of the present application may be integrated into the terminal device as a software module and/or a hardware module. For example, the automatic temperature control system 300 for electronic grade hydrogen peroxide preparation may be a software module in the operating system of the terminal device, or may be an application program developed for the terminal device; of course, the automatic temperature control system 300 for preparing electronic grade hydrogen peroxide can be one of a plurality of hardware modules of the terminal device.

Alternatively, in another example, the temperature automatic control system 300 for electronic grade hydrogen peroxide preparation and the terminal device may be separate devices, and the temperature automatic control system 300 for electronic grade hydrogen peroxide preparation may be connected to the terminal device through a wired and/or wireless network, and transmit the interaction information according to a agreed data format.

An exemplary method is: fig. 8 is a flowchart of a method for automatically controlling temperature for electronic grade hydrogen peroxide preparation according to an embodiment of the present application. As shown in fig. 8, the method for automatically controlling the temperature for preparing electronic grade hydrogen peroxide according to the embodiment of the application includes the following steps: s110, acquiring temperature values of hydrogen peroxide solution to be purified at a plurality of preset time points in a preset time period; s120, after arranging the temperature values of the plurality of preset time points into temperature time sequence input vectors according to a time dimension, calculating the difference value between the temperature values of every two adjacent positions in the temperature time sequence input vectors to obtain temperature change time sequence input vectors; s130, performing association coding on the temperature time sequence input vector and the temperature change time sequence input vector to obtain a temperature static-dynamic association matrix; s140, performing matrix division on the temperature static-dynamic correlation matrix to obtain a plurality of temperature static-dynamic correlation submatrices; s150, passing the plurality of temperature static-dynamic correlation submatrices through a convolutional neural network model serving as a filter to obtain a plurality of temperature static-dynamic local correlation feature vectors; s160, enabling the plurality of temperature static-dynamic local association feature vectors to pass through a context encoder based on a converter to obtain decoding feature vectors; and S170, passing the decoding characteristic vector through a decoder to obtain a decoding value, wherein the decoding value is used for representing a recommended flow velocity value of circulating water in the cooling tower at the current time point.

In one example, in the above-mentioned automatic temperature control method for preparing electronic grade hydrogen peroxide, the step S130 includes: performing association coding on the temperature time sequence input vector and the temperature change time sequence input vector by using the following association formula to obtain a temperature static-dynamic association matrix; wherein, the formula is:

wherein->

Representing the temperature timing input vector, +.>

A transpose vector representing the temperature timing input vector, < >>

Representing the temperature variation time sequence input vector, +.>

Representing said obtained temperature static-dynamic correlation matrix,/->

Representing vector multiplication.

In one example, in the above-mentioned automatic temperature control method for preparing electronic grade hydrogen peroxide, the step S150 includes: each layer of the convolutional neural network model used as the filter performs the following steps on input data in forward transfer of the layer: carrying out convolution processing on input data to obtain a convolution characteristic diagram; pooling the convolution feature images based on a feature matrix to obtain pooled feature images; performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the convolutional neural network as a filter is the plurality of temperature static-dynamic local correlation eigenvectors, and the input of the first layer of the convolutional neural network as a filter is the plurality of temperature static-dynamic correlation submatrices.

In one example, in the above-mentioned automatic temperature control method for preparing electronic grade hydrogen peroxide, the step S160 includes: mapping each of the plurality of temperature static-dynamic local correlation feature vectors into a temperature static-dynamic local correlation embedded vector by using an embedded layer of the context encoder based on the converter so as to obtain a sequence of temperature static-dynamic local correlation embedded vectors; performing global context semantic coding based on a converter thought on the sequence of the temperature static-dynamic local association embedded vectors by using a converter of the converter-based context encoder to obtain a plurality of global context semantic temperature static-dynamic local association feature vectors; and cascading the plurality of global context semantic temperature static-dynamic local association feature vectors to obtain the decoding feature vector. The converter of the context encoder based on the converter is used for carrying out global context semantic coding based on a converter idea on the sequence of the temperature static-dynamic local association embedded vectors to obtain a plurality of global context semantic temperature static-dynamic local association feature vectors, and the method comprises the steps of carrying out one-dimensional arrangement on the sequence of the temperature static-dynamic local association embedded vectors to obtain global temperature static-dynamic local association feature vectors; calculating the product between the global feature vector and the transpose vector of each temperature static-dynamic local correlation embedded vector in the sequence of temperature static-dynamic local correlation embedded vectors to obtain a plurality of self-attention correlation matrices; respectively carrying out standardization processing on each self-attention correlation matrix in the plurality of self-attention correlation matrices to obtain a plurality of standardized self-attention correlation matrices; obtaining a plurality of probability values by using a Softmax classification function through each normalized self-attention correlation matrix in the normalized self-attention correlation matrices; weighting each temperature static-dynamic local association embedded vector in the sequence of the temperature static-dynamic local association embedded vectors by taking each probability value in the plurality of probability values as a weight so as to obtain a plurality of context semantic temperature static-dynamic local association feature vectors; and cascading the plurality of context semantic temperature static-dynamic local association feature vectors to obtain the plurality of global context semantic temperature static-dynamic local association feature vectors.

In summary, the temperature automatic control method for preparing electronic grade hydrogen peroxide according to the embodiment of the application is clarified, by adopting a neural network model based on deep learning to dig out time sequence change characteristic information of a temperature value of hydrogen peroxide solution to be purified, the flow velocity value of circulating water in a cooling tower is further adaptively controlled based on the time sequence change characteristic of the temperature value of the hydrogen peroxide solution, and thus, decomposition of hydrogen peroxide in the purification process can be avoided, and the preparation quality of electronic grade hydrogen peroxide is improved.

Exemplary electronic device: next, an electronic device according to an embodiment of the present application is described with reference to fig. 9.

Fig. 9 illustrates a block diagram of an electronic device according to an embodiment of the present application.

As shown in fig. 9, the electronic device 10 includes one or more processors 11 and a memory 12.

The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

Memory 12 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like. One or more computer program instructions may be stored on the computer readable storage medium that may be executed by the processor 11 to perform the functions in the automatic temperature control system for electronic grade hydrogen peroxide preparation and/or other desired functions of the various embodiments of the present application described above. Various content such as temperature static-dynamic local association feature vectors may also be stored in the computer-readable storage medium.

In one example, the electronic device 10 may further include: an input device 13 and an output device 14, which are interconnected by a bus system and/or other forms of connection mechanisms (not shown).

The input means 13 may comprise, for example, a keyboard, a mouse, etc.

The output device 14 can output various information including a decoded value and the like to the outside. The output means 14 may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, etc.

Of course, only some of the components of the electronic device 10 relevant to the present application are shown in fig. 9 for simplicity, components such as buses, input/output interfaces, etc. being omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.

Exemplary computer program product and computer readable storage Medium: in addition to the methods and apparatus described above, embodiments of the present application may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform steps in the functions of the automatic temperature control method for electronic grade hydrogen peroxide production described in the "exemplary systems" section of the present specification, according to various embodiments of the present application.

The computer program product may write program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server.

Furthermore, embodiments of the present application may also be a computer-readable storage medium, having stored thereon computer program instructions that, when executed by a processor, cause the processor to perform steps in the functions of the automatic temperature control method for electronic grade hydrogen peroxide production described in the above-described "exemplary systems" section of the present specification, according to various embodiments of the present application.

The computer readable storage medium may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

The basic principles of the present application have been described above in connection with specific embodiments, however, it should be noted that the advantages, benefits, effects, etc. mentioned in the present application are merely examples and not limiting, and these advantages, benefits, effects, etc. are not to be considered as necessarily possessed by the various embodiments of the present application. Furthermore, the specific details disclosed herein are for purposes of illustration and understanding only, and are not intended to be limiting, as the application is not intended to be limited to the details disclosed herein as such.

The block diagrams of the devices, apparatuses, devices, systems referred to in this application are only illustrative examples and are not intended to require or imply that the connections, arrangements, configurations must be made in the manner shown in the block diagrams. As will be appreciated by one of skill in the art, the devices, apparatuses, devices, systems may be connected, arranged, configured in any manner. Words such as "including," "comprising," "having," and the like are words of openness and mean "including but not limited to," and are used interchangeably therewith. The terms "or" and "as used herein refer to and are used interchangeably with the term" and/or "unless the context clearly indicates otherwise. The term "such as" as used herein refers to, and is used interchangeably with, the phrase "such as, but not limited to.

It is also noted that in the apparatus, devices and methods of the present application, the components or steps may be disassembled and/or assembled. Such decomposition and/or recombination should be considered as equivalent to 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, this description is not intended to limit the embodiments of the application to the form disclosed herein. Although a number of example aspects and embodiments have been discussed above, a person of ordinary skill in the art will recognize certain variations, modifications, alterations, additions, and subcombinations thereof.

Claims

1. A temperature automatic control system for electronic grade hydrogen peroxide preparation, characterized by comprising: the temperature data acquisition module is used for acquiring temperature values of the hydrogen peroxide solution to be purified at a plurality of preset time points in a preset time period; the temperature relative data construction module is used for arranging the temperature values of the plurality of preset time points into temperature time sequence input vectors according to a time dimension, and then calculating the difference value between the temperature values of every two adjacent positions in the temperature time sequence input vectors to obtain temperature change time sequence input vectors; the temperature correlation coding module is used for performing correlation coding on the temperature time sequence input vector and the temperature change time sequence input vector to obtain a temperature static-dynamic correlation matrix; the matrix dividing module is used for carrying out matrix division on the temperature static-dynamic correlation matrix to obtain a plurality of temperature static-dynamic correlation submatrices; the temperature time sequence local association coding module is used for enabling the plurality of temperature static-dynamic association submatrices to pass through a convolutional neural network model serving as a filter to obtain a plurality of temperature static-dynamic local association feature vectors; a global context-dependent encoding module for passing the plurality of temperature static-dynamic local-dependent feature vectors through a converter-based context encoder to obtain a decoded feature vector; and a flow rate control module for passing the decoded feature vector through a decoder to obtain a decoded value, wherein the decoded value is used for representing a recommended flow rate value of circulating water in the cooling tower at the current time point.

2. The automatic temperature control system for preparing electronic grade hydrogen peroxide according to claim 1, wherein the temperature-related encoding module is configured to: performing association coding on the temperature time sequence input vector and the temperature change time sequence input vector by using the following association formula to obtain a temperature static-dynamic association matrix; wherein, the formula is:

wherein->

Representing the temperature timing input vector, +.>

A transpose vector representing the temperature timing input vector, < >>

Representing the temperature variation time sequence input vector, +.>

Representing the resulting temperature static-dynamic correlation matrix,

representing vector multiplication.

3. The automatic temperature control system for preparing electronic grade hydrogen peroxide according to claim 2, wherein the temperature time sequence local association coding module is used for: each layer of the convolutional neural network model used as the filter performs the following steps on input data in forward transfer of the layer: carrying out convolution processing on input data to obtain a convolution characteristic diagram; pooling the convolution feature images based on a feature matrix to obtain pooled feature images; non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map; wherein the output of the last layer of the convolutional neural network as a filter is the plurality of temperature static-dynamic local correlation eigenvectors, and the input of the first layer of the convolutional neural network as a filter is the plurality of temperature static-dynamic correlation submatrices.

4. The automatic temperature control system for electronic grade hydrogen peroxide production of claim 3, wherein the global context-dependent encoding module comprises: the word embedding unit is used for mapping each temperature static-dynamic local association characteristic vector in the plurality of temperature static-dynamic local association characteristic vectors into a temperature static-dynamic local association embedding vector by using an embedding layer of the context encoder based on the converter so as to obtain a sequence of the temperature static-dynamic local association embedding vectors; a context coding unit, configured to perform global context semantic coding based on a converter thought on the sequence of temperature static-dynamic local association embedded vectors by using a converter of the converter-based context encoder to obtain a plurality of global context semantic temperature static-dynamic local association feature vectors; and a concatenation unit, configured to concatenate the plurality of global context semantic temperature static-dynamic local association feature vectors to obtain the decoded feature vector.

5. The automatic temperature control system for electronic grade hydrogen peroxide production of claim 4, wherein the context encoding unit comprises: the query vector construction subunit is used for carrying out one-dimensional arrangement on the sequence of the temperature static-dynamic local association embedded vector to obtain a global temperature static-dynamic local association feature vector; a self-attention subunit, configured to calculate a product between the global feature vector and a transpose vector of each temperature static-dynamic local association embedding vector in the sequence of temperature static-dynamic local association embedding vectors to obtain a plurality of self-attention association matrices; the normalization subunit is used for respectively performing normalization processing on each self-attention correlation matrix in the plurality of self-attention correlation matrices to obtain a plurality of normalized self-attention correlation matrices; the attention calculating subunit is used for obtaining a plurality of probability values through a Softmax classification function by each normalized self-attention correlation matrix in the normalized self-attention correlation matrices; an attention applying subunit, configured to weight each of the temperature static-dynamic local association embedding vectors in the sequence of temperature static-dynamic local association embedding vectors with each of the plurality of probability values as a weight to obtain the plurality of context semantic temperature static-dynamic local association feature vectors; and a cascading subunit, configured to cascade the plurality of context semantic temperature static-dynamic local association feature vectors to obtain the plurality of global context semantic temperature static-dynamic local association feature vectors.

6. The automatic temperature control system for electronic grade hydrogen peroxide production of claim 5, further comprising a training module that trains the convolutional neural network model as a filter, the converter-based context encoder, and the decoder.

7. The automatic temperature control system for electronic grade hydrogen peroxide production of claim 6, wherein the training module comprises: the training data acquisition module is used for acquiring training temperature values of the hydrogen peroxide solution to be purified at a plurality of preset time points in a preset time period and a true value of a recommended flow velocity value of circulating water in the cooling tower at the current time point; the training relative data construction module is used for arranging the training temperature values of the plurality of preset time points into training temperature time sequence input vectors according to the time dimension, and then calculating the difference value between the training temperature values of every two adjacent positions in the training temperature time sequence input vectors to obtain training temperature change time sequence input vectors; the training temperature association coding module is used for carrying out association coding on the training temperature time sequence input vector and the training temperature change time sequence input vector to obtain a training temperature static-dynamic association matrix; the feature optimization module is used for carrying out eigenvoice bitwise displacement association matching optimization on the training temperature static-dynamic association matrix to obtain an optimized training temperature static-dynamic association matrix; the training matrix dividing module is used for dividing the matrix of the optimized training temperature static-dynamic correlation matrix to obtain a plurality of training temperature static-dynamic correlation submatrices; the training temperature time sequence local correlation coding module is used for enabling the training temperature static-dynamic correlation submatrices to pass through the convolutional neural network model serving as a filter to obtain training temperature static-dynamic local correlation feature vectors; a training global context-dependent encoding module for passing the plurality of training temperature static-dynamic local-dependent feature vectors through the converter-based context encoder to obtain a training decoded feature vector; and a decoding loss module for passing the training decoding feature vector through the decoder to obtain a decoding loss function value; and a training module for training the convolutional neural network model as a filter, the converter-based context encoder, and the decoder based on the decoding loss function value and by back propagation of gradient descent.

8. The automatic temperature control system for electronic grade hydrogen peroxide production of claim 7, wherein the feature optimization module is configured to: performing eigen unitized bitwise displacement association matching optimization on the training temperature static-dynamic association matrix by using the following reinforcement formula to obtain the optimized training temperature static-dynamic association matrix; wherein, the formula is:

wherein->

Is the training temperature static-dynamic correlation matrix, < >>

To->

The number of eigenvalues is set,

for said->

And->

Are all diagonal matrix>

Representing matrix multiplication +.>

Representing matrix addition, ++>

Representing multiplication by location +.>

9. The automatic temperature control system for electronic grade hydrogen peroxide production of claim 8, wherein the decode-and-lose module is configured to: performing decoding regression on the training decoding feature vector using the decoder in the following formula to obtain a decoding loss function value; wherein, the formula is:

Wherein->

Representing the training decoded feature vector, +.>

Is the decoded value,/->

Is a weight matrix, < >>

Representing matrix multiplication. />