CN116678258A

CN116678258A - Cold-heat exchanger for pressure vessel and control method thereof

Info

Publication number: CN116678258A
Application number: CN202310629221.4A
Authority: CN
Inventors: 叶挺朋; 吴素珍
Original assignee: Zhejiang Jinggang Pressure Vessel Co ltd
Current assignee: Zhejiang Jinggang Pressure Vessel Co ltd
Priority date: 2023-05-31
Filing date: 2023-05-31
Publication date: 2023-09-01

Abstract

The application relates to the field of intelligent control, and particularly discloses a cold-heat exchanger for a pressure vessel and a control method thereof.

Description

Cold-heat exchanger for pressure vessel and control method thereof

Technical Field

The present application relates to the field of intelligent control, and more particularly, to a cold-heat exchanger for a pressure vessel and a control method thereof.

Background

The pressure vessel is a closed device for containing gas or liquid and needs to bear a certain pressure. The pressure vessel is widely applied to various fields of petroleum, chemical industry, aerospace, medical treatment and the like, and is divided into a low-pressure vessel, a medium-pressure vessel, a high-pressure vessel and an ultrahigh-pressure vessel according to pressure class. The horizontal pressure vessel is one of the most common and widely used pressure vessels.

In use, the fluid or gas in the pressure vessel is often required to reach a certain temperature, so a cold heat exchanger is typically installed in the pressure vessel to maintain the fluid temperature within a predetermined fluctuation range by the cold heat exchanger. However, in the actual use process, the temperature of the fluid is found to exceed the preset fluctuation range, which causes trouble to the normal use of the fluid. The reason is found that: in use, the fluid velocity changes, and the control of the working power of the cold-heat exchanger has hysteresis, and once the fluid flow velocity changes, the cold-heat exchanger does not track in time, so that the fluid temperature exceeds a preset fluctuation range.

Therefore, an optimized cold-heat exchanger for a pressure vessel is desired.

Disclosure of Invention

The present application has been made to solve the above-mentioned technical problems. The embodiment of the application provides a cold-heat exchanger for a pressure vessel and a control method thereof, which are used for accurately evaluating and predicting the fluid flow velocity value at the next time point by adopting a neural network model based on deep learning to mine dynamic change characteristic information of the fluid flow velocity value in the time dimension, so as to accurately control the power of the cold-heat exchanger at the next time point, thereby maintaining the fluid temperature within a preset fluctuation range and ensuring the normal and safe use of the pressure vessel.

According to one aspect of the present application, there is provided a cold-heat exchanger for a pressure vessel, comprising: the flow rate acquisition module is used for acquiring fluid flow rate values of a plurality of preset time points in a preset time period; a vector construction module for arranging the fluid flow velocity values of the plurality of preset time points into a flow velocity input vector according to a time dimension; the relative flow velocity calculation module is used for calculating the difference value between the fluid flow velocity values of every two adjacent positions in the flow velocity input vector to obtain a flow velocity change input vector; the flow velocity change feature extraction module is used for respectively passing the flow velocity input vector and the flow velocity change input vector through the multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector; the multi-scale change feature fusion module is used for fusing the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector based on a Gaussian density chart to obtain a fusion feature matrix; the fusion optimization module is used for carrying out vector spectrum clustering agent learning fusion optimization on the fusion feature matrix to obtain an optimized fusion feature matrix; the decoding module is used for enabling the optimized fusion characteristic matrix to pass through a decoder to obtain a decoding value, wherein the decoding value is used for representing a fluid flow velocity value at the next time point; and

And the instruction generation module is used for generating a power control instruction of the cold-heat exchanger at the next time point based on the decoding value.

In the above-mentioned cold-heat exchanger for a pressure vessel, the multi-scale neighborhood feature extraction module comprises: the device comprises a first convolution layer, a second convolution layer parallel to the first convolution layer and a multi-scale feature fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first length, and the second convolution layer uses a one-dimensional convolution kernel with a second length.

In the above-described cold-heat exchanger for a pressure vessel, the flow rate variation feature extraction module includes: the first neighborhood scale feature extraction unit is used for inputting the flow velocity input vector and the flow velocity change input vector into a first convolution layer of the multi-scale neighborhood feature extraction module respectively to obtain a first neighborhood scale flow velocity time sequence feature vector and a first neighborhood scale flow velocity change time sequence feature vector, wherein the first convolution layer is provided with a first one-dimensional convolution kernel with a first length; a second neighborhood scale feature extraction unit, configured to input the flow velocity input vector and the flow velocity variation input vector into a second convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a second neighborhood scale flow velocity time sequence feature vector and a second neighborhood scale flow velocity variation time sequence feature vector, where the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and the multi-scale cascading unit is used for cascading the first neighborhood scale flow velocity time sequence feature vector and the second neighborhood scale flow velocity time sequence feature vector to obtain the flow velocity time sequence feature vector, and cascading the first neighborhood scale flow velocity change time sequence feature vector and the second neighborhood scale flow velocity change time sequence feature vector to obtain the flow velocity change time sequence feature vector.

In the above-mentioned cold-heat exchanger for a pressure vessel, the multi-scale variation feature fusion module includes: a gaussian fusion unit, configured to fuse the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector by using the Gao Simi degree map to obtain an initial fused gaussian density map, where a mean vector of the initial fused gaussian density map is a mean vector between the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector, and a value of each position in a covariance matrix of the initial fused gaussian density map is a variance between feature values of corresponding two positions in the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector; the Gaussian discretization unit is used for carrying out Gaussian discretization on the Gaussian distribution of each position in the initial fusion Gaussian density map so as to obtain an initial fusion feature matrix; the weighted feature extraction unit is used for enabling the initial fusion feature matrix to pass through a convolutional neural network model with the same channel number as the length of the flow velocity time sequence feature vector so as to obtain a weighted feature map; the distinguishing unit is used for modeling the weighted feature map by a feature correlation accumulation distinguishing mechanism so as to obtain a weighted feature vector; the weighting optimization unit is used for performing point multiplication on the weighting feature vector and the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector respectively to obtain an optimized flow velocity time sequence feature vector and an optimized flow velocity change time sequence feature vector; and the fusion unit is used for carrying out association coding on the optimized flow velocity time sequence feature vector and the optimized flow velocity change time sequence feature vector so as to obtain the fusion feature matrix.

In the above-described cold-heat exchanger for a pressure vessel, the weighted feature extraction unit is configured to: performing point multiplication on the weighted eigenvector and the flow velocity time sequence eigenvector and the flow velocity change time sequence eigenvector respectively by the following formula to obtain an optimized flow velocity time sequence eigenvectorAnd optimizing the flow rate variation timing feature vector; wherein, the formula is:wherein->Representing the weighted feature vector,/->Representing the flow rate timing characteristic vector and the flow rate variation timing characteristic vector,/or->Representing the optimized flow rate timing feature vector and the optimized flow rate variation timing feature vector, +.>Representing multiplication by location.

In the above-described cold-heat exchanger for a pressure vessel, the differentiating unit is further configured to: modeling the weighted feature map by a feature correlation accumulation differentiation mechanism according to the following formula to obtain the weighted feature vector; wherein, the formula is:wherein->Representing the weighted feature map, ">And->Respectively representing the single-layer convolution operation based on different convolution kernels on the feature map,/and>representation->Activating function->Representation->Activate function, and->Representing global pooling of each feature matrix of the feature map,/for each feature matrix >Representing addition by position +.>Representing the weighted feature vector.

In the above-described cold-heat exchanger for a pressure vessel, the fusion unit is configured to: performing association coding on the optimized flow velocity time sequence feature vector and the optimized flow velocity change time sequence feature vector by using the following formula to obtain the fusion feature matrix; wherein, the formula is:wherein->Representing the optimized flow rate timing feature vector, +.>Transpose of the optimized flow timing feature vector, < >>Time sequence characteristic vector representing the optimized flow rate change, < >>Representing the fusion feature matrix,>representing vector multiplication.

In the above-described cold-heat exchanger for a pressure vessel, the fusion is excellentA chemical module for: vector spectral clustering agent learning fusion optimization is carried out on the fusion feature matrix according to the following formula so as to obtain the optimized fusion feature matrix; wherein, the formula is:wherein (1)>Is the fusion feature matrix,>is the optimized fusion feature matrix, +.>Representing individual row feature vectors of the fusion feature matrix, and +.>Is a distance matrix composed of the distances between every two corresponding row feature vectors of the fusion feature matrix,/and >An exponential operation representing a matrix representing a natural exponential function value raised to a power by a characteristic value of each position in the matrix, ">And->Respectively representing dot-by-location multiplication and matrix addition.

In the above-described cold-heat exchanger for a pressure vessel, the decoding module is configured to: passing the optimized fusion feature matrix through a decoder using the decoder to obtain a decoded value representing a fluid flow rate value at a next point in time; wherein, the formula is:wherein->Representing the optimized fusion feature matrix, +.>Is the decoded value,/->Is a weight matrix, < >>Representing matrix multiplication.

According to another aspect of the present application, there is provided a control method of a cold heat exchanger for a pressure vessel, comprising: acquiring fluid flow rate values at a plurality of preset time points in a preset time period; arranging the fluid flow velocity values of the plurality of preset time points into flow velocity input vectors according to a time dimension; calculating the difference between the fluid flow velocity values of every two adjacent positions in the flow velocity input vector to obtain a flow velocity change input vector; respectively passing the flow velocity input vector and the flow velocity change input vector through a multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector; fusing the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector based on a Gaussian density chart to obtain a fusion feature matrix; passing the optimized fusion feature matrix through a decoder to obtain a decoded value, wherein the decoded value is used for representing a fluid flow velocity value at a next time point; and

And generating a power control instruction of the cold-heat exchanger at the next time point based on the decoded value.

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 control method for a cold heat exchanger of a pressure vessel as described above.

According to a further 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 control method for a cold heat exchanger of a pressure vessel as described above.

Compared with the prior art, the cold and heat exchanger for the pressure vessel and the control method thereof provided by the application have the advantages that the dynamic change characteristic information of the fluid flow velocity value in the time dimension is mined by adopting the neural network model based on deep learning, so that the fluid flow velocity value at the next time point is accurately estimated and predicted, the power of the cold and heat exchanger at the next time point is accurately controlled, the fluid temperature is maintained within the preset fluctuation range, and the normal and safe use of the pressure vessel is ensured.

Drawings

The above and other objects, features and advantages of the present application will become more apparent by describing embodiments of the present application in more detail with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and together with the embodiments of the application, and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.

FIG. 1 is an application scenario diagram of a cold heat exchanger for a pressure vessel according to an embodiment of the present application; FIG. 2 is a block diagram of a cold heat exchanger for a pressure vessel according to an embodiment of the present application; FIG. 3 is a system architecture diagram of a cold heat exchanger for a pressure vessel according to an embodiment of the present application; FIG. 4 is a block diagram of a flow rate variation feature extraction module in a cold-heat exchanger for a pressure vessel in accordance with an embodiment of the application; FIG. 5 is a block diagram of a multi-scale variation feature fusion module in a cold-heat exchanger for a pressure vessel in accordance with an embodiment of the present application; FIG. 6 is a flow chart of a method of controlling a cold heat exchanger for a pressure vessel according to an embodiment of the present application; fig. 7 is a block diagram of an electronic device according to an embodiment of the application.

Detailed Description

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

Summary of the application: as described above, in the process of maintaining the fluid temperature within the predetermined fluctuation range using the cold-heat exchanger in practice, it is found that the fluid temperature often goes beyond the predetermined fluctuation range, which causes trouble to the normal use of the fluid. The reason is found that: in use, the fluid velocity changes, and the control of the working power of the cold-heat exchanger has hysteresis, and once the fluid flow velocity changes, the cold-heat exchanger does not track in time, so that the fluid temperature exceeds a preset fluctuation range. Therefore, an optimized cold-heat exchanger for a pressure vessel is desired.

Accordingly, considering that in the actual working process of the cold-heat exchanger, if the working power of the cold-heat exchanger is to be accurately controlled in real time so as to achieve the purpose of maintaining the fluid temperature within the predetermined fluctuation range, so as to avoid the fluid temperature exceeding the predetermined fluctuation range, it is necessary to evaluate and predict the fluid flow rate value at the next time point based on the change condition of the fluid flow rate in the time dimension, and further correspondingly control the power of the cold-heat exchanger. However, since the time-series variation of the fluid flow rate value is small-scale information with respect to the fluid flow rate value, it is difficult to capture and acquire such small-scale variation information, which reduces the accuracy of prediction of the fluid flow rate value for the next time point. Therefore, in this process, the difficulty lies in how to mine the dynamic change characteristic information of the fluid flow velocity value in the time dimension, so as to accurately evaluate and predict the fluid flow velocity value at the next time point, and further accurately control the power of the cold-heat exchanger at the next time point, so as to maintain the fluid temperature within the preset fluctuation range, and ensure the normal and safe use of the pressure vessel.

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 new solutions and solutions for mining the dynamically changing characteristic information of the fluid flow velocity values in the time dimension.

Specifically, in the technical scheme of the present application, first, fluid flow velocity values at a plurality of predetermined time points within a predetermined period are acquired. Next, considering that the fluid flow velocity value has a dynamic change rule in the time dimension, in order to facilitate the extraction of the dynamic change characteristic information of the fluid flow velocity value in the time dimension, the fluid flow velocity values at a plurality of preset time points are further arranged into flow velocity input vectors according to the time dimension, so that the distribution information of the fluid flow velocity value in the time sequence is integrated.

Then, in order to extract the dynamic change characteristic of the fluid flow rate value in the time dimension, so as to accurately predict the fluid flow rate value at the next time point, it is necessary to extract the time sequence change characteristic about the fluid flow rate in the flow rate input vector, and considering that the time sequence change condition of the fluid flow rate value is small-scale information relative to the fluid flow rate value, if the evaluation prediction of the fluid flow rate value at the next time point is performed by absolute change information, not only the calculated amount is large, but also the small-scale characteristic information of the flow rate change is difficult to capture and detect, and the accuracy of the subsequent decoding prediction is affected. Therefore, in the technical solution of the present application, it is desirable to use a combination of a relative change feature and an absolute change feature of the fluid flow velocity value in the time dimension to improve capturing of time-series dynamic changes of the fluid flow velocity value, so as to improve the prediction accuracy of the fluid flow velocity value at the next time point.

Specifically, first, the difference between the fluid flow velocity values at every adjacent two positions in the flow velocity input vector is calculated to obtain a flow velocity variation input vector. Then, considering that the fluid flow velocity value has volatility and uncertainty in the time dimension and has different dynamic change mode information under different time period spans in the preset time period, in the technical scheme of the application, the flow velocity input vector and the flow velocity change input vector are further subjected to feature mining respectively through a multi-scale neighborhood feature extraction module so as to extract dynamic multi-scale neighborhood associated features of the relative quantity and the absolute quantity of the fluid flow velocity under different time spans in the preset time period respectively, thereby obtaining a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector.

Next, in order to fuse the relative time series dynamic change feature and the absolute time series dynamic change feature of the fluid flow velocity value, further considering that the flow velocity time series feature vector and the flow velocity change time series feature vector each correspond to one feature distribution manifold in a high-dimensional feature space, and these feature distribution manifolds are very easy to sink into local extremum points when finding an optimum point by gradient descent, if the time series dynamic change feature of the fluid flow velocity is represented by cascading only the feature vectors of the relative amount data and the absolute amount data of the fluid flow velocity value, it would be quite easy to superimpose these feature distribution manifolds in original positions and shapes, so that the boundaries of the newly obtained feature distribution manifolds become very irregular and complex, and global optimum points cannot be obtained. Therefore, it is necessary to further appropriately fuse the flow velocity timing feature vector and the flow velocity variation timing feature vector so that the respective feature distributions can converge on the profile with respect to each other.

Further, considering that gaussian density maps are widely used in deep learning for a priori based estimation of target posterior, they can be used to correct data distribution to achieve the above objective. Specifically, in the technical scheme of the application, firstly, a flow velocity Gaussian density chart of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector is constructed, so that an absolute quantity dynamic multi-scale change feature and a relative quantity dynamic multi-scale change feature of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector in relation to the fluid flow velocity in a time dimension are fused; and then, carrying out Gaussian discretization processing on the flow velocity Gaussian density map so as not to generate information loss when the data characteristics are amplified, and further improving the accuracy of subsequent decoding, thereby obtaining a fusion characteristic matrix.

Then, the fusion feature matrix is subjected to decoding processing in a decoder to obtain a decoded value for representing the fluid flow velocity value at the next time point. That is, a decoding regression is performed based on time-series dynamic change characteristic information of relative and absolute amounts of the fluid flow rate values, thereby accurately evaluating predictions for fluid flow rate values at the next time point based on change characteristics of the fluid flow rate values in the time dimension. Then, based on the decoded value, a power control command of the cold-heat exchanger at the next time point is generated, so that the power of the cold-heat exchanger at the next time point is accurately controlled, and the fluid temperature is maintained within a predetermined fluctuation range. Accordingly, in one specific example of the present application, in performing the power control of the heat and cold exchanger, if the decoded value is greater than the fluid flow rate value at the current time point, the power control command of the heat and cold exchanger at the next time point is to increase the power of the heat and cold exchanger, and if the decoded value is less than the fluid flow rate value at the current time point, the power control command of the heat and cold exchanger at the next time point is to decrease the power of the heat and cold exchanger.

Here, since the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector are fused based on the gaussian density chart, the mean feature vector of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector is subjected to gaussian discretization based on two-dimensional gaussian probability distribution, so that the fused feature matrix does not distinguish the importance of feature values of each position of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector, namely, does not reflect the confidence of each feature value of the feature vector, thereby influencing the expression effect of the fused feature matrix.

Based on this, in the technical scheme of the application, the fusion feature matrix is first passed through a convolutional neural network model with the same channel number as the length of the flow velocity time sequence feature vector to obtain a weighted feature mapThe weighted feature map is further +.>Modeling of a feature correlation cumulative discrimination mechanism to obtain a weighted feature vector +.>，/>Expressed as:wherein (1)>And->Respectively, the weighted characteristic diagram is subjected to single-layer convolution operation based on different convolution kernels, and +.>Representing a global pooling operation on each feature matrix of the weighted feature graph.

Here, the feature correlation accumulating and distinguishing mechanism modeling firstly generates two new local association units of the weighted feature map through convolution operation, then uses Sigmoid function and ReLU function to perform simple embedding, resetting and updating similar to a neural network architecture on the local association features, and then accumulates the correlation of the local features relative to the whole features through global average pooling operation, so that the feature importance sequence is explicitly modeled by using the feature distinguishing mechanism, and then the proper weighting factors in the channel dimension can be determined based on the feature accumulating and distinguishing mechanism of each feature matrix of the weighted feature map.

Then, the weighted feature vector is further processedRespectively with the flow velocity time sequenceAnd carrying out point multiplication on the sign vector and the flow velocity change time sequence feature vector to distinguish the importance of the feature values of each position of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector based on the importance degree of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector in the associated feature domain, thereby improving the expression effect of the fusion feature matrix obtained after fusion. Therefore, the fluid flow velocity value at the next time point can be accurately estimated and predicted based on the actual fluid flow velocity change condition, and the power of the cold-heat exchanger at the next time point is accurately controlled, so that the temperature of the fluid is maintained within a preset fluctuation range, and the normal and safe use of the pressure vessel is ensured.

Here, since the flow velocity time series eigenvector and the flow velocity change time series eigenvector are fused based on a gaussian density chart, gaussian discretization based on two-dimensional gaussian probability distribution is performed on each eigenvalue of the flow velocity time series eigenvector and the mean eigenvector of the flow velocity change time series eigenvector based on the corresponding row and column of the variance matrix of the flow velocity time series eigenvector and the flow velocity change time series eigenvector, which makes the fused eigenvector can be regarded as an eigenvector obtained by stitching the individual row or column eigenvectors. Therefore, when classifying the fusion feature matrix, the integration of the feature distribution of the fusion feature matrix is expected to be improved, so that the classification effect of the fusion feature matrix is improved.

Based on this, the applicant of the present application refers to the fusion feature matrix, e.g. denoted asVector spectral clustering agent learning fusion optimization is carried out, and the vector spectral clustering agent learning fusion optimization is expressed as follows: />Wherein (1)>Representing the fusion feature matrix->Is a column or row feature vector, and +.>Is a distance matrix of distances between the respective vectors.

Here, in the fusion feature matrixWhen the feature vectors of each row or column are spliced and then classified by a classifier, the internal quasi-regression semantic features of the feature vectors of each row or column are mixed with the synthesized noise features, so that the ambiguity of the boundary between the meaningful quasi-regression semantic features and the noise features is caused, and therefore, the vector spectral clustering agent learning fusion optimization utilizes the conceptual information used for representing the association between the quasi-regression semantic features and the quasi-regression scene to carry out hidden supervision propagation on the potential association attribute between the feature vectors of each row or column by introducing the spectral clustering agent learning used for representing the distribution layout and the semantic distribution similarity between the feature vectors, so that the overall distribution dependence of the fusion feature matrix serving as the synthesized feature of the feature vectors of each row or column is improved, and the classification effect of the fusion feature matrix through the classifier is improved.

Based on this, the application proposes a cold-heat exchanger for a pressure vessel, comprising: the flow rate acquisition module is used for acquiring fluid flow rate values of a plurality of preset time points in a preset time period; a vector construction module for arranging the fluid flow velocity values of the plurality of preset time points into a flow velocity input vector according to a time dimension; the relative flow velocity calculation module is used for calculating the difference value between the fluid flow velocity values of every two adjacent positions in the flow velocity input vector to obtain a flow velocity change input vector; the flow velocity change feature extraction module is used for respectively passing the flow velocity input vector and the flow velocity change input vector through the multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector; the multi-scale change feature fusion module is used for fusing the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector based on a Gaussian density chart to obtain a fusion feature matrix; the fusion optimization module is used for carrying out vector spectrum clustering agent learning fusion optimization on the fusion feature matrix to obtain an optimized fusion feature matrix; the decoding module is used for enabling the optimized fusion characteristic matrix to pass through a decoder to obtain a decoding value, wherein the decoding value is used for representing a fluid flow velocity value at the next time point; and the instruction generation module is used for generating a power control instruction of the cold-heat exchanger at the next time point based on the decoded value.

Fig. 1 is an application scenario diagram of a cold-heat exchanger for a pressure vessel according to an embodiment of the present application. As shown in fig. 1, in this application scenario. Fluid flow rate values at a plurality of predetermined time points within a predetermined period of time are acquired by a flow rate sensor (e.g., V as illustrated in fig. 1). Next, the above data is input to a server (e.g., S in fig. 1) where a control algorithm for the cold-heat exchanger of the pressure vessel is deployed, wherein the server is capable of processing the above input data with the control algorithm for the cold-heat exchanger of the pressure vessel to generate a decoded value, and generating a power control command for the cold-heat exchanger at a next point in time based on the decoded value.

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 a cold-heat exchanger for a pressure vessel according to an embodiment of the present application. As shown in fig. 2, the cold-heat exchanger 300 for a pressure vessel according to an embodiment of the present application includes: a flow rate acquisition module 310; a vector construction module 320; a relative flow rate calculation module 330; a flow rate variation feature extraction module 340; a multi-scale variation feature fusion module 350; the fusion optimization module 360; a decoding module 370; and an instruction generation module 380.

The flow rate acquisition module 310 is configured to acquire fluid flow rate values at a plurality of predetermined time points within a predetermined time period; the vector construction module 320 is configured to arrange the fluid flow velocity values at the plurality of predetermined time points into a flow velocity input vector according to a time dimension; the relative flow rate calculation module 330 is configured to calculate a difference between the fluid flow rate values of each two adjacent positions in the flow rate input vector to obtain a flow rate variation input vector; the flow velocity change feature extraction module 340 is configured to pass the flow velocity input vector and the flow velocity change input vector through a multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector; the multiscale variation feature fusion module 350 is configured to fuse the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector based on a gaussian density map to obtain a fusion feature matrix; the fusion optimization module 360 is configured to perform vector spectrum clustering agent learning fusion optimization on the fusion feature matrix to obtain an optimized fusion feature matrix; the decoding module 370 is configured to pass the optimized fusion feature matrix through a decoder to obtain a decoded value, where the decoded value is used to represent a fluid flow velocity value at a next time point; and an instruction generation module 380 for generating a power control instruction of the cold-heat exchanger at a next point in time based on the decoded value.

Fig. 3 is a system architecture diagram of a cold-heat exchanger for a pressure vessel according to an embodiment of the present application. As shown in fig. 3, in the network architecture, fluid flow rate values at a plurality of predetermined time points within a predetermined time period are first acquired by the flow rate acquisition module 310; next, the vector construction module 320 arranges the fluid flow velocity values at a plurality of predetermined time points acquired by the flow velocity acquisition module 310 into a flow velocity input vector according to a time dimension; the relative flow rate calculation module 330 calculates a difference between the fluid flow rate values of each adjacent two positions in the flow rate input vector obtained by the vector construction module 320 to obtain a flow rate variation input vector; the flow velocity change feature extraction module 340 respectively passes the flow velocity input vector obtained by the vector construction module 320 and the flow velocity change input vector obtained by the relative flow velocity calculation module 330 through a multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector; then, the multi-scale variation feature fusion module 350 fuses the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector based on a gaussian density map to obtain a fusion feature matrix; the fusion optimization module 360 performs vector spectrum clustering agent learning fusion optimization on the fusion feature matrix obtained by the multi-scale variation feature fusion module 350 to obtain an optimized fusion feature matrix; the decoding module 370 passes the optimized fusion feature matrix obtained by the multi-scale variation feature fusion module 360 through a decoder to obtain a decoded value, where the decoded value is used to represent a fluid flow velocity value at a next time point; further, the command generation module 380 generates a power control command for the heat and cold exchanger at the next time point based on the decoded value.

Specifically, during operation of the cold heat exchanger 300 for a pressure vessel, the flow rate acquisition module 310 is configured to acquire fluid flow rate values at a plurality of predetermined time points within a predetermined time period. It should be understood that, during the operation of the actual cold-heat exchanger, if the operation power of the cold-heat exchanger is to be accurately controlled in real time, so as to achieve the purpose of maintaining the fluid temperature within the predetermined fluctuation range, so as to avoid the fluid temperature exceeding the predetermined fluctuation range, it is necessary to evaluate and predict the fluid flow rate value at the next time point based on the change condition of the fluid flow rate in the time dimension, and accordingly control the power of the cold-heat exchanger. Thus, in one specific example of the present application, first, fluid flow rate values at a plurality of predetermined time points within a predetermined period of time may be acquired by a flow rate sensor.

Specifically, during operation of the cold heat exchanger 300 for a pressure vessel, the vector construction module 320 is configured to arrange the fluid flow velocity values at the plurality of predetermined time points into a flow velocity input vector according to a time dimension. In order to facilitate extraction of the dynamic change characteristic information of the fluid flow velocity values in the time dimension, the fluid flow velocity values at a plurality of preset time points are further arranged into flow velocity input vectors according to the time dimension, so that distribution information of the fluid flow velocity values in time sequence is integrated.

Specifically, during operation of the cold heat exchanger 300 for a pressure vessel, the relative flow rate calculation module 330 is configured to calculate a difference between fluid flow rate values at each adjacent two locations in the flow rate input vector to obtain a flow rate variation input vector. In the technical scheme of the application, the absolute change characteristic of the fluid flow velocity value in the time dimension is obtained by calculating the difference value between the fluid flow velocity values of every two adjacent positions in the flow velocity input vector.

Specifically, during the operation of the cold/heat exchanger 300 for a pressure vessel, the flow velocity variation feature extraction module 340 is configured to pass the flow velocity input vector and the flow velocity variation input vector through a multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity variation time sequence feature vector, respectively. The flow velocity input vector and the flow velocity change input vector are respectively subjected to feature mining through a multi-scale neighborhood feature extraction module, so that dynamic multi-scale neighborhood associated features of the relative quantity and the absolute quantity of the fluid flow velocity in different time spans in the preset time period are extracted, and a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector are obtained. Wherein, the multiscale neighborhood feature extraction module comprises: the device comprises a first convolution layer, a second convolution layer parallel to the first convolution layer and a multi-scale feature fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first length, and the second convolution layer uses a one-dimensional convolution kernel with a second length.

Fig. 4 is a block diagram of a flow rate variation feature extraction module in a cold-heat exchanger for a pressure vessel according to an embodiment of the application. As shown in fig. 4, the flow rate variation feature extraction module includes: a first neighborhood scale feature extraction unit 341, configured to input the flow velocity input vector and the flow velocity variation input vector into a first convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a first neighborhood scale flow velocity time sequence feature vector and a first neighborhood scale flow velocity variation time sequence feature vector, where the first convolution layer has a first one-dimensional convolution kernel with a first length; a second neighborhood scale feature extraction unit 342 for inputting the flow rateThe input vector and the flow velocity change input vector are respectively input into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second neighborhood scale flow velocity time sequence feature vector and a second neighborhood scale flow velocity change time sequence feature vector, wherein the second convolution layer is provided with a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; a kind of electronic device with a high-pressure air-conditioning system. The multi-scale cascade unit 343 is configured to cascade the first neighborhood scale flow velocity time sequence feature vector and the second neighborhood scale flow velocity time sequence feature vector to obtain the flow velocity time sequence feature vector, and cascade the first neighborhood scale flow velocity change time sequence feature vector and the second neighborhood scale flow velocity change time sequence feature vector to obtain the flow velocity change time sequence feature vector. Specifically, inputting the flow velocity input vector and the flow velocity variation input vector into a first convolution layer of the multi-scale neighborhood feature extraction module respectively to obtain a first neighborhood scale flow velocity time sequence feature vector and a first neighborhood scale flow velocity variation time sequence feature vector, including: using a first convolution layer of the multi-scale neighborhood feature extraction module to respectively perform one-dimensional convolution coding on the flow velocity input vector and the flow velocity change input vector according to the following formula so as to obtain a first neighborhood scale flow velocity time sequence feature vector and a first neighborhood scale flow velocity change time sequence feature vector; wherein, the formula is: Wherein (1)>For the first convolution kernel at->Width in direction, ++>For the first convolution kernel parameter vector, +.>For a local vector matrix operating with a convolution kernel function, < ->For the size of the first convolution kernel, +.>Representing the flow rate input vector and the flow rate variation input vector; more specifically, inputting the flow velocity input vector and the flow velocity variation input vector into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second neighborhood scale flow velocity time sequence feature vector and a second neighborhood scale flow velocity variation time sequence feature vector, respectively, including: using a second convolution layer of the multi-scale neighborhood feature extraction module to respectively perform one-dimensional convolution coding on the flow velocity input vector and the flow velocity change input vector according to the following formula so as to obtain a second neighborhood scale flow velocity time sequence feature vector and a second neighborhood scale flow velocity change time sequence feature vector; wherein, the formula is: />Wherein (1)>For the second convolution kernel>Width in direction, ++>For a second convolution kernel parameter vector, +.>For a local vector matrix operating with a convolution kernel function, < ->For the size of the second convolution kernel, +.>Representing the flow rate input vector and the flow rate change input vector.

In particular during operation of the cold and heat exchanger 300 for a pressure vesselThe multiscale variation feature fusion module 350 is configured to fuse the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector based on a gaussian density map to obtain a fusion feature matrix. Considering that the flow velocity time series feature vector and the flow velocity variation time series feature vector each correspond to a feature distribution manifold in a high-dimensional feature space, and these feature distribution manifolds are very easily trapped in local extremum points when finding an optimum point by gradient descent, so that global optimum points cannot be obtained if the time series dynamic variation feature of the fluid flow velocity is represented by concatenating only feature vectors of relative quantity data and absolute quantity data of the fluid flow velocity values. Therefore, it is necessary to further appropriately fuse the flow velocity timing feature vector and the flow velocity variation timing feature vector so that the respective feature distributions can converge on the profile with respect to each other. Also considering that gaussian density maps are widely used for prior-based estimation of target posterior in deep learning, they can be used to correct data distribution, thereby achieving the above objective. Specifically, in the technical scheme of the application, firstly, a flow velocity Gaussian density chart of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector is constructed, so that an absolute quantity dynamic multi-scale change feature and a relative quantity dynamic multi-scale change feature of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector in relation to the fluid flow velocity in a time dimension are fused; and then, carrying out Gaussian discretization processing on the flow velocity Gaussian density map so as not to generate information loss when the data characteristics are amplified, and further improving the accuracy of subsequent decoding, thereby obtaining a fusion characteristic matrix. In the technical scheme of the application, because the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector are fused based on the Gaussian density chart, the mean feature vector of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector is subjected to Gaussian discretization based on two-dimensional Gaussian probability distribution, so that the fusion feature matrix is not capable of And distinguishing the importance of the characteristic values of each position of the flow velocity time sequence characteristic vector and the flow velocity change time sequence characteristic vector, namely, not reflecting the confidence of each characteristic value of the characteristic vector, thereby influencing the expression effect of the fusion characteristic matrix. Based on this, in the technical scheme of the application, the fusion feature matrix is first passed through a convolutional neural network model with the same channel number as the length of the flow velocity time sequence feature vector to obtain a weighted feature mapThe weighted feature map is further +.>Modeling of a feature correlation cumulative discrimination mechanism to obtain a weighted feature vector +.>，/>Expressed as: />Wherein->Representing the weighted feature map, ">Andrespectively representing the single-layer convolution operation based on different convolution kernels on the feature map,/and>representation->Activating function->Representation->Activate function, and->Representing global pooling of each feature matrix of the feature map,/for each feature matrix>Representing addition by position +.>Representing the weighted feature vector. Here, the feature correlation accumulating and distinguishing mechanism modeling firstly generates two new local association units of the weighted feature map through convolution operation, then uses Sigmoid function and ReLU function to perform simple embedding, resetting and updating similar to a neural network architecture on the local association features, and then accumulates the correlation of the local features relative to the whole features through global average pooling operation, so that the feature importance sequence is explicitly modeled by using the feature distinguishing mechanism, and then the proper weighting factors in the channel dimension can be determined based on the feature accumulating and distinguishing mechanism of each feature matrix of the weighted feature map. Then, the weighted feature vector is added again >And respectively carrying out dot multiplication with the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector to distinguish the importance of the feature values of each position of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector based on the importance degree of the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector in the associated feature domain, thereby improving the expression effect of the fusion feature matrix obtained after fusion. Therefore, the fluid flow velocity value at the next time point can be accurately estimated and predicted based on the actual fluid flow velocity change condition, and the power of the cold-heat exchanger at the next time point is accurately controlled, so that the temperature of the fluid is maintained within a preset fluctuation range, and the normal and safe use of the pressure vessel is ensured.

FIG. 5 is a schematic illustration of an embodiment of the applicationA block diagram of a multi-scale varying feature fusion module in a cold-heat exchanger for a pressure vessel. As shown in fig. 5, the multi-scale variant feature fusion module 350 includes: a gaussian fusion unit 351, configured to fuse the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector by using the Gao Simi degree graph to obtain an initial fused gaussian density graph, where a mean vector of the initial fused gaussian density graph is a mean vector between the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector, and a value of each position in a covariance matrix of the initial fused gaussian density graph is a variance between feature values of corresponding two positions in the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector; the gaussian discretization unit 352 is configured to perform gaussian discretization on the gaussian distribution of each position in the initial fusion gaussian density map to obtain an initial fusion feature matrix; a weighted feature extraction unit 353, configured to pass the initial fusion feature matrix through a convolutional neural network model with the same channel number and the same length of the flow velocity time sequence feature vector, so as to obtain a weighted feature map; a differentiating unit 354 for modeling a feature correlation cumulative differentiating mechanism of the weighted feature map to obtain a weighted feature vector; a weighted optimization unit 355, configured to perform point multiplication on the weighted feature vector and the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector to obtain an optimized flow velocity time sequence feature vector and an optimized flow velocity change time sequence feature vector; and a fusion unit 356, configured to perform association encoding on the optimized flow velocity time sequence feature vector and the optimized flow velocity variation time sequence feature vector to obtain the fusion feature matrix. Wherein, the weighted feature extraction unit is used for: respectively carrying out point multiplication on the weighted feature vector and the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector by using the following formula to obtain an optimized flow velocity time sequence feature vector and an optimized flow velocity change time sequence feature vector; wherein, the formula is: Wherein->Representing the weighted feature vector,/->Representing the flow rate timing characteristic vector and the flow rate variation timing characteristic vector,/or->Representing the optimized flow rate timing feature vector and the optimized flow rate variation timing feature vector, +.>Representing multiplication by location. More specifically, the fusion unit is configured to: performing association coding on the optimized flow velocity time sequence feature vector and the optimized flow velocity change time sequence feature vector by using the following formula to obtain the fusion feature matrix; wherein, the formula is: />Wherein->Representing the optimized flow rate timing feature vector, +.>Transpose of the optimized flow timing feature vector, < >>Time sequence characteristic vector representing the optimized flow rate change, < >>Representing the fusion feature matrix,>representing vector multiplication.

Specifically, during the operation of the cold/heat exchanger 300 for a pressure vessel, the fusion optimization module 360 is configured to perform vector spectral clustering agent learning fusion optimization on the fusion feature matrix to obtain an optimized fusion feature matrix. It should be appreciated that due to the high basisWhen the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector are fused by the Gaussian density diagram, each feature value of the flow velocity time sequence feature vector and the mean feature vector of the flow velocity change time sequence feature vector is subjected to Gaussian discretization based on two-dimensional Gaussian probability distribution on the basis of corresponding rows and columns of the flow velocity time sequence feature vector and the variance matrix of the flow velocity change time sequence feature vector, so that the fused feature matrix can be regarded as a feature matrix obtained by splicing the feature vectors of each row or column. Therefore, when classifying the fusion feature matrix, the integration of the feature distribution of the fusion feature matrix is expected to be improved, so that the classification effect of the fusion feature matrix is improved. Based on this, the applicant of the present application refers to the fusion feature matrix, e.g. denoted as And carrying out vector spectrum clustering agent learning fusion optimization.

Specifically, the fusion optimization module is used for: vector spectral clustering agent learning fusion optimization is carried out on the fusion feature matrix according to the following formula so as to obtain the optimized fusion feature matrix; wherein, the formula is:wherein (1)>Is the fusion feature matrix,>is the optimized fusion feature matrix, +.>Representing individual row feature vectors of the fusion feature matrix, and +.>Is a distance matrix composed of the distances between every two corresponding row feature vectors of the fusion feature matrix,/and>an exponential operation representing a matrix representing a natural exponential function value raised to a power by a characteristic value of each position in the matrix, ">And->Respectively representing dot-by-location multiplication and matrix addition.

Specifically, during operation of the cold heat exchanger 300 for a pressure vessel, the decoding module 370 and the instruction generating module 380 are configured to pass the optimized fusion feature matrix through a decoder to obtain decoded values, where the decoded values are used to represent fluid flow rate values at a next point in time; and generating a power control instruction of the cold-heat exchanger at the next time point based on the decoded value. That is, the optimized fusion feature matrix is passed through a decoder to obtain a decoded value, in particular, the optimized fusion feature matrix is passed through a decoder using the decoder in the following formula to obtain a decoded valueA value representing a fluid flow rate value at a next point in time; wherein, the formula is:wherein->Representing the optimized fusion feature matrix, +.>Is the decoded value,/->Is a weight matrix, < >>Representing matrix multiplication. That is, a decoding regression is performed based on time-series dynamic change characteristic information of relative and absolute amounts of the fluid flow rate values, thereby accurately evaluating predictions for fluid flow rate values at the next time point based on change characteristics of the fluid flow rate values in the time dimension. Then, based on the decoded value, a power control command of the cold-heat exchanger at the next time point is generated, so that the power of the cold-heat exchanger at the next time point is accurately controlled, and the fluid temperature is maintained within a predetermined fluctuation range. Accordingly, in one specific example of the present application, in performing the power control of the heat and cold exchanger, if the decoded value is greater than the fluid flow rate value at the current time point, the power control command of the heat and cold exchanger at the next time point is to increase the power of the heat and cold exchanger, and if the decoded value is less than the fluid flow rate value at the current time point, the power control command of the heat and cold exchanger at the next time point is to decrease the power of the heat and cold exchanger.

In summary, the cold-hot exchanger 300 for a pressure vessel according to the embodiment of the application is illustrated, which uses a neural network model based on deep learning to mine the dynamic variation characteristic information of the fluid flow velocity value in the time dimension, so as to accurately evaluate and predict the fluid flow velocity value at the next time point, and further accurately control the power of the cold-hot exchanger at the next time point, so as to maintain the fluid temperature within the predetermined fluctuation range, and ensure the normal and safe use of the pressure vessel.

As described above, the heat and cold exchanger for a pressure vessel according to the embodiment of the present application can be implemented in various terminal devices. In one example, the cold heat exchanger 300 for a pressure vessel according to an embodiment of the present application may be integrated into the terminal device as one software module and/or hardware module. For example, the cold heat exchanger 300 for the pressure vessel may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the cold heat exchanger 300 for the pressure vessel can equally be one of the plurality of hardware modules of the terminal device.

Alternatively, in another example, the cold heat exchanger 300 for a pressure vessel and the terminal device may be separate devices, and the cold heat exchanger 300 for a pressure vessel may be connected to the terminal device through a wired and/or wireless network and transmit interactive information in a agreed data format.

An exemplary method is: fig. 6 is a flowchart of a control method of a cold heat exchanger for a pressure vessel according to an embodiment of the present application. As shown in fig. 6, a control method of a cold-heat exchanger for a pressure vessel according to an embodiment of the present application includes the steps of: s110, acquiring fluid flow velocity values at a plurality of preset time points in a preset time period; s120, arranging the fluid flow velocity values of the plurality of preset time points into flow velocity input vectors according to a time dimension; s130, calculating the difference value between the fluid flow velocity values of every two adjacent positions in the flow velocity input vector to obtain a flow velocity change input vector; s140, respectively passing the flow velocity input vector and the flow velocity change input vector through a multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector; s150, fusing the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector based on a Gaussian density chart to obtain a fusion feature matrix; s160, the optimized fusion feature matrix passes through a decoder to obtain a decoding value, wherein the decoding value is used for representing a fluid flow velocity value at the next time point; s170, the optimized fusion feature matrix is passed through a decoder to obtain a decoding value, wherein the decoding value is used for representing a fluid flow velocity value at the next time point; and S180, generating a power control instruction of the cold-heat exchanger at the next time point based on the decoded value.

In one example, in the above control method for a cold-heat exchanger of a pressure vessel, the step S140 includes: respectively inputting the flow velocity input vector and the flow velocity change input vector into a first convolution layer of the multi-scale neighborhood feature extraction module to obtain a first neighborhood scale flow velocity time sequence feature vector and a first neighborhood scale flow velocity change time sequence feature vector, wherein the first convolution layer is provided with a first one-dimensional convolution kernel with a first length; respectively inputting the flow velocity input vector and the flow velocity change input vector into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second neighborhood scale flow velocity time sequence feature vector and a second neighborhood scale flow velocity change time sequence feature vector, wherein the second convolution layer is provided with a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and concatenating the first neighborhood scale flow velocity timing feature vector and the second neighborhood scale flow velocity timing feature vector to obtain the flow velocity timing feature vector, and concatenating the first neighborhood scale flow velocity variation timing feature vector and the second neighborhood scale flow velocity variation timing feature vector to obtain the flow velocity variation timing feature vector. Wherein, the multiscale neighborhood feature extraction module comprises: the device comprises a first convolution layer, a second convolution layer parallel to the first convolution layer and a multi-scale feature fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first length, and the second convolution layer uses a one-dimensional convolution kernel with a second length.

In one example, in the above control method for a cold-heat exchanger of a pressure vessel, the step S150 includes: usingThe Gaussian density map is used for fusing the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector to obtain an initial fused Gaussian density map, wherein the mean value vector of the initial fused Gaussian density map is the mean value vector between the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector, and the value of each position in the covariance matrix of the initial fused Gaussian density map is the variance between the characteristic values of the corresponding two positions in the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector; performing Gaussian discretization on the Gaussian distribution of each position in the initial fusion Gaussian density map to obtain an initial fusion feature matrix; the initial fusion feature matrix passes through a convolutional neural network model with the same channel number as the length of the flow velocity time sequence feature vector so as to obtain a weighted feature diagram; modeling the weighted feature map by a feature correlation accumulation and differentiation mechanism to obtain a weighted feature vector; respectively carrying out point multiplication on the weighted feature vector and the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector to obtain an optimized flow velocity time sequence feature vector and an optimized flow velocity change time sequence feature vector; and performing association coding on the optimized flow velocity time sequence feature vector and the optimized flow velocity change time sequence feature vector to obtain the fusion feature matrix. The method for performing point multiplication on the weighted feature vector and the flow velocity time sequence feature vector to obtain an optimized flow velocity time sequence feature vector and an optimized flow velocity change time sequence feature vector respectively comprises the following steps: respectively carrying out point multiplication on the weighted feature vector and the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector by using the following formula to obtain an optimized flow velocity time sequence feature vector and an optimized flow velocity change time sequence feature vector; wherein, the formula is: Wherein->Representing the weighted feature vector,/->Representing the flow rate timing characteristic vector and the flow rate variation timing characteristic vector,/or->Representing the optimized flow rate timing feature vector and the optimized flow rate variation timing feature vector, +.>Representing multiplication by location. More specifically, performing association coding on the optimized flow velocity time sequence feature vector and the optimized flow velocity change time sequence feature vector to obtain the fusion feature matrix, including: performing association coding on the optimized flow velocity time sequence feature vector and the optimized flow velocity change time sequence feature vector by using the following formula to obtain the fusion feature matrix; wherein, the formula is: />Wherein->Representing the optimized flow rate timing feature vector, +.>Transpose of the optimized flow timing feature vector, < >>Time sequence characteristic vector representing the optimized flow rate change, < >>Representing the fusion feature matrix,>representing vector multiplication. More specifically, modeling the weighted feature map for a feature correlation cumulative discrimination mechanism to obtain a weighted feature vector includes: modeling the weighted feature map by a feature correlation accumulation differentiation mechanism according to the following formula to obtain the weighted feature vector; wherein, the formula is: Wherein->Representing the weighted feature map, ">And->Respectively representing the single-layer convolution operation based on different convolution kernels on the feature map,/and>representation->The function is activated and the function is activated,representation->Activate function, and->Representing global pooling of each feature matrix of the feature map,/for each feature matrix>Representing addition by position +.>Representing the weighted feature vector.

In one example, in the above control method for a cold-heat exchanger of a pressure vessel, the step S160 includes: vector spectral clustering agent learning fusion optimization is carried out on the fusion feature matrix according to the following formula so as to obtain the optimized fusion feature matrix; wherein, the formula is:wherein (1)>Is the fusion feature matrix,>is the optimized fusion feature matrix, +.>Representing individual row feature vectors of the fusion feature matrix, and +.>Is a distance matrix composed of distances between every two corresponding row feature vectors of the fusion feature matrix,an exponential operation representing a matrix representing a natural exponential function value raised to a power by a characteristic value of each position in the matrix, ">And->Respectively representing dot-by-location multiplication and matrix addition.

In one example, in the above control method for a cold-heat exchanger of a pressure vessel, the step S170 includes: passing the optimized fusion feature matrix through a decoder using the decoder to obtain a decoded value representing a fluid flow rate value at a next point in time; wherein, the formula is: Wherein->Representing the optimized fusion feature matrix, +.>Is the decoded value,/->Is a weight matrix, < >>Representing matrix multiplication.

In summary, the control method for the cold and heat exchanger of the pressure vessel according to the embodiment of the application is explained, which adopts the neural network model based on deep learning to mine the dynamic change characteristic information of the fluid flow velocity value in the time dimension, so as to accurately evaluate and predict the fluid flow velocity value at the next time point, and further accurately control the power of the cold and heat exchanger at the next time point, so as to maintain the fluid temperature within the preset fluctuation range, and ensure the normal and safe use of the pressure vessel.

Exemplary electronic device: next, an electronic device according to an embodiment of the present application is described with reference to fig. 7.

Fig. 7 illustrates a block diagram of an electronic device according to an embodiment of the application.

As shown in fig. 7, 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 cold heat exchanger for a pressure vessel and/or other desired functions of the various embodiments of the present application described above. Various content such as a fusion feature matrix 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 that are relevant to the present application are shown in fig. 7 for simplicity, components such as buses, input/output interfaces, etc. are 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 application may also be a computer program product comprising computer program instructions which, when executed by a processor, cause the processor to perform steps in the methods of controlling a cold heat exchanger for a pressure vessel according to the various embodiments of the application described in the "exemplary systems" section of this specification.

The computer program product may write program code for performing 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, on which computer program instructions are stored, which, when being executed by a processor, cause the processor to perform the steps in the functions in the control method for a cold-heat exchanger of a pressure vessel according to the various embodiments of the present application described in the "exemplary systems" section above of the present specification.

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 intended to be limiting, and these advantages, benefits, effects, etc. are not to be considered as essential to 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 necessarily limited to practice with the above described specific details.

The block diagrams of the devices, apparatuses, devices, systems referred to in the present 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 aspects of the present application.

The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit 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 cold-heat exchanger for a pressure vessel, comprising: the flow rate acquisition module is used for acquiring fluid flow rate values of a plurality of preset time points in a preset time period; a vector construction module for arranging the fluid flow velocity values of the plurality of preset time points into a flow velocity input vector according to a time dimension; the relative flow velocity calculation module is used for calculating the difference value between the fluid flow velocity values of every two adjacent positions in the flow velocity input vector to obtain a flow velocity change input vector; the flow velocity change feature extraction module is used for respectively passing the flow velocity input vector and the flow velocity change input vector through the multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector; the multi-scale change feature fusion module is used for fusing the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector based on a Gaussian density chart to obtain a fusion feature matrix; the fusion optimization module is used for carrying out vector spectrum clustering agent learning fusion optimization on the fusion feature matrix to obtain an optimized fusion feature matrix; the decoding module is used for enabling the optimized fusion characteristic matrix to pass through a decoder to obtain a decoding value, wherein the decoding value is used for representing a fluid flow velocity value at the next time point; and the instruction generation module is used for generating a power control instruction of the cold-heat exchanger at the next time point based on the decoded value.

2. The cold-heat exchanger for a pressure vessel of claim 1, wherein the multi-scale neighborhood feature extraction module comprises: the device comprises a first convolution layer, a second convolution layer parallel to the first convolution layer and a multi-scale feature fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first length, and the second convolution layer uses a one-dimensional convolution kernel with a second length.

3. The cold-heat exchanger for a pressure vessel according to claim 2, wherein the flow rate variation feature extraction module comprises: the first neighborhood scale feature extraction unit is used for inputting the flow velocity input vector and the flow velocity change input vector into a first convolution layer of the multi-scale neighborhood feature extraction module respectively to obtain a first neighborhood scale flow velocity time sequence feature vector and a first neighborhood scale flow velocity change time sequence feature vector, wherein the first convolution layer is provided with a first one-dimensional convolution kernel with a first length; a second neighborhood scale feature extraction unit, configured to input the flow velocity input vector and the flow velocity variation input vector into a second convolution layer of the multi-scale neighborhood feature extraction module, respectively, to obtain a second neighborhood scale flow velocity time sequence feature vector and a second neighborhood scale flow velocity variation time sequence feature vector, where the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and the multi-scale cascading unit is used for cascading the first neighborhood scale flow velocity time sequence feature vector and the second neighborhood scale flow velocity time sequence feature vector to obtain the flow velocity time sequence feature vector, and cascading the first neighborhood scale flow velocity change time sequence feature vector and the second neighborhood scale flow velocity change time sequence feature vector to obtain the flow velocity change time sequence feature vector.

4. A cold-heat exchanger for a pressure vessel according to claim 3, wherein the multiscale varying feature fusion module comprises: a gaussian fusion unit, configured to fuse the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector by using the Gao Simi degree map to obtain an initial fused gaussian density map, where a mean vector of the initial fused gaussian density map is a mean vector between the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector, and a value of each position in a covariance matrix of the initial fused gaussian density map is a variance between feature values of corresponding two positions in the flow velocity time sequence feature vector and the flow velocity variation time sequence feature vector; the Gaussian discretization unit is used for carrying out Gaussian discretization on the Gaussian distribution of each position in the initial fusion Gaussian density map so as to obtain an initial fusion feature matrix; the weighted feature extraction unit is used for enabling the initial fusion feature matrix to pass through a convolutional neural network model with the same channel number as the length of the flow velocity time sequence feature vector so as to obtain a weighted feature map; the distinguishing unit is used for modeling the weighted feature map by a feature correlation accumulation distinguishing mechanism so as to obtain a weighted feature vector; the weighting optimization unit is used for performing point multiplication on the weighting feature vector and the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector respectively to obtain an optimized flow velocity time sequence feature vector and an optimized flow velocity change time sequence feature vector; and the fusion unit is used for carrying out association coding on the optimized flow velocity time sequence feature vector and the optimized flow velocity change time sequence feature vector so as to obtain the fusion feature matrix.

5. The cold-heat exchanger for a pressure vessel according to claim 4, wherein the weighted feature extraction unit is configured to: respectively carrying out point multiplication on the weighted feature vector and the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector by using the following formula to obtain an optimized flow velocity time sequence feature vector and an optimized flow velocity change time sequence feature vector; wherein, the formula is:wherein->Representing the weighted feature vector,/->Representing the flow rate timing characteristic vector and the flow rate variation timing characteristic vector,/or->Representing the optimized flow rate timing feature vector and the optimized flow rate variation timing feature vector, +.>Representing multiplication by location.

6. The heat and cold exchanger for a pressure vessel of claim 5, wherein the differentiating unit is further configured to: modeling the weighted feature map by a feature correlation accumulation differentiation mechanism according to the following formula to obtain the weighted feature vector; wherein, the formula is:wherein->Representing the weighted feature map, ">And->Respectively representing a single-layer convolution operation based on different convolution kernels on the feature map,representation->Activating function- >Representation->Activate function, and->Representing global pooling of each feature matrix of the feature map,/for each feature matrix>Representing addition by position +.>Representing the weighted feature vector.

7. The heat and cold exchanger for a pressure vessel according to claim 6, wherein the fusion unit is configured to: performing association coding on the optimized flow velocity time sequence feature vector and the optimized flow velocity change time sequence feature vector by using the following formula to obtain the fusion feature matrix; wherein, the formula is:wherein->Representing the saidOptimizing flow timing feature vector,/->Transpose of the optimized flow timing feature vector, < >>Time sequence characteristic vector representing the optimized flow rate change, < >>Representing the fusion feature matrix,>representing vector multiplication.

8. The cold-heat exchanger for a pressure vessel of claim 7, wherein the fusion optimization module is configured to: vector spectral clustering agent learning fusion optimization is carried out on the fusion feature matrix according to the following formula so as to obtain the optimized fusion feature matrix; wherein, the formula is:wherein (1)>Is the fusion feature matrix,>is the optimized fusion feature matrix, +. >Representing individual row feature vectors of the fusion feature matrix, and +.>Is a distance matrix composed of distances between every two corresponding row feature vectors of the fusion feature matrix,an exponential operation representing a matrix representing a natural exponential function value raised to a power by a characteristic value of each position in the matrix, ">And->Respectively representing dot-by-location multiplication and matrix addition.

9. The cold-heat exchanger for a pressure vessel of claim 8, wherein the decoding module is configured to: passing the optimized fusion feature matrix through a decoder using the decoder to obtain a decoded value representing a fluid flow rate value at a next point in time; wherein, the formula is:wherein->Representing the optimized fusion feature matrix, +.>Is the decoded value,/->Is a weight matrix, < >>Representing matrix multiplication.

10. A control method for a cold-heat exchanger of a pressure vessel, comprising: acquiring fluid flow rate values at a plurality of preset time points in a preset time period; arranging the fluid flow velocity values of the plurality of preset time points into flow velocity input vectors according to a time dimension; calculating the difference between the fluid flow velocity values of every two adjacent positions in the flow velocity input vector to obtain a flow velocity change input vector; respectively passing the flow velocity input vector and the flow velocity change input vector through a multi-scale neighborhood feature extraction module to obtain a flow velocity time sequence feature vector and a flow velocity change time sequence feature vector; fusing the flow velocity time sequence feature vector and the flow velocity change time sequence feature vector based on a Gaussian density chart to obtain a fusion feature matrix; vector spectrum clustering agent learning fusion optimization is carried out on the fusion feature matrix so as to obtain an optimized fusion feature matrix; passing the optimized fusion feature matrix through a decoder to obtain a decoded value, wherein the decoded value is used for representing a fluid flow velocity value at a next time point; and generating a power control instruction of the cold-heat exchanger at the next time point based on the decoded value.