CN116835540B - Preparation method of phosphorus pentafluoride - Google Patents
Preparation method of phosphorus pentafluoride Download PDFInfo
- Publication number
- CN116835540B CN116835540B CN202310479969.0A CN202310479969A CN116835540B CN 116835540 B CN116835540 B CN 116835540B CN 202310479969 A CN202310479969 A CN 202310479969A CN 116835540 B CN116835540 B CN 116835540B
- Authority
- CN
- China
- Prior art keywords
- feature
- matrix
- contact temperature
- vector
- phosphorus pentafluoride
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- OBCUTHMOOONNBS-UHFFFAOYSA-N phosphorus pentafluoride Chemical compound FP(F)(F)(F)F OBCUTHMOOONNBS-UHFFFAOYSA-N 0.000 title claims abstract description 132
- 238000002360 preparation method Methods 0.000 title abstract description 35
- 238000006243 chemical reaction Methods 0.000 claims abstract description 72
- 238000000034 method Methods 0.000 claims abstract description 58
- 239000012535 impurity Substances 0.000 claims abstract description 29
- 238000003062 neural network model Methods 0.000 claims abstract description 19
- -1 hexafluorophosphate Chemical compound 0.000 claims abstract description 18
- 229910001512 metal fluoride Inorganic materials 0.000 claims abstract description 13
- 238000005979 thermal decomposition reaction Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 199
- 239000013598 vector Substances 0.000 claims description 175
- 238000013527 convolutional neural network Methods 0.000 claims description 34
- 230000004043 responsiveness Effects 0.000 claims description 33
- 238000012545 processing Methods 0.000 claims description 31
- 238000011176 pooling Methods 0.000 claims description 24
- 238000000605 extraction Methods 0.000 claims description 23
- 230000004044 response Effects 0.000 claims description 21
- 230000004913 activation Effects 0.000 claims description 18
- 230000003247 decreasing effect Effects 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 16
- 238000012546 transfer Methods 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 37
- 230000008859 change Effects 0.000 abstract description 16
- 238000013507 mapping Methods 0.000 abstract description 6
- 238000007711 solidification Methods 0.000 abstract 1
- 230000008023 solidification Effects 0.000 abstract 1
- 238000009826 distribution Methods 0.000 description 14
- 239000000047 product Substances 0.000 description 9
- 230000000694 effects Effects 0.000 description 8
- 239000002994 raw material Substances 0.000 description 8
- 239000006227 byproduct Substances 0.000 description 6
- 238000004519 manufacturing process Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000005065 mining Methods 0.000 description 4
- 239000000203 mixture Substances 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000009835 boiling Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000000746 purification Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000005094 computer simulation Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 238000004378 air conditioning Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
Classifications
-
- C—CHEMISTRY; METALLURGY
- C01—INORGANIC CHEMISTRY
- C01B—NON-METALLIC ELEMENTS; COMPOUNDS THEREOF; METALLOIDS OR COMPOUNDS THEREOF NOT COVERED BY SUBCLASS C01C
- C01B25/00—Phosphorus; Compounds thereof
- C01B25/10—Halides or oxyhalides of phosphorus
-
- C—CHEMISTRY; METALLURGY
- C01—INORGANIC CHEMISTRY
- C01P—INDEXING SCHEME RELATING TO STRUCTURAL AND PHYSICAL ASPECTS OF SOLID INORGANIC COMPOUNDS
- C01P2006/00—Physical properties of inorganic compounds
- C01P2006/80—Compositional purity
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Landscapes
- Chemical & Material Sciences (AREA)
- Organic Chemistry (AREA)
- Inorganic Chemistry (AREA)
- Feedback Control In General (AREA)
Abstract
The application relates to the technical field of intelligent preparation, and particularly discloses a preparation method of phosphorus pentafluoride, which comprises the following steps: passing phosphorus pentafluoride comprising gaseous impurities through a metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate; and, subjecting the hexafluorophosphate to thermal decomposition to obtain purified phosphorus pentafluoride; in particular, in the process of solidifying phosphorus pentafluoride containing gas impurities in the form of hexafluorophosphate by metal fluoride, a mapping relation between time sequence cooperative change of contact temperature and reaction pressure and time sequence change of flow rate value of phosphorus pentafluoride is established through a deep neural network model, so that the flow rate value of phosphorus pentafluoride is adaptively adjusted accurately in real time based on the time sequence cooperative change condition of actual contact temperature value and reaction pressure value, and the efficiency of a solidification process and the preparation quality of phosphorus pentafluoride are optimized.
Description
Technical Field
The application relates to the technical field of intelligent preparation, and in particular relates to a preparation method of phosphorus pentafluoride.
Background
Phosphorus pentafluoride (phosphorus pentafluoride) has wide application in the semiconductor field and the battery field, and in the preparation process of phosphorus pentafluoride, the mixture in the raw materials of the product can have adverse effects on the performance of the product and the safety of the preparation process, so that in the actual preparation process of phosphorus pentafluoride, the quality and purity requirements on the raw materials of the product are high.
However, in the conventional scheme, impurities such as raw materials, byproducts of the manufacturing method and the like, which are derived from the phosphorus pentafluoride, are generated in the process of preparing the phosphorus pentafluoride, so that the purity of the prepared phosphorus pentafluoride is low. However, when phosphorus pentafluoride is purified by a purification method using a common gas such as a difference in boiling point, it is not easy to obtain phosphorus pentafluoride of high purity by removing a mixture of byproducts contained in a raw material or derived from a reaction by using a simple apparatus.
Thus, an optimized phosphorus pentafluoride preparation scheme is desired.
Disclosure of Invention
The application provides a preparation method of phosphorus pentafluoride, which can adaptively adjust the flowing speed value of the phosphorus pentafluoride based on the time sequence cooperative change condition of an actual contact temperature value and a reaction pressure value in real time so as to optimize the efficiency of a curing process and the preparation quality of the phosphorus pentafluoride.
In a first aspect, there is provided a process for the preparation of phosphorus pentafluoride, the process comprising: passing phosphorus pentafluoride comprising gaseous impurities through a metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate; a kind of electronic device with a high-pressure air-conditioning system. The hexafluorophosphate is thermally decomposed to obtain purified phosphorus pentafluoride.
With reference to the first aspect, in one implementation of the first aspect, passing phosphorus pentafluoride containing gas impurities through a metal fluoride to solidify the phosphorus pentafluoride in the form of hexafluorophosphate includes obtaining a flow rate value of the phosphorus pentafluoride containing gas impurities at a plurality of predetermined time points over a predetermined period of time, a contact temperature value of the plurality of predetermined time points, and a reaction pressure value of the plurality of predetermined time points; after the contact temperature values at the preset time points and the reaction pressure values at the preset time points are respectively arranged into a contact temperature input vector and a reaction pressure input vector according to the time dimension, carrying out association coding on the contact temperature input vector and the reaction pressure input vector to obtain a contact temperature-pressure association matrix; the contact temperature-pressure correlation matrix is passed through a convolutional neural network model comprising a block structural feature extraction module to obtain a contact temperature-pressure correlation feature matrix; arranging the inlet flow velocity values of the plurality of preset time points into flow velocity time sequence input vectors according to time dimensions, and then obtaining flow velocity time sequence feature vectors through a one-dimensional convolutional neural network model; multiplying the contact temperature-pressure correlation feature matrix and the flow velocity time sequence feature vector by a matrix to obtain a classification feature vector; based on the classification feature vector, performing feature expression constraint on the contact temperature-pressure correlation feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix; multiplying the optimized contact temperature-pressure correlation feature matrix and the flow velocity time sequence feature vector by a matrix to obtain an optimized classification feature vector; and passing the optimized classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow rate value of the current time point is increased or decreased.
With reference to the first aspect, in an implementation manner of the first aspect, the step of passing the contact temperature-pressure correlation matrix through a convolutional neural network model including a block structural feature extraction module to obtain a contact temperature-pressure correlation feature matrix includes: each layer of the convolutional neural network model comprising the block structural feature extraction module respectively carries out input data in forward transfer of the layer: performing convolution processing, pooling processing and nonlinear processing on the input data based on a first two-dimensional convolution kernel to obtain a first activation feature map; performing convolution processing, pooling processing and nonlinear activation processing based on a second two-dimensional convolution kernel on the first activation feature map to obtain a second activation feature map, wherein the first two-dimensional convolution kernel and the second two-dimensional convolution kernel are transposed with each other; the input of the first layer of the convolution neural network model containing the block structure feature extraction module is the contact temperature-pressure correlation matrix, and the output of the last layer of the convolution neural network model containing the block structure feature extraction module is the contact temperature-pressure correlation matrix.
With reference to the first aspect, in an implementation manner of the first aspect, the arranging, by a time dimension, the flow velocity values of the plurality of predetermined time points into flow velocity time sequence input vectors, and then obtaining flow velocity time sequence feature vectors by using a one-dimensional convolutional neural network model includes: each layer of the one-dimensional convolutional neural network model is used for respectively carrying out forward transfer on input data in the layers: performing convolution processing based on a one-dimensional convolution kernel on the input data to obtain a convolution feature map; pooling processing is carried out on the convolution feature images based on feature matrixes to obtain pooled feature images; non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map; the input of the first layer of the one-dimensional convolutional neural network model is the flow velocity time sequence input vector, and the output of the last layer of the one-dimensional convolutional neural network model is the flow velocity time sequence feature vector.
With reference to the first aspect, in an implementation manner of the first aspect, based on the classification feature vector, performing feature expression constraint on the contact temperature-pressure correlation feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix, including: calculating the responsiveness estimation of the classification feature vector relative to the flow velocity time sequence feature vector to obtain a responsiveness estimation feature matrix; and performing convolution dictionary contrast response learning on the contact temperature-pressure correlation characteristic matrix based on the response estimation characteristic matrix to obtain an optimized contact temperature-pressure correlation characteristic matrix.
With reference to the first aspect, in an implementation manner of the first aspect, calculating a responsiveness estimate of the classification feature vector relative to the flow velocity timing feature vector to obtain a responsiveness estimate feature matrix includes: further used for: calculating a responsiveness estimate of the classification feature vector relative to the flow timing feature vector to obtain the responsiveness estimate feature matrix with the following formula; wherein, the formula is: Wherein/> Representing the classification feature vector,/>Representing the flow velocity time sequence characteristic vector,/>Representing matrix multiplication,/>Representing the responsiveness estimation feature matrix.
With reference to the first aspect, in an implementation manner of the first aspect, performing convolutional dictionary contrast response learning on the contact temperature-pressure correlation feature matrix based on the responsiveness estimation feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix includes:
Performing convolution dictionary contrast response learning on the contact temperature-pressure correlation feature matrix according to the response estimation feature matrix by using the following optimization formula so as to obtain the optimized contact temperature-pressure correlation feature matrix;
wherein, the formula is: Wherein/> And/>Respectively the responsiveness estimation feature matrix and the contact temperature-pressure correlation feature matrix,/>Frobenius norms,/>, representing matricesRepresenting matrix subtraction,/>Representing matrix multiplication,/>Representing the optimized contact temperature-pressure correlation characteristic matrix.
With reference to the first aspect, in an implementation manner of the first aspect, multiplying the optimized contact temperature-pressure correlation feature matrix with the flow velocity time sequence feature vector to obtain an optimized classification feature vector includes: calculating the optimized contact temperature-pressure correlation characteristic matrix and the flow velocity time sequence characteristic vector by using the following formula, and multiplying the optimized contact temperature-pressure correlation characteristic matrix and the flow velocity time sequence characteristic vector by using the matrix to obtain an optimized classification characteristic vector; wherein, the formula is: Wherein/> Representing the flow velocity time sequence characteristic vector,/>Representing the optimized classification feature vector,/>Representing the optimized contact temperature-pressure correlation characteristic matrix,/>Representing a matrix multiplication.
With reference to the first aspect, in an implementation manner of the first aspect, the optimizing the classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate that the value of the flow rate of the current time point should be increased or should be decreased, and includes: processing the transfer vector using the classifier in the following formula to obtain the classification result; wherein, the formula is: wherein/> To/>Is a weight matrix,/>To/>Is bias vector,/>For transfer vector,/>Representing a normalized exponential function.
According to the preparation method of phosphorus pentafluoride, provided by the application, the flowing-in flow velocity value of phosphorus pentafluoride can be adaptively adjusted in real time and accurately based on the time sequence cooperative change condition of the actual contact temperature value and the reaction pressure value, so that the efficiency of a curing process and the preparation quality of phosphorus pentafluoride are optimized.
Drawings
Fig. 1 is an application scenario diagram of a preparation method of phosphorus pentafluoride according to an embodiment of the application.
Fig. 2 is a schematic flow chart of a method for producing phosphorus pentafluoride according to an embodiment of the application.
Fig. 3 is a schematic flow chart of a method for preparing phosphorus pentafluoride according to an embodiment of the application, in which phosphorus pentafluoride containing gas impurities is passed through metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate.
Fig. 4 is a schematic diagram of a model architecture of passing phosphorus pentafluoride containing gas impurities through a metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate in the preparation method of phosphorus pentafluoride according to the embodiment of the application.
Detailed Description
The technical scheme of the application will be described below with reference to the accompanying drawings.
Because of the deep learning-based deep neural network model, related terms and concepts of the deep neural network model that may be related to embodiments of the present application are described below.
1. Deep neural network model: in the deep neural network model, the hidden layers may be convolutional layers and pooled layers. The set of weight values corresponding to the convolutional layer is referred to as a filter, also referred to as a convolutional kernel. The filter and the input eigenvalue are both represented as a multi-dimensional matrix, correspondingly, the filter represented as a multi-dimensional matrix is also called a filter matrix, the input eigenvalue represented as a multi-dimensional matrix is also called an input eigenvalue, of course, besides the input eigenvalue, the eigenvector can also be input, and the input eigenvector is only exemplified by the input eigenvector. The operation of the convolution layer is called a convolution operation, which is to perform an inner product operation on a part of eigenvalues of the input eigenvalue matrix and weight values of the filter matrix.
The operation process of each convolution layer in the deep neural network model can be programmed into software, and then the output result of each layer of network, namely the output characteristic matrix, is obtained by running the software in an operation device. For example, the software performs inner product operation by taking the upper left corner of the input feature matrix of each layer of network as a starting point and taking the size of the filter as a window in a sliding window mode, and extracting data of one window from the feature value matrix each time. After the inner product operation is completed between the data of the right lower corner window of the input feature matrix and the filter, a two-dimensional output feature matrix of each layer of network can be obtained. The software repeats the above process until the entire output feature matrix for each layer of network is generated.
The convolution layer operation process is to slide a window with a filter size across the whole input image (i.e. the input feature matrix), and at each moment, to perform inner product operation on the input feature value covered in the window and the filter, wherein the step length of window sliding is 1. Specifically, the upper left corner of the input feature matrix is used as a starting point, the size of the filter is used as a window, the sliding step length of the window is 1, the input feature value of one window is extracted from the feature value matrix each time and the filter performs inner product operation, and when the data of the lower right corner of the input feature matrix and the filter complete inner product operation, a two-dimensional output feature matrix of the input feature matrix can be obtained.
Since it is often necessary to reduce the number of training parameters, the convolutional layer often requires a periodic introduction of a pooling layer, the only purpose of which is to reduce the spatial size of the image during image processing. The pooling layer may include an average pooling operator and/or a maximum pooling operator for sampling the input image to obtain a smaller size image. The average pooling operator may calculate pixel values in the image over a particular range to produce an average as a result of the average pooling. The max pooling operator may take the pixel with the largest value in a particular range as the result of max pooling. In addition, just as the size of the weighting matrix used in the convolutional layer should be related to the image size, the operators in the pooling layer should also be related to the image size. The size of the image output after the processing by the pooling layer can be smaller than the size of the image input to the pooling layer, and each pixel point in the image output by the pooling layer represents the average value or the maximum value of the corresponding sub-region of the image input to the pooling layer.
Since the functions actually required to be simulated in the deep neural network are nonlinear, the rolling and pooling can only simulate linear functions, in order to introduce nonlinear factors in the deep neural network model to increase the characterization capacity of the whole network, an activation layer is further arranged after the pooling layer, an activation function is arranged in the activation layer, and common excitation functions comprise sigmoid, tanh, reLU functions.
2. Softmax classification function: the Softmax classification function is also called soft maximum function, normalized exponential function. One K-dimensional vector containing arbitrary real numbers can be "compressed" into another K-dimensional real vector such that each element ranges between (0, 1) and the sum of all elements is 1. The Softmax classification function is commonly used to classify problems.
Having described the relevant terms and concepts of the deep neural network model to which embodiments of the present application may relate, the following description of the basic principles of the present application will be presented for ease of understanding by those skilled in the art.
As described above, in the conventional scheme, impurities such as raw materials, byproducts of the manufacturing process, and the like, derived from phosphorus pentafluoride, are generated in the process of preparing phosphorus pentafluoride, resulting in lower purity of the prepared phosphorus pentafluoride. However, when phosphorus pentafluoride is purified by a purification method using a common gas such as a difference in boiling point, it is not easy to obtain phosphorus pentafluoride of high purity by removing a mixture of byproducts contained in a raw material or derived from a reaction by using a simple apparatus. Thus, an optimized phosphorus pentafluoride preparation scheme is desired.
Specifically, in the technical scheme of the application, a preparation method of phosphorus pentafluoride is provided, which comprises the following steps: passing phosphorus pentafluoride comprising gaseous impurities through a metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate; and, subjecting the hexafluorophosphate to thermal decomposition to obtain purified phosphorus pentafluoride.
Accordingly, it is considered that the control of parameters of the contact temperature value and the reaction pressure value has an important influence on the curing efficiency of phosphorus pentafluoride in the curing process when curing phosphorus pentafluoride containing gas impurities in the actual process of preparing phosphorus pentafluoride, and the flow rate value of the phosphorus pentafluoride in the actual curing process also affects the contact surface area of the curing, and further affects the curing efficiency and quality. Accordingly, in the technical scheme of the application, it is desirable to comprehensively control the curing process based on the flowing-in flow rate value, the contact temperature value and the reaction pressure value of the phosphorus pentafluoride containing the gas impurities, so as to optimize the efficiency of the curing process and the preparation quality of the phosphorus pentafluoride. However, it is considered that the flow rate value of the phosphorus pentafluoride, the contact temperature value and the reaction pressure value all have a specific change rule in the time dimension, and the contact temperature value and the reaction pressure value also have mutual influence in time sequence, so that the curing effect of the phosphorus pentafluoride is determined together with the flow rate value of the phosphorus pentafluoride. Therefore, in this process, it is difficult to establish a mapping relationship between the time-series cooperative variation of the contact temperature and the reaction pressure and the time-series variation of the flow rate value of the phosphorus pentafluoride, thereby adaptively adjusting the flow rate value of the phosphorus pentafluoride based on the actual time-series cooperative variation of the contact temperature value and the reaction pressure value to optimize the efficiency of the curing process and the preparation quality of the phosphorus pentafluoride.
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.
Deep learning and development of a neural network provide new solutions and schemes for mining complex mapping relations between time sequence cooperative changes of the contact temperature and the reaction pressure and time sequence changes of the flowing-in flow velocity value of the phosphorus pentafluoride. Those of ordinary skill in the art will appreciate that a deep learning based deep neural network model may adjust parameters of the deep neural network model by appropriate training strategies, such as by a gradient descent back-propagation algorithm, to enable modeling of complex nonlinear correlations between things, which is obviously suitable for modeling and establishing complex mappings between time-series coordinated changes in contact temperature and reaction pressure and time-series changes in the phosphorus pentafluoride's turn-in flow rate values.
Specifically, in the technical scheme of the application, firstly, the flowing-in flow rate values of the phosphorus pentafluoride containing the gas impurities at a plurality of preset time points in a preset time period, the contact temperature values of the preset time points and the reaction pressure values of the preset time points are obtained. Then, it is considered that the contact temperature value and the reaction pressure value have time-sequential dynamic variation characteristics in the time dimension, and the parameter data of the contact temperature value and the reaction pressure value also have parameter-synergetic relevance characteristic distribution information in the time dimension. Therefore, in the technical scheme of the application, in order to accurately express the time sequence collaborative correlation characteristic of the contact temperature value and the reaction pressure value, the contact temperature values at a plurality of preset time points and the reaction pressure values at a plurality of preset time points are further arranged into a contact temperature input vector and a reaction pressure input vector according to a time dimension respectively, and then the contact temperature input vector and the reaction pressure input vector are subjected to correlation coding, so that the correlation relation of the contact temperature value and the reaction pressure value on the time sequence distribution is established, and a contact temperature-pressure correlation matrix is obtained.
Then, a time sequence correlation characteristic extraction of the contact temperature value and the reaction pressure value is performed by using a convolution neural network model with excellent performance in terms of implicit correlation characteristic extraction, specifically, the contact temperature-pressure correlation matrix is processed by the convolution neural network model comprising a block structure characteristic extraction module, so as to extract time sequence cooperative correlation characteristic distribution information of the contact temperature value and the reaction pressure value, and thus a contact temperature-pressure correlation characteristic matrix is obtained.
Regarding the inflow rate value of the phosphorus pentafluoride containing the gas impurity, it is considered that since the inflow rate value also has a time-series variation law in the time dimension, such variation law has an important influence on the curing contact surface area of the phosphorus pentafluoride, that is, has an influence on the curing efficiency. Therefore, in order to enhance the expression effect of the time sequence change characteristics of the inlet flow velocity value of phosphorus pentafluoride containing the gas impurities, the inlet flow velocity values of the plurality of preset time points are arranged into flow velocity time sequence input vectors according to the time dimension, and then feature mining is carried out in a one-dimensional convolutional neural network model, so that time sequence dynamic associated feature information of the inlet flow velocity values in the time dimension is extracted, and the flow velocity time sequence feature vectors are obtained.
Further, the contact temperature-pressure correlation characteristic matrix and the flow velocity time sequence characteristic vector are multiplied by a matrix so as to fuse time sequence cooperative correlation characteristic information of the contact temperature value and the reaction pressure value with time sequence dynamic change characteristic information of the flowing-in flow velocity value, thereby obtaining a classification characteristic vector of correlation characteristic distribution information with time sequence cooperative correlation characteristics of the contact temperature and the reaction pressure and time sequence change characteristics of the flowing-in flow velocity. The classification feature vector is then passed through a classifier to obtain a classification result that indicates whether the flow rate value at the current point in time should be increased or decreased.
That is, in the technical solution of the present application, the labels of the classifier include that the current time point of the flow rate value should be increased (first label) and that the current time point of the flow rate value should be decreased (second label), wherein the classifier determines to which classification label the classification feature vector belongs through a soft maximum function. It should be noted that the first tag p1 and the second tag p2 do not include the concept of artificial setting, and in fact, during the training process, the computer model does not have the concept of "the current time point of the flow rate value should be increased or should be decreased", which is simply that there are two kinds of classification tags and the probability that the output feature is at the two classification tags sign, i.e., the sum of p1 and p2 is one. Therefore, the classification result that the flow rate value of the current time point should be increased or decreased is actually converted into the classified probability distribution conforming to the natural rule through classifying the tag, and the physical meaning of the natural probability distribution of the tag is essentially used instead of the language text meaning that the flow rate value of the current time point should be increased or decreased. It should be understood that, in the technical solution of the present application, the classification label of the classifier is a control policy label of the current time point of the flow rate value, so after the classification result is obtained, the current time point of the flow rate value can be adaptively adjusted based on the classification result, so as to optimize the efficiency of the curing process and the preparation quality of phosphorus pentafluoride.
In particular, in the technical solution of the present application, when the contact temperature-pressure correlation feature matrix is multiplied by the flow velocity time sequence feature vector to obtain the classification feature vector, the time sequence correlation feature of the flow velocity expressed by the flow velocity time sequence feature vector is mapped into the high-order cross structural correlation feature of the temperature-pressure value, so that the present application considers that if the feature representation of the contact temperature-pressure correlation feature matrix can be further constrained, the feature expression effect of the contact temperature-pressure correlation feature matrix can be improved, and the accuracy of the classification result of the classification feature vector is improved.
Based on this, the present application uses the responsiveness of the classification feature vector to the flow velocity timing feature vector as a constraint on the feature representation of the contact temperature-pressure correlation feature matrix. Specifically, a responsiveness estimate of the classification feature vector relative to the flow rate timing feature vector is first calculated to obtain a responsiveness estimate feature matrix, e.g., denoted asEstimating a feature matrix/>, based on the responsivenessFor the contact temperature-pressure correlation characteristic matrix, for example, it is noted thatConvolved dictionary-against-response learning to optimize the contact temperature-pressure correlation feature matrix, e.g., denoted/>The method specifically comprises the following steps: /(I)Wherein/>Representing the Frobenius norm of the matrix.
That is, a feature matrix is estimated based on the responsivenessNeighborhood operator attributes characterized by convolution kernels of the convolutional neural network, associated feature matrix/>, for the contact temperature-pressure by convolutional dictionary contrast learning based on differential feature flows between corresponding featuresThe n-level (n-hop) neighbor of the eigenvalue of the (n-hop) is subjected to eigenvalue expression of an eigenvalue prior structure, and prior knowledge under low-rank expression is used as the characteristic response reference of high-dimensional characteristic distribution, so that the interpretive response among the characteristics is learned, and the optimized contact temperature-pressure correlation characteristic matrix/>, is improvedThereby improving the calculation of the optimized contact temperature-pressure correlation characteristic matrix/>Accuracy of the classification result of the classification feature vector obtained by matrix multiplication of the flow velocity time sequence feature vector. Therefore, the flowing-in flow speed value of the phosphorus pentafluoride can be adaptively adjusted in real time and accurately based on the time sequence cooperative change condition of the actual contact temperature value and the reaction pressure value, so that the efficiency of a curing process and the preparation quality of the phosphorus pentafluoride are optimized.
Fig. 1 is an application scenario diagram of a preparation method of phosphorus pentafluoride according to an embodiment of the application. As shown in fig. 1, in this application scenario, the inflow flow rate values of phosphorus pentafluoride containing gas impurities at a plurality of predetermined time points within a predetermined period of time are acquired by a flow rate sensor (e.g., V illustrated in fig. 1), the contact temperature values at the plurality of predetermined time points are acquired by a temperature sensor (e.g., T illustrated in fig. 1), and the reaction pressure values at the plurality of predetermined time points are acquired by a pressure sensor (e.g., P illustrated in fig. 1). Then, the collected inflow flow rate values at the plurality of predetermined time points, the contact temperature values at the plurality of predetermined time points, and the reaction pressure values at the plurality of predetermined time points are input into a server (e.g., S illustrated in fig. 1) in which a preparation algorithm of phosphorus pentafluoride is deployed, wherein the server is capable of processing input data using the preparation algorithm of phosphorus pentafluoride to generate a classification result indicating that the inflow flow rate value at the current time point should be increased or decreased.
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.
As described above, in the conventional scheme, impurities such as raw materials, byproducts of the manufacturing process, and the like, derived from phosphorus pentafluoride, are generated in the process of preparing phosphorus pentafluoride, resulting in lower purity of the prepared phosphorus pentafluoride. However, when phosphorus pentafluoride is purified by a purification method using a common gas such as a difference in boiling point, it is not easy to obtain phosphorus pentafluoride of high purity by removing a mixture of byproducts contained in a raw material or derived from a reaction by using a simple apparatus. Fig. 2 is a schematic flow chart of a method for producing phosphorus pentafluoride according to an embodiment of the application. As shown in fig. 2, a method for preparing phosphorus pentafluoride, the method comprising: s110, passing phosphorus pentafluoride containing gas impurities through a metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate; and S120, carrying out thermal decomposition on the hexafluorophosphate to obtain purified phosphorus pentafluoride.
In the actual process of preparing phosphorus pentafluoride, the control of parameters of a contact temperature value and a reaction pressure value has an important influence on the curing efficiency of phosphorus pentafluoride in a curing process when curing phosphorus pentafluoride containing gas impurities, and the flowing-in flow rate value of phosphorus pentafluoride also influences the curing contact surface area and further influences the curing efficiency and quality in the actual curing process. Accordingly, in the technical scheme of the application, it is desirable to comprehensively control the curing process based on the flowing-in flow rate value, the contact temperature value and the reaction pressure value of the phosphorus pentafluoride containing the gas impurities, so as to optimize the efficiency of the curing process and the preparation quality of the phosphorus pentafluoride. However, it is considered that the flow rate value of the phosphorus pentafluoride, the contact temperature value and the reaction pressure value all have a specific change rule in the time dimension, and the contact temperature value and the reaction pressure value also have mutual influence in time sequence, so that the curing effect of the phosphorus pentafluoride is determined together with the flow rate value of the phosphorus pentafluoride. Therefore, in this process, it is difficult to establish a mapping relationship between the time-series cooperative variation of the contact temperature and the reaction pressure and the time-series variation of the flow rate value of the phosphorus pentafluoride, thereby adaptively adjusting the flow rate value of the phosphorus pentafluoride based on the actual time-series cooperative variation of the contact temperature value and the reaction pressure value to optimize the efficiency of the curing process and the preparation quality of the phosphorus pentafluoride. Fig. 3 is a schematic flow chart of a method for preparing phosphorus pentafluoride according to an embodiment of the application, in which phosphorus pentafluoride containing gas impurities is passed through metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate. S210, obtaining the flowing flow rate value of the phosphorus pentafluoride containing the gas impurities at a plurality of preset time points in a preset time period, the contact temperature value of the preset time points and the reaction pressure value of the preset time points; s220, after the contact temperature values at the plurality of preset time points and the reaction pressure values at the plurality of preset time points are respectively arranged into a contact temperature input vector and a reaction pressure input vector according to a time dimension, carrying out association coding on the contact temperature input vector and the reaction pressure input vector to obtain a contact temperature-pressure association matrix; s230, passing the contact temperature-pressure correlation matrix through a convolutional neural network model comprising a block structural feature extraction module to obtain a contact temperature-pressure correlation feature matrix; s240, arranging the inlet flow velocity values of the plurality of preset time points into flow velocity time sequence input vectors according to time dimensions, and then obtaining flow velocity time sequence feature vectors through a one-dimensional convolutional neural network model; s250, multiplying the contact temperature-pressure correlation feature matrix and the flow velocity time sequence feature vector by a matrix to obtain a classification feature vector; s260, based on the classification feature vector, performing feature expression constraint on the contact temperature-pressure correlation feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix; s270, multiplying the optimized contact temperature-pressure correlation feature matrix and the flow velocity time sequence feature vector by a matrix to obtain an optimized classification feature vector; and S280, passing the optimized classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow rate value of the current time point is increased or decreased.
Fig. 4 is a schematic diagram of a model architecture of passing phosphorus pentafluoride containing gas impurities through a metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate in the preparation method of phosphorus pentafluoride according to the embodiment of the application. As shown in fig. 4, the input of the model architecture for passing phosphorus pentafluoride containing gas impurities through metal fluoride to solidify the phosphorus pentafluoride in the form of hexafluorophosphate is a flow-in flow rate value at a plurality of predetermined time points, a contact temperature value at a plurality of predetermined time points, and a reaction pressure value at a plurality of predetermined time points. And then, after the contact temperature values at the plurality of preset time points and the reaction pressure values at the plurality of preset time points are respectively arranged into a contact temperature input vector and a reaction pressure input vector according to the time dimension, carrying out association coding on the contact temperature input vector and the reaction pressure input vector to obtain a contact temperature-pressure association matrix. Then, the contact temperature-pressure correlation matrix is passed through a convolutional neural network model comprising a block structural feature extraction module to obtain a contact temperature-pressure correlation feature matrix. And meanwhile, arranging the flowing-in flow velocity values of the plurality of preset time points into a flow velocity time sequence input vector according to a time dimension, and then obtaining a flow velocity time sequence feature vector through a one-dimensional convolutional neural network model. And then, multiplying the contact temperature-pressure correlation characteristic matrix with the flow velocity time sequence characteristic vector by a matrix to obtain a classification characteristic vector. And then, based on the classification feature vector, performing feature expression constraint on the contact temperature-pressure correlation feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix. And finally, multiplying the optimized contact temperature-pressure correlation characteristic matrix with the flow velocity time sequence characteristic vector to obtain an optimized classification characteristic vector, and enabling the optimized classification characteristic vector to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow velocity value of the current time point is increased or decreased.
Step S210, obtaining the flowing-in flow rate values of the phosphorus pentafluoride containing the gas impurities at a plurality of preset time points in a preset time period, the contact temperature values of the preset time points and the reaction pressure values of the preset time points. As described above, the difficulty establishes a mapping relationship between the time-series cooperative variation of the contact temperature and the reaction pressure and the time-series variation of the flow rate value of the phosphorus pentafluoride, thereby adaptively adjusting the flow rate value of the phosphorus pentafluoride based on the actual time-series cooperative variation of the contact temperature value and the reaction pressure value to optimize the efficiency of the curing process and the preparation quality of the phosphorus pentafluoride. So as to facilitate the subsequent mining of the time sequence characteristics and time sequence cooperative characteristics in the inlet flow velocity values of the plurality of preset time points, the contact temperature values of the plurality of preset time points and the contact temperature values of the plurality of preset time points through the deep neural network model,
Step S220, after arranging the contact temperature values at the predetermined time points and the reaction pressure values at the predetermined time points into a contact temperature input vector and a reaction pressure input vector according to a time dimension, performing association encoding on the contact temperature input vector and the reaction pressure input vector to obtain a contact temperature-pressure association matrix. It should be understood that, considering that the contact temperature value and the reaction pressure value have time-sequential dynamic variation characteristics in the time dimension, the two parameter data also have parameter-collaborative relevance characteristic distribution information in the time dimension. Therefore, in the technical scheme of the application, in order to accurately express the time sequence collaborative correlation characteristic of the contact temperature value and the reaction pressure value, the contact temperature values at a plurality of preset time points and the reaction pressure values at a plurality of preset time points are further arranged into a contact temperature input vector and a reaction pressure input vector according to a time dimension respectively, and then the contact temperature input vector and the reaction pressure input vector are subjected to correlation coding, so that the correlation relation of the contact temperature value and the reaction pressure value on the time sequence distribution is established, and a contact temperature-pressure correlation matrix is obtained.
Step S230, passing the contact temperature-pressure correlation matrix through a convolutional neural network model including a block structural feature extraction module to obtain a contact temperature-pressure correlation feature matrix. It should be understood that the time-series correlation feature extraction of the contact temperature value and the reaction pressure value is performed using a convolutional neural network model having excellent performance in terms of implicit correlation feature extraction, specifically, the contact temperature-pressure correlation matrix is processed through the convolutional neural network model including a block structure feature extraction module to extract time-series cooperative correlation feature distribution information of the contact temperature value and the reaction pressure value, thereby obtaining a contact temperature-pressure correlation feature matrix.
Optionally, in an embodiment of the present application, passing the contact temperature-pressure correlation matrix through a convolutional neural network model including a block structural feature extraction module to obtain a contact temperature-pressure correlation feature matrix includes: each layer of the convolutional neural network model comprising the block structural feature extraction module respectively carries out input data in forward transfer of the layer: performing convolution processing, pooling processing and nonlinear processing on the input data based on a first two-dimensional convolution kernel to obtain a first activation feature map; performing convolution processing, pooling processing and nonlinear activation processing based on a second two-dimensional convolution kernel on the first activation feature map to obtain a second activation feature map, wherein the first two-dimensional convolution kernel and the second two-dimensional convolution kernel are transposed with each other; the input of the first layer of the convolution neural network model containing the block structure feature extraction module is the contact temperature-pressure correlation matrix, and the output of the last layer of the convolution neural network model containing the block structure feature extraction module is the contact temperature-pressure correlation matrix.
Step S240, arranging the inlet flow velocity values of the plurality of preset time points into flow velocity time sequence input vectors according to a time dimension, and obtaining flow velocity time sequence feature vectors through a one-dimensional convolutional neural network model. It will be appreciated that considering that the flow rate values also have a time-dependent variation in time dimension, this variation has an important effect on the curing contact surface area of the phosphorus pentafluoride, i.e. on the curing efficiency. Therefore, in order to enhance the expression effect of the time sequence change characteristics of the inlet flow velocity value of phosphorus pentafluoride containing the gas impurities, the inlet flow velocity values of the plurality of preset time points are arranged into flow velocity time sequence input vectors according to the time dimension, and then feature mining is carried out in a one-dimensional convolutional neural network model, so that time sequence dynamic associated feature information of the inlet flow velocity values in the time dimension is extracted, and the flow velocity time sequence feature vectors are obtained.
Optionally, in an embodiment of the present application, the arranging the flow velocity values of the plurality of predetermined time points into the flow velocity time sequence input vector according to a time dimension, and then obtaining the flow velocity time sequence feature vector through a one-dimensional convolutional neural network model includes: each layer of the one-dimensional convolutional neural network model is used for respectively carrying out forward transfer on input data in the layers: performing convolution processing based on a one-dimensional convolution kernel on the input data to obtain a convolution feature map; pooling processing is carried out on the convolution feature images based on feature matrixes to obtain pooled feature images; non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map; the input of the first layer of the one-dimensional convolutional neural network model is the flow velocity time sequence input vector, and the output of the last layer of the one-dimensional convolutional neural network model is the flow velocity time sequence feature vector.
And S250, multiplying the contact temperature-pressure correlation characteristic matrix with the flow velocity time sequence characteristic vector by a matrix to obtain a classification characteristic vector. It is to be understood that the contact temperature-pressure correlation feature matrix is matrix-multiplied by the flow rate timing feature vector to fuse the timing cooperative correlation feature information of the contact temperature value and the reaction pressure value with the timing dynamic variation feature information of the inflow flow rate value, thereby obtaining a classification feature vector having the timing cooperative correlation feature of the contact temperature and the reaction pressure and the correlation feature distribution information of the timing variation feature of the inflow flow rate.
And step S260, carrying out feature expression constraint on the contact temperature-pressure correlation feature matrix based on the classification feature vector so as to obtain an optimized contact temperature-pressure correlation feature matrix. It should be understood that, in the technical solution of the present application, when the contact temperature-pressure correlation feature matrix is multiplied by the flow velocity time sequence feature vector to obtain the classification feature vector, the time sequence correlation feature of the flow velocity expressed by the flow velocity time sequence feature vector is mapped into the high-order cross structural correlation feature of the temperature-pressure value, so that the present application considers that if the feature representation of the contact temperature-pressure correlation feature matrix can be further constrained, the feature expression effect of the contact temperature-pressure correlation feature matrix can be improved, and the accuracy of the classification result of the classification feature vector is improved.
Based on this, the present application uses the responsiveness of the classification feature vector to the flow velocity timing feature vector as a constraint on the feature representation of the contact temperature-pressure correlation feature matrix. Specifically, a responsiveness estimate of the classification feature vector relative to the flow rate timing feature vector is first calculated to obtain a responsiveness estimate feature matrix, e.g., denoted asEstimating a feature matrix/>, based on the responsivenessFor the contact temperature-pressure correlation characteristic matrix, for example, it is noted thatConvolved dictionary-against-response learning to optimize the contact temperature-pressure correlation feature matrix, e.g., denoted/>。
Optionally, in an embodiment of the present application, based on the classification feature vector, performing feature expression constraint on the contact temperature-pressure correlation feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix, including: calculating the responsiveness estimation of the classification feature vector relative to the flow velocity time sequence feature vector to obtain a responsiveness estimation feature matrix; and performing convolution dictionary contrast response learning on the contact temperature-pressure correlation characteristic matrix based on the response estimation characteristic matrix to obtain an optimized contact temperature-pressure correlation characteristic matrix.
Optionally, in an embodiment of the present application, calculating a responsiveness estimate of the classification feature vector relative to the flow rate timing feature vector to obtain a responsiveness estimate feature matrix includes: further used for: calculating a responsiveness estimate of the classification feature vector relative to the flow timing feature vector to obtain the responsiveness estimate feature matrix with the following formula; wherein, the formula is: Wherein/> Representing the classification feature vector,/>Representing the flow velocity time sequence characteristic vector,/>Representing matrix multiplication,/>Representing the responsiveness estimation feature matrix.
Optionally, in an embodiment of the present application, performing convolutional dictionary contrast response learning on the contact temperature-pressure correlation feature matrix based on the responsiveness estimation feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix, including: performing convolution dictionary contrast response learning on the contact temperature-pressure correlation feature matrix according to the response estimation feature matrix by using the following optimization formula so as to obtain the optimized contact temperature-pressure correlation feature matrix; wherein, the formula is: Wherein/> And/>Respectively the responsiveness estimation feature matrix and the contact temperature-pressure correlation feature matrix,/>Frobenius norms,/>, representing matricesRepresenting matrix subtraction,/>Representing matrix multiplication,/>Representing the optimized contact temperature-pressure correlation characteristic matrix.
That is, a feature matrix is estimated based on the responsivenessNeighborhood operator attributes characterized by convolution kernels of the convolutional neural network, associated feature matrix/>, for the contact temperature-pressure by convolutional dictionary contrast learning based on differential feature flows between corresponding featuresThe n-level (n-hop) neighbor of the eigenvalue of the (n-hop) is subjected to eigenvalue expression of an eigenvalue prior structure, and prior knowledge under low-rank expression is used as the characteristic response reference of high-dimensional characteristic distribution, so that the interpretive response among the characteristics is learned, and the optimized contact temperature-pressure correlation characteristic matrix/>, is improvedThereby improving the calculation of the optimized contact temperature-pressure correlation characteristic matrix/>Accuracy of the classification result of the classification feature vector obtained by matrix multiplication of the flow velocity time sequence feature vector. Therefore, the flowing-in flow speed value of the phosphorus pentafluoride can be adaptively adjusted in real time and accurately based on the time sequence cooperative change condition of the actual contact temperature value and the reaction pressure value, so that the efficiency of a curing process and the preparation quality of the phosphorus pentafluoride are optimized.
And step S270, multiplying the optimized contact temperature-pressure correlation characteristic matrix with the flow velocity time sequence characteristic vector by a matrix to obtain an optimized classification characteristic vector. It is to be understood that the optimized contact temperature-pressure correlation feature matrix is matrix multiplied by the flow velocity time sequence feature vector to fuse time sequence cooperative correlation feature information of the contact temperature value and the reaction pressure value in the optimized contact temperature-pressure correlation feature matrix with time sequence dynamic change feature information of the flow velocity value in the flow velocity time sequence feature vector, thereby obtaining an optimized classification feature vector with time sequence cooperative correlation feature of the contact temperature and the reaction pressure and time sequence change feature of the flow velocity.
Optionally, in an embodiment of the present application, multiplying the optimized contact temperature-pressure correlation feature matrix with the flow velocity time sequence feature vector to obtain an optimized classification feature vector includes: calculating the optimized contact temperature-pressure correlation characteristic matrix and the flow velocity time sequence characteristic vector by using the following formula, and multiplying the optimized contact temperature-pressure correlation characteristic matrix and the flow velocity time sequence characteristic vector by using the matrix to obtain an optimized classification characteristic vector; wherein, the formula is: Wherein/> Representing the flow velocity time sequence characteristic vector,/>Representing the optimized classification feature vector,/>Representing the optimized contact temperature-pressure correlation characteristic matrix,/>Representing a matrix multiplication.
Step S280, passing the optimized classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate that the value of the flow rate at the current time point should be increased or decreased. That is, in the technical solution of the present application, the labels of the classifier include that the current time point of the flow rate value should be increased (first label) and that the current time point of the flow rate value should be decreased (second label), wherein the classifier determines to which classification label the classification feature vector belongs through a soft maximum function. It should be noted that the first tag p1 and the second tag p2 do not include the concept of artificial setting, and in fact, during the training process, the computer model does not have the concept of "the current time point of the flow rate value should be increased or should be decreased", which is simply that there are two kinds of classification tags and the probability that the output feature is at the two classification tags sign, i.e., the sum of p1 and p2 is one. Therefore, the classification result that the flow rate value of the current time point should be increased or decreased is actually converted into the classified probability distribution conforming to the natural rule through classifying the tag, and the physical meaning of the natural probability distribution of the tag is essentially used instead of the language text meaning that the flow rate value of the current time point should be increased or decreased. It should be understood that, in the technical solution of the present application, the classification label of the classifier is a control policy label of the current time point of the flow rate value, so after the classification result is obtained, the current time point of the flow rate value can be adaptively adjusted based on the classification result, so as to optimize the efficiency of the curing process and the preparation quality of phosphorus pentafluoride.
Optionally, in an embodiment of the present application, the optimizing the classification feature vector is passed through a classifier to obtain a classification result, where the classification result is used to indicate that the value of the flow rate at the current time point should be increased or should be decreased, and the method includes: processing the transfer vector using the classifier in the following formula to obtain the classification result; wherein, the formula is: wherein/> To/>Is a weight matrix,/>To/>Is bias vector,/>For transfer vector,/>Representing a normalized exponential function.
In summary, the preparation method of phosphorus pentafluoride provided by the application can adaptively adjust the flowing speed value of phosphorus pentafluoride based on the time sequence cooperative change condition of the actual contact temperature value and the reaction pressure value, so as to optimize the efficiency of the curing process and the preparation quality of phosphorus pentafluoride.
It will be appreciated that the specific examples herein are intended only to assist those skilled in the art in better understanding the embodiments of the application and are not intended to limit the scope of the embodiments of the application.
It should also be understood that, in various embodiments of the present application, the sequence number of each process does not mean that the execution sequence of each process should be determined by its functions and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
It should also be understood that the various embodiments described in this specification may be implemented alone or in combination, and that the present embodiments are not limited in this regard.
Unless defined otherwise, all technical and scientific terms used in the embodiments of the application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the present application is for the purpose of describing particular embodiments only and is not intended to limit the scope of the present application. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items. As used in this application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
In the several embodiments provided by the present application, it should be understood that the disclosed systems, devices, and methods may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (8)
1. A method for preparing phosphorus pentafluoride, comprising the steps of:
passing phosphorus pentafluoride comprising gaseous impurities through a metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate; and
Performing thermal decomposition on the hexafluorophosphate to obtain purified phosphorus pentafluoride;
Wherein passing phosphorus pentafluoride containing gaseous impurities through a metal fluoride to cure the phosphorus pentafluoride in the form of hexafluorophosphate comprises:
acquiring the flowing-in flow rate values of the phosphorus pentafluoride containing the gas impurities at a plurality of preset time points in a preset time period, the contact temperature values of the preset time points and the reaction pressure values of the preset time points;
after the contact temperature values at the preset time points and the reaction pressure values at the preset time points are respectively arranged into a contact temperature input vector and a reaction pressure input vector according to the time dimension, carrying out association coding on the contact temperature input vector and the reaction pressure input vector to obtain a contact temperature-pressure association matrix;
the contact temperature-pressure correlation matrix is passed through a convolutional neural network model comprising a block structural feature extraction module to obtain a contact temperature-pressure correlation feature matrix;
Arranging the inlet flow velocity values of the plurality of preset time points into flow velocity time sequence input vectors according to time dimensions, and then obtaining flow velocity time sequence feature vectors through a one-dimensional convolutional neural network model;
Performing matrix multiplication on the contact temperature-pressure correlation characteristic matrix and the flow velocity time sequence characteristic vector to obtain a classification characteristic vector;
Based on the classification feature vector, performing feature expression constraint on the contact temperature-pressure correlation feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix;
performing matrix multiplication on the optimized contact temperature-pressure correlation feature matrix and the flow velocity time sequence feature vector to obtain an optimized classification feature vector; and
And the optimized classification feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow rate value of the current time point is increased or decreased.
2. The method for preparing phosphorus pentafluoride according to claim 1, wherein passing the contact temperature-pressure correlation matrix through a convolutional neural network model including a block structural feature extraction module to obtain a contact temperature-pressure correlation feature matrix comprises: each layer of the convolutional neural network model comprising the block structural feature extraction module respectively carries out input data in forward transfer of the layer:
performing convolution processing, pooling processing and nonlinear processing on the input data based on a first two-dimensional convolution kernel to obtain a first activation feature map; and
Performing convolution processing, pooling processing and nonlinear activation processing based on a second two-dimensional convolution kernel on the first activation feature map to obtain a second activation feature map, wherein the first two-dimensional convolution kernel and the second two-dimensional convolution kernel are transposed with each other;
the input of the first layer of the convolution neural network model containing the block structure feature extraction module is the contact temperature-pressure correlation matrix, and the output of the last layer of the convolution neural network model containing the block structure feature extraction module is the contact temperature-pressure correlation matrix.
3. The method for preparing phosphorus pentafluoride according to claim 2, wherein the step of arranging the flow velocity values of the plurality of predetermined time points into the flow velocity time sequence input vector according to the time dimension and then obtaining the flow velocity time sequence feature vector through a one-dimensional convolutional neural network model comprises the following steps: each layer of the one-dimensional convolutional neural network model is used for respectively carrying out forward transfer on input data in the layers:
performing convolution processing based on a one-dimensional convolution kernel on the input data to obtain a convolution feature map;
pooling processing is carried out on the convolution feature images based on feature matrixes to obtain pooled feature images;
non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map;
the input of the first layer of the one-dimensional convolutional neural network model is the flow velocity time sequence input vector, and the output of the last layer of the one-dimensional convolutional neural network model is the flow velocity time sequence feature vector.
4. The method for preparing phosphorus pentafluoride according to claim 3, characterized in that performing feature expression constraint on the contact temperature-pressure correlation feature matrix based on the classification feature vector to obtain an optimized contact temperature-pressure correlation feature matrix, comprises:
calculating the responsiveness estimation of the classification feature vector relative to the flow velocity time sequence feature vector to obtain a responsiveness estimation feature matrix; and
And performing convolution dictionary contrast response learning on the contact temperature-pressure correlation characteristic matrix based on the responsiveness estimation characteristic matrix to obtain an optimized contact temperature-pressure correlation characteristic matrix.
5. The method of claim 4, wherein calculating a responsiveness estimate of the classification feature vector relative to the flow timing feature vector to obtain a responsiveness estimate feature matrix comprises: for the purpose of: calculating a responsiveness estimate of the classification feature vector relative to the flow timing feature vector to obtain the responsiveness estimate feature matrix with the following formula;
wherein, the formula is:
，
Wherein the method comprises the steps of Representing the classification feature vector,/>Representing the flow velocity time sequence characteristic vector,/>Representing a matrix multiplication of the number of bits,Representing the responsiveness estimation feature matrix.
6. The method of claim 5, wherein performing convolutional dictionary contrast response learning on the contact temperature-pressure correlation feature matrix based on the estimated response feature matrix to obtain an optimized contact temperature-pressure correlation feature matrix comprises:
Performing convolution dictionary contrast response learning on the contact temperature-pressure correlation feature matrix according to the response estimation feature matrix by using the following optimization formula so as to obtain the optimized contact temperature-pressure correlation feature matrix;
wherein, the formula is:
，
Wherein the method comprises the steps of And/>The responsiveness estimated feature matrix and the contact temperature-pressure associated feature matrix, respectively, and/>Frobenius norms,/>, representing matricesRepresenting matrix subtraction,/>Representing matrix multiplication,/>Representing the optimized contact temperature-pressure correlation characteristic matrix.
7. The method for preparing phosphorus pentafluoride according to claim 6, characterized in that the matrix multiplication of the optimized contact temperature-pressure correlation feature matrix and the flow velocity time sequence feature vector is performed to obtain an optimized classification feature vector, comprising: calculating the optimized contact temperature-pressure correlation characteristic matrix and the flow velocity time sequence characteristic vector by using the following formula, and performing matrix multiplication to obtain an optimized classification characteristic vector;
wherein, the formula is:
，
Wherein the method comprises the steps of Representing the optimized classification feature vector,/>Representing the flow velocity time sequence characteristic vector,/>Representing the optimized contact temperature-pressure correlation characteristic matrix,/>Representing a matrix multiplication.
8. The method for preparing phosphorus pentafluoride according to claim 7, wherein the step of passing the optimized classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating that the flow rate value of the current time point should be increased or decreased, comprises the following steps: processing the optimized classification feature vector using the classifier in the following formula to obtain the classification result;
wherein, the formula is: wherein/> To/>As a matrix of weights, the weight matrix,To/>Is bias vector,/>To optimize the classification feature vector,/>Representing a normalized exponential function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310479969.0A CN116835540B (en) | 2023-04-28 | 2023-04-28 | Preparation method of phosphorus pentafluoride |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310479969.0A CN116835540B (en) | 2023-04-28 | 2023-04-28 | Preparation method of phosphorus pentafluoride |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116835540A CN116835540A (en) | 2023-10-03 |
| CN116835540B true CN116835540B (en) | 2024-05-21 |
Family
ID=88167729
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310479969.0A Active CN116835540B (en) | 2023-04-28 | 2023-04-28 | Preparation method of phosphorus pentafluoride |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116835540B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105593165A (en) * | 2013-10-04 | 2016-05-18 | 关东电化工业株式会社 | Method for purifying phosphorus pentafluoride |
| CN112553817A (en) * | 2020-12-11 | 2021-03-26 | 河南省纺织产品质量监督检验院 | Method and system for real-time control and adjustment of textile dyeing process based on artificial intelligence |
| CN112766303A (en) * | 2020-12-23 | 2021-05-07 | 湖南大学 | CNN-based aeroengine fault diagnosis method |
| CN114444790A (en) * | 2022-01-19 | 2022-05-06 | 天津大学 | Method for predicting time sequence of various measuring points on gas turbine based on steady-state feature picture |
| CN115586755A (en) * | 2022-10-28 | 2023-01-10 | 福建省龙德新能源有限公司 | Production management control system and method for preparing lithium hexafluorophosphate |
| CN115720964A (en) * | 2022-10-27 | 2023-03-03 | 内蒙古国方农牧业科技有限公司 | Processing method and device for degrading and puffing shrubs into protein forage grass |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20210033769A (en) * | 2019-09-19 | 2021-03-29 | 삼성전자주식회사 | Electric apparatus and operation method of the electric apparatus |
| US12044562B2 (en) * | 2020-06-26 | 2024-07-23 | Schlumberger Technology Corporation | Multiphase flowmeter and related methods |
-
2023
- 2023-04-28 CN CN202310479969.0A patent/CN116835540B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105593165A (en) * | 2013-10-04 | 2016-05-18 | 关东电化工业株式会社 | Method for purifying phosphorus pentafluoride |
| CN112553817A (en) * | 2020-12-11 | 2021-03-26 | 河南省纺织产品质量监督检验院 | Method and system for real-time control and adjustment of textile dyeing process based on artificial intelligence |
| CN112766303A (en) * | 2020-12-23 | 2021-05-07 | 湖南大学 | CNN-based aeroengine fault diagnosis method |
| CN114444790A (en) * | 2022-01-19 | 2022-05-06 | 天津大学 | Method for predicting time sequence of various measuring points on gas turbine based on steady-state feature picture |
| CN115720964A (en) * | 2022-10-27 | 2023-03-03 | 内蒙古国方农牧业科技有限公司 | Processing method and device for degrading and puffing shrubs into protein forage grass |
| CN115586755A (en) * | 2022-10-28 | 2023-01-10 | 福建省龙德新能源有限公司 | Production management control system and method for preparing lithium hexafluorophosphate |
Non-Patent Citations (2)
| Title |
|---|
| Machine Learning Based Methods for Obtaining Correlations between Microstructures and Thermal Stresses;Akshay Bhutada 等;Metals;20210722;第11卷;1-17 * |
| 回转窑生产过程工况识别研究;高振洋;中国优秀硕士学位论文全文数据库工程科技Ⅰ辑;20220215(第2期);B015-410页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116835540A (en) | 2023-10-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yuan et al. | Tensor ring decomposition with rank minimization on latent space: An efficient approach for tensor completion | |
| CN108549839B (en) | Adaptive feature fusion multi-scale correlation filtering visual tracking method | |
| CN115936682A (en) | Waste recovery system and method for printed circuit board | |
| CN115393779A (en) | Control system and control method for laser cladding metal ball manufacturing | |
| CN116680987A (en) | Forming method and system of graphite crucible | |
| CN109376787A (en) | Manifold learning network and computer visual image collection classification method based on it | |
| CN116434117A (en) | Preparation method of composite polyacrylamide oil displacement agent | |
| CN116683092B (en) | Disassembling system and method for new energy automobile battery | |
| CN116184972A (en) | Control system and method for air separation device | |
| CN116086790A (en) | Performance detection method and system for high-pressure valve of hydrogen fuel cell | |
| CN116796248A (en) | Forest health environment assessment system and method thereof | |
| CN110363713B (en) | Hyperspectral image noise reduction method based on recursive sample scaling and bilinear factorization | |
| Wayahdi et al. | Evaluation of the K-Nearest Neighbor Model With K-Fold Cross Validation on Image Classification | |
| CN117404853B (en) | External circulating water cooling system and method for tunnel boring machine | |
| CN117274726B (en) | Picture classification method and system based on multi-view supplementary tag | |
| CN113177970B (en) | Multi-scale filtering target tracking method based on self-adaptive feature fusion | |
| CN114511785A (en) | Remote sensing image cloud detection method and system based on bottleneck attention module | |
| CN112215241B (en) | Image feature extraction device based on small sample learning | |
| CN116835540B (en) | Preparation method of phosphorus pentafluoride | |
| CN114202694A (en) | Small sample remote sensing scene image classification method based on manifold mixed interpolation and contrast learning | |
| CN106960225B (en) | sparse image classification method based on low-rank supervision | |
| CN117710841A (en) | Small target detection method and device for aerial image of unmanned aerial vehicle | |
| CN116177858A (en) | Preparation method and system of high-purity quartz crucible | |
| CN109215057B (en) | High-performance visual tracking method and device | |
| CN112183299A (en) | Pedestrian attribute prediction method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |