WO2024045243A1 - 用于六氟磷酸锂制备的自动配料系统及其配料方法 - Google Patents

用于六氟磷酸锂制备的自动配料系统及其配料方法 Download PDF

Info

Publication number
WO2024045243A1
WO2024045243A1 PCT/CN2022/120829 CN2022120829W WO2024045243A1 WO 2024045243 A1 WO2024045243 A1 WO 2024045243A1 CN 2022120829 W CN2022120829 W CN 2022120829W WO 2024045243 A1 WO2024045243 A1 WO 2024045243A1
Authority
WO
WIPO (PCT)
Prior art keywords
temperature
matrix
flow rate
feature
predetermined time
Prior art date
Application number
PCT/CN2022/120829
Other languages
English (en)
French (fr)
Inventor
杨永淮
郭剑煌
华杭州
傅少鹏
夏维亮
张永炎
Original Assignee
福建省龙德新能源有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 福建省龙德新能源有限公司 filed Critical 福建省龙德新能源有限公司
Publication of WO2024045243A1 publication Critical patent/WO2024045243A1/zh

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B01—PHYSICAL OR CHEMICAL PROCESSES OR APPARATUS IN GENERAL
    • B01F—MIXING, e.g. DISSOLVING, EMULSIFYING OR DISPERSING
    • B01F35/00—Accessories for mixers; Auxiliary operations or auxiliary devices; Parts or details of general application
    • B01F35/20—Measuring; Control or regulation
    • B01F35/22—Control or regulation
    • B01F35/222—Control or regulation of the operation of the driving system, e.g. torque, speed or power of motors; of the position of mixing devices or elements
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B01—PHYSICAL OR CHEMICAL PROCESSES OR APPARATUS IN GENERAL
    • B01F—MIXING, e.g. DISSOLVING, EMULSIFYING OR DISPERSING
    • B01F35/00—Accessories for mixers; Auxiliary operations or auxiliary devices; Parts or details of general application
    • B01F35/20—Measuring; Control or regulation
    • B01F35/22—Control or regulation
    • B01F35/221—Control or regulation of operational parameters, e.g. level of material in the mixer, temperature or pressure
    • B01F35/2211—Amount of delivered fluid during a period
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B01—PHYSICAL OR CHEMICAL PROCESSES OR APPARATUS IN GENERAL
    • B01F—MIXING, e.g. DISSOLVING, EMULSIFYING OR DISPERSING
    • B01F35/00—Accessories for mixers; Auxiliary operations or auxiliary devices; Parts or details of general application
    • B01F35/20—Measuring; Control or regulation
    • B01F35/22—Control or regulation
    • B01F35/221—Control or regulation of operational parameters, e.g. level of material in the mixer, temperature or pressure
    • B01F35/2214—Speed during the operation
    • B01F35/22141—Speed of feeding of at least one component to be mixed
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B01—PHYSICAL OR CHEMICAL PROCESSES OR APPARATUS IN GENERAL
    • B01F—MIXING, e.g. DISSOLVING, EMULSIFYING OR DISPERSING
    • B01F35/00—Accessories for mixers; Auxiliary operations or auxiliary devices; Parts or details of general application
    • B01F35/20—Measuring; Control or regulation
    • B01F35/22—Control or regulation
    • B01F35/221—Control or regulation of operational parameters, e.g. level of material in the mixer, temperature or pressure
    • B01F35/2215—Temperature
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B01—PHYSICAL OR CHEMICAL PROCESSES OR APPARATUS IN GENERAL
    • B01F—MIXING, e.g. DISSOLVING, EMULSIFYING OR DISPERSING
    • B01F35/00—Accessories for mixers; Auxiliary operations or auxiliary devices; Parts or details of general application
    • B01F35/80—Forming a predetermined ratio of the substances to be mixed
    • B01F35/83—Forming a predetermined ratio of the substances to be mixed by controlling the ratio of two or more flows, e.g. using flow sensing or flow controlling devices
    • B01F35/831—Forming a predetermined ratio of the substances to be mixed by controlling the ratio of two or more flows, e.g. using flow sensing or flow controlling devices using one or more pump or other dispensing mechanisms for feeding the flows in predetermined proportion, e.g. one of the pumps being driven by one of the flows
    • 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
    • C01B25/16—Oxyacids of phosphorus; Salts thereof
    • C01B25/26—Phosphates
    • C01B25/455—Phosphates containing halogen
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • 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

Definitions

  • the present invention relates to the field of intelligent manufacturing, and more specifically, to an automatic batching system for preparing lithium hexafluorophosphate and a batching method thereof.
  • lithium hexafluorophosphate (LiPF6) has been selected as the electrolyte of lithium ion secondary batteries.
  • Lithium hexafluorophosphate is dissolved in certain non-aqueous organic solvents to form the electrolyte of lithium ion secondary batteries.
  • Lithium-ion secondary batteries are internationally recognized as an ideal chemical energy source. They are widely used in mobile phones, laptops and laptops due to their small size, large capacity, ability to be repeatedly charged and discharged 500 times, and only a 3% reduction in capacity. Cameras, etc., are also used in the electronics industry to make wafer dopants and catalysts for organic synthesis.
  • Lithium ions move through the electrolyte solution, with the result that in order to maintain battery performance, such as service life, the purity of the electrolyte in the electrolyte solution should be strictly controlled.
  • LiPF6 lithium hexafluorophosphate
  • an automatic batching system for the preparation of lithium hexafluorophosphate is expected to improve the preparation efficiency of lithium hexafluorophosphate products while simplifying the preparation process of lithium hexafluorophosphate.
  • the embodiments of the present application provide an automatic batching system for the preparation of lithium hexafluorophosphate and a batching method thereof, which uses an artificial intelligence control method and a convolutional neural network model to control phosphorus pentafluoride gas at multiple predetermined time points.
  • the correlation matrix between the gas flow rate value and the reaction temperature value and the liquid chromatogram of the reaction liquid are implicitly extracted to determine the phosphorus pentafluoride gas introduced during the reaction process during the preparation process of the lithium hexafluorophosphate product.
  • the flow rate is controlled in real-time and dynamically with the reaction temperature change, thereby improving the preparation efficiency of the lithium hexafluorophosphate product.
  • an automatic batching system for the preparation of lithium hexafluorophosphate which includes:
  • the gas batching data acquisition module is used to obtain the gas flow rate value of phosphorus pentafluoride gas flowing into the anhydrous hydrogen fluoride solution containing lithium fluoride at multiple predetermined time points within a predetermined time period;
  • the reaction data acquisition module is used to obtain The reaction temperature values at multiple predetermined time points within the predetermined time period and the liquid chromatogram of the reaction liquid;
  • a discrete data vectorization module for converting the phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period.
  • the gas flow rate values and reaction temperature values are arranged into temperature input vectors and flow rate input vectors according to the time dimension respectively;
  • a data-level correlation module used to calculate the product between the transpose vector of the temperature input vector and the flow rate input vector to obtain the temperature-flow rate correlation matrix
  • a correction module used to calculate the position of the temperature-flow rate correlation matrix based on the Information, correct the temperature-flow velocity correlation matrix to obtain the corrected temperature-flow velocity correlation matrix-flow velocity correlation matrix
  • the feature extraction module is used to pass the corrected temperature-flow velocity correlation matrix through the first volume of the filter Accumulate the neural network to obtain the temperature-flow rate correlation feature vector
  • the product data encoding module is used to pass the liquid chromatogram of the reaction solution at multiple predetermined time points within the predetermined time period through the second convolution using a three-dimensional convolution kernel a neural network to obtain a product feature vector
  • a responsiveness estimation module for calculating a responsiveness estimate of the product feature vector relative to the temperature-flow rate correlation feature vector to obtain a classification feature matrix
  • a batching control result generation module is used to pass the classification feature matrix through a classifier to obtain a classification result.
  • the classification result is used to indicate that the gas flow rate value of phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • the correction module includes: a transposition unit for calculating the transpose matrix of the temperature-flow rate correlation matrix; a first convolution unit for calculating the temperature - The flow velocity correlation matrix is convolved to obtain the first convolution feature matrix; a fusion unit is used to calculate the transpose matrix of the temperature-flow velocity correlation matrix and the position-wise points of the first convolution feature matrix to obtain a preliminary Fusion feature matrix; a second convolution unit, used to perform convolution processing on the preliminary fusion feature matrix to obtain a second convolution feature matrix; a position information extraction unit, used to convert each position in the temperature-flow velocity correlation matrix The two-dimensional position coordinates are mapped to one-dimensional values to obtain the position information matrix; and, a re-fusion unit is used to calculate the second convolution feature matrix and the position information matrix by position point to obtain the corrected temperature -Flow velocity correlation matrix.
  • M represents the temperature-flow velocity correlation matrix
  • M 1 represents the position information matrix
  • P M represents the (x, y) coordinate matrix of the temperature-flow velocity correlation matrix, Used to map two-dimensional position coordinates to one-dimensional values.
  • the feature extraction module is further used to: each layer of the first convolutional neural network as a filter performs on the input data respectively in the forward transmission of the layer: Perform convolution processing on the input data to obtain a convolution feature map; perform mean pooling based on the local feature matrix on the convolution feature map to obtain a pooled feature map; and perform nonlinear activation on the pooled feature map.
  • the output of the last layer of the first convolutional neural network as a filter is the temperature-flow velocity correlation feature vector
  • the output of the first convolutional neural network as a filter is The input of one layer is the corrected temperature-flow rate correlation matrix.
  • the product data encoding module is further used: the second convolutional neural network using a three-dimensional convolution kernel performs on the input data respectively in the forward transmission of the layer: Perform three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution feature map; perform mean pooling processing based on a local feature matrix on the convolution feature map to obtain a pooled feature map; and The pooled feature map performs nonlinear activation to obtain an activation feature map; wherein, the output of the last layer of the second convolutional neural network is the product feature vector, and the first layer of the second convolutional neural network
  • the input of the layer is the liquid chromatogram of the reaction solution at multiple predetermined time points within the predetermined time period.
  • the responsiveness estimation module is further used to: calculate the responsiveness estimate of the product feature vector relative to the temperature-flow rate correlation feature vector using the following formula to obtain the Classification feature matrix; wherein, the formula is:
  • F represents the product feature vector
  • T represents the classification feature matrix
  • R represents the temperature-flow rate correlation feature vector
  • the batching control result generation module processes the classification feature matrix with the following formula to generate a classification result, wherein the formula is :softmax ⁇ (W n ,B n ):...:(W 1 ,B 1 )
  • a batching method of an automatic batching system for preparing lithium hexafluorophosphate which includes:
  • the classification feature matrix is passed through a classifier to obtain a classification result, which is used to indicate that the gas flow rate value of phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • the temperature-flow rate correlation matrix is corrected to obtain the corrected temperature-flow rate correlation matrix-flow rate correlation matrix , including: calculating the transpose matrix of the temperature-flow velocity correlation matrix; performing convolution processing on the temperature-flow velocity correlation matrix to obtain the first convolution feature matrix; calculating the transpose matrix sum of the temperature-flow velocity correlation matrix
  • the first convolution feature matrix is added by position points to obtain a preliminary fusion feature matrix; the preliminary fusion feature matrix is convolved to obtain a second convolution feature matrix; each position in the temperature-flow velocity correlation matrix is The two-dimensional position coordinates are mapped to one-dimensional values to obtain the position information matrix; and, the second convolution feature matrix and the position information matrix are calculated and added by position points to obtain the corrected temperature-flow rate correlation matrix.
  • mapping the two-dimensional position coordinates of each position in the temperature-flow rate correlation matrix to a one-dimensional value to obtain the position information matrix includes: using the following formula to The two-dimensional position coordinates of each position in the temperature-flow velocity correlation matrix are mapped into one-dimensional values to obtain the position information matrix;
  • M represents the temperature-flow velocity correlation matrix
  • M 1 represents the position information matrix
  • P M represents the (x, y) coordinate matrix of the temperature-flow velocity correlation matrix, Used to map two-dimensional position coordinates to one-dimensional values.
  • the corrected temperature-flow rate correlation matrix is passed through the first convolutional neural network as a filter to obtain the temperature-flow rate correlation feature vector, including: Each layer of the first convolutional neural network of the filter performs a convolution process on the input data in the forward pass of the layer to obtain a convolution feature map; performs a convolution process on the convolution feature map based on local features.
  • the output is the temperature-flow rate correlation feature vector
  • the input of the first layer of the first convolutional neural network serving as a filter is the corrected temperature-flow rate correlation matrix.
  • the liquid chromatograms of the reaction liquid at multiple predetermined time points within the predetermined time period are passed through the second convolutional neural network using a three-dimensional convolution kernel to obtain
  • the product feature vector includes: the second convolutional neural network using a three-dimensional convolution kernel performs three-dimensional convolution processing on the input data in the forward pass of the layer: based on the three-dimensional convolution kernel to perform three-dimensional convolution processing on the input data.
  • calculating the response estimate of the product feature vector relative to the temperature-flow rate correlation feature vector to obtain a classification feature matrix includes: calculating the product with the following formula The responsiveness of the feature vector relative to the temperature-flow velocity correlation feature vector is estimated to obtain the classification feature matrix; wherein, the formula is:
  • F represents the product feature vector
  • T represents the classification feature matrix
  • R represents the temperature-flow rate correlation feature vector
  • passing the classification feature matrix through a classifier to obtain a classification result includes: using the classifier to process the classification feature matrix with the following formula to generate a classification
  • the formula is: softmaxP(W n ,B n ):...:(W 1 ,B 1 )
  • the automatic batching system and batching method provided by this application for the preparation of lithium hexafluorophosphate uses artificial intelligence control methods and a convolutional neural network model to control pentafluorination at multiple predetermined time points.
  • the correlation matrix between the gas flow rate value of the phosphorus gas and the reaction temperature value and the liquid chromatogram of the reaction liquid are used for implicit feature extraction, so that in the preparation process of the lithium hexafluorophosphate product, the phosphorus pentafluoride gas in the reaction process
  • the introduced flow rate is controlled in real-time and dynamically with the reaction temperature change, thereby improving the preparation efficiency of the lithium hexafluorophosphate product.
  • Figure 1 is an application scenario diagram of an automatic batching system for preparing lithium hexafluorophosphate according to an embodiment of the present application.
  • Figure 2 is a block diagram of an automatic batching system for preparing lithium hexafluorophosphate according to an embodiment of the present application.
  • Figure 3 is a block diagram of a calibration module in an automatic batching system for preparing lithium hexafluorophosphate according to an embodiment of the present application.
  • Figure 4 is a flow chart of a batching method of an automatic batching system for preparing lithium hexafluorophosphate according to an embodiment of the present application.
  • Figure 5 is a schematic structural diagram of a batching method of an automatic batching system for preparing lithium hexafluorophosphate according to an embodiment of the present application.
  • lithium hexafluorophosphate (LiPF6) has been selected as the electrolyte of lithium ion secondary batteries.
  • Lithium hexafluorophosphate is dissolved in certain non-aqueous organic solvents to form the electrolyte of lithium ion secondary batteries.
  • Lithium-ion secondary batteries are internationally recognized as an ideal chemical energy source. They are widely used in mobile phones, laptops and laptops due to their small size, large capacity, ability to be repeatedly charged and discharged 500 times, and only a 3% reduction in capacity. Cameras, etc., are also used in the electronics industry to make wafer dopants and catalysts for organic synthesis.
  • Lithium ions move through the electrolyte solution, with the result that in order to maintain battery performance, such as service life, the purity of the electrolyte in the electrolyte solution should be strictly controlled.
  • LiPF6 lithium hexafluorophosphate
  • an automatic batching system for the preparation of lithium hexafluorophosphate is expected to improve the preparation efficiency of lithium hexafluorophosphate products while simplifying the preparation process of lithium hexafluorophosphate.
  • Phosphorus fluoride gas specifically, here, the reaction temperature of the hexafluorophosphoric acid and fuming sulfuric acid is 5°C-32°C, and the reaction time is 2-4 hours; (3) Dissolve high-purity lithium fluoride in anhydrous In the hydrogen fluoride solution, an anhydrous hydrogen fluoride solution of lithium fluoride is formed; (4) After cooling the phosphorus pentafluoride gas at -40°C, it is introduced into an anhydrous hydrogen fluoride solution containing lithium fluoride, and after reaction, crystallization, Separate and dry to obtain pure lithium hexafluorophosphate product;
  • the inventor of the present application found that in the above-mentioned preparation process of the lithium hexafluorophosphate product, the synergy of the flow rate of the phosphorus pentafluoride gas introduction and the control strategy of the reaction temperature can effectively improve the preparation efficiency of the lithium hexafluorophosphate product. Therefore, in the technical solution of the present application, it is expected that in the preparation step of the lithium hexafluorophosphate product, the flow rate of the phosphorus pentafluoride gas introduced during the reaction process can be controlled in real time and dynamically with the change of the reaction temperature, thereby improving the efficiency of the lithium hexafluorophosphate product. Product preparation efficiency.
  • a flow rate sensor is used to obtain the gas flow rate value of phosphorus pentafluoride gas flowing into an anhydrous hydrogen fluoride solution containing lithium fluoride at multiple predetermined time points within a predetermined time period. It should be understood that if you want to control the flow rate of the phosphorus pentafluoride gas introduced in real-time and dynamically with the change of the reaction temperature during the preparation process of the lithium hexafluorophosphate product, you also need to obtain the temperature data within the predetermined time period through a temperature sensor.
  • the reaction temperature value at a predetermined time point is used, and the liquid chromatogram of the reaction liquid at multiple predetermined time points within the predetermined time period is collected by a liquid chromatograph as the result characteristic of the reaction. Furthermore, in this way, the gas flow rate value of the phosphorus pentafluoride gas can be controlled based on the change characteristic information of the reaction temperature and the dynamic change characteristics of the liquid chromatogram of the reaction liquid.
  • the gas flow rate values and reaction temperature values of phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period are arranged into temperature input vectors and flow rate input vectors according to the time dimension, respectively. Integrate the gas flow rate value and reaction temperature value information of the phosphorus pentafluoride gas at each time point.
  • the temperature-flow velocity correlation feature vector can have a good expression for the local temperature-flow velocity correlation of the temperature-flow velocity correlation matrix However, it is expected to further enhance the ability of the temperature-flow rate correlation feature vector to express the global temperature-flow rate correlation of the temperature-flow rate correlation matrix.
  • the temperature-flow velocity correlation matrix M is subjected to position proposal local reasoning transformation:
  • M represents the temperature-flow velocity correlation matrix
  • Cov 1 () and Cov 2 () are both single convolution layers, Used to map two-dimensional position coordinates to one-dimensional values
  • P M represents the (x, y) coordinate matrix of matrix M
  • ⁇ represents the position-wise dot multiplication.
  • the position information can be used as a proposal to pass the convolutional layer
  • the bias of the local perceptual field and transposed structure is used to reason about the global scene semantics, thereby comprehensively integrating the captured local temperature-flow velocity correlation semantics and further deriving the global temperature-flow velocity semantics.
  • the ability of the temperature-flow velocity correlation feature vector to express the global temperature-flow velocity correlation of the temperature-flow velocity correlation matrix can be strengthened, thereby improving the accuracy of classification. sex.
  • the corrected temperature-flow rate correlation matrix can be further processed through the first convolutional neural network as a filter to extract the corrected temperature-flow rate correlation matrix corresponding to the pentafluoride
  • the high-dimensional implicit correlation feature between the gas flow rate value of the phosphorus gas and the reaction temperature value is used to obtain a temperature-flow rate correlation feature vector.
  • the liquid chromatograms of the reaction liquid at multiple predetermined time points within the predetermined time period are passed through the second convolutional neural network using a three-dimensional convolution kernel. Get the product feature vector.
  • the classification feature matrix can be passed through the classifier to obtain a classification result indicating that the gas flow rate value of the phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • this application proposes an automatic batching system for the preparation of lithium hexafluorophosphate, which includes: a gas batching data acquisition module, used to obtain the phosphorus pentafluoride gas at multiple predetermined time points within a predetermined time period and pass it into the container.
  • the gas flow rate value of the anhydrous hydrogen fluoride solution of lithium fluoride ; a reaction data acquisition module, used to obtain the reaction temperature values at multiple predetermined time points within the predetermined time period and the liquid chromatogram of the reaction liquid; a discrete data vectorization module , used to arrange the gas flow rate values and reaction temperature values of the phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period into temperature input vectors and flow rate input vectors according to the time dimension; the data level correlation module is used to Calculate the product between the transposed vector of the temperature input vector and the flow rate input vector to obtain a temperature-flow rate correlation matrix; a correction module for, based on the position information of the temperature-flow rate correlation matrix, correct the temperature-flow rate correlation matrix The flow velocity correlation matrix is corrected to obtain the corrected temperature-flow velocity correlation matrix-flow velocity correlation matrix; the feature extraction module is used to pass the corrected temperature-flow velocity correlation matrix through the first convolutional neural network as a filter to obtain the temperature-flow velocity correlation matrix.
  • Flow rate related feature vector Flow rate related feature vector
  • product data encoding module used to pass the liquid chromatogram of the reaction liquid at multiple predetermined time points within the predetermined time period through a second convolutional neural network using a three-dimensional convolution kernel to obtain the product feature vector ;
  • Responsiveness estimation module used to calculate the responsiveness estimation of the product feature vector relative to the temperature-flow rate related feature vector to obtain a classification feature matrix;
  • a batching control result generation module used to convert the classification feature matrix
  • the classification result is obtained through the classifier, and the classification result is used to indicate that the gas flow rate value of the phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • FIG 1 illustrates an application scenario diagram of an automatic batching system for preparing lithium hexafluorophosphate according to an embodiment of the present application.
  • each sensor for example, the flow rate sensor T1 and the temperature sensor T2 illustrated in Figure 1
  • the phosphorus gas for example, P as shown in Figure 1
  • the anhydrous hydrogen fluoride solution for example, N as shown in Figure 1 containing lithium fluoride
  • a liquid chromatograph for example, L as shown in Figure 1 collects liquid chromatograms of the reaction liquid at multiple predetermined time points within the predetermined time period.
  • the obtained gas flow rate values and reaction temperature values at multiple predetermined time points within the predetermined time period and the liquid chromatogram of the reaction liquid are input into a server deployed with an automatic batching algorithm for the preparation of lithium hexafluorophosphate (for example, , the cloud server S) as shown in Figure 1, wherein the server can use an automatic batching algorithm for the preparation of lithium hexafluorophosphate to calculate gas flow rate values and reaction temperature values at multiple predetermined time points within the predetermined time period, as well as the
  • the liquid chromatogram of the reaction liquid is processed to generate a classification result indicating whether the gas flow rate value of the phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • FIG. 2 illustrates a block diagram of an automatic batching system for the preparation of lithium hexafluorophosphate according to an embodiment of the present application.
  • the automatic batching system 200 for the preparation of lithium hexafluorophosphate according to the embodiment of the present application includes: a gas batching data acquisition module 210, used to obtain the gas flow of phosphorus pentafluoride at multiple predetermined time points within a predetermined time period.
  • the reaction data acquisition module 220 is used to obtain the reaction temperature values at multiple predetermined time points within the predetermined time period and the liquid chromatogram of the reaction solution;
  • the discrete data vectorization module 230 is used to arrange the gas flow rate values and reaction temperature values of the phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period into temperature input vectors and flow rate input vectors according to the time dimension;
  • data The stage correlation module 240 is used to calculate the product between the transpose vector of the temperature input vector and the flow speed input vector to obtain the temperature-flow speed correlation matrix;
  • the correction module 250 is used to calculate the temperature-flow speed correlation matrix based on the temperature-flow speed correlation matrix.
  • the feature extraction module 260 is used to pass the corrected temperature-flow rate correlation matrix through the third filter as a filter.
  • a convolutional neural network to obtain the temperature-flow rate correlation feature vector; the product data encoding module 270 is used to pass the liquid chromatograms of the reaction liquid at multiple predetermined time points within the predetermined time period through the first three-dimensional convolution kernel.
  • the generation module 290 is used to pass the classification feature matrix through a classifier to obtain a classification result.
  • the classification result is used to indicate that the gas flow rate value of phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • the gas batching data collection module 210 and the reaction data collection module 220 are used to obtain the phosphorus pentafluoride gas at multiple predetermined time points within a predetermined time period and pass it into the container.
  • the gas flow rate value of the anhydrous hydrogen fluoride solution of lithium fluoride is obtained, and the reaction temperature values at multiple predetermined time points within the predetermined time period and the liquid chromatogram of the reaction liquid are obtained.
  • the synergy between the flow rate of the phosphorus pentafluoride gas introduced and the reaction temperature control strategy can effectively improve the preparation efficiency of the lithium hexafluorophosphate product.
  • the flow rate of the phosphorus pentafluoride gas introduced during the reaction process can be controlled in real time and dynamically with the change of the reaction temperature, thereby improving the efficiency of the lithium hexafluorophosphate product.
  • the flow rate sensor is used to obtain the gas of phosphorus pentafluoride gas flowing into the anhydrous hydrogen fluoride solution containing lithium fluoride at multiple predetermined time points within a predetermined time period.
  • Flow rate value It should be understood that if you want to control the flow rate of the phosphorus pentafluoride gas introduced in real-time and dynamically with the change of the reaction temperature during the preparation process of the lithium hexafluorophosphate product, you also need to obtain the temperature data within the predetermined time period through a temperature sensor.
  • the reaction temperature value at a predetermined time point is used, and the liquid chromatogram of the reaction liquid at multiple predetermined time points within the predetermined time period is collected by a liquid chromatograph as the result characteristic of the reaction. Furthermore, in this way, the gas flow rate value of the phosphorus pentafluoride gas can be controlled based on the change characteristic information of the reaction temperature and the dynamic change characteristics of the liquid chromatogram of the reaction liquid.
  • the discrete data vectorization module 230 and the data level correlation module 240 are used to calculate the gas flow rate of phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period.
  • the value and reaction temperature value are arranged into a temperature input vector and a flow rate input vector according to the time dimension respectively, and the product between the transposed vector of the temperature input vector and the flow rate input vector is calculated to obtain the temperature-flow rate correlation matrix. It should be understood that since there is an implicit correlation between the gas flow rate value of the phosphorus pentafluoride gas and the reaction temperature value, the pentafluorination is performed in order to extract this implicit correlation feature.
  • the gas flow rate values and reaction temperature values of phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period are arranged respectively according to the time dimension as
  • the temperature input vector and the flow rate input vector are used to respectively integrate the gas flow rate value and reaction temperature value information of the phosphorus pentafluoride gas at each time point.
  • calculate the transpose vector between the temperature input vector and the flow rate input vector are calculated. Multiply to obtain the temperature-flow rate correlation matrix.
  • the correction module 250 is used to correct the temperature-flow rate correlation matrix based on the position information of the temperature-flow rate correlation matrix to obtain a corrected temperature-flow rate correlation matrix - Flow velocity correlation matrix.
  • the temperature-flow velocity correlation feature vector can have a good expression for the local temperature-flow velocity correlation of the temperature-flow velocity correlation matrix
  • the correction module includes: first, calculating the transpose matrix of the temperature-flow rate correlation matrix. Next, the temperature-flow velocity correlation matrix is convolved to obtain a first convolution feature matrix. Then, calculate the transpose matrix of the temperature-flow velocity correlation matrix and the location points of the first convolution feature matrix to obtain a preliminary fusion feature matrix. Next, the preliminary fusion feature matrix is subjected to convolution processing to obtain a second convolution feature matrix. Then, the two-dimensional position coordinates of each position in the temperature-flow velocity correlation matrix are mapped into one-dimensional numerical values to obtain a position information matrix.
  • the corrected temperature-flow rate correlation matrix is obtained by calculating the position-wise points of the second convolution feature matrix and the position information matrix. That is, in a specific example, based on the position information of the temperature-flow velocity correlation matrix, the temperature-flow velocity correlation matrix is corrected with the following formula to obtain the corrected temperature-flow velocity correlation matrix-flow velocity correlation matrix
  • M represents the temperature-flow velocity correlation matrix
  • Cov 1 () and Cov 2 () are both single convolution layers
  • P M represents the (x, y) coordinate matrix of the temperature-flow velocity correlation matrix
  • ⁇ represents the position-wise dot multiplication.
  • the position information can be used as a proposal to pass
  • the local perceptual field and the bias of the transposed structure of the convolutional layer are used to reason about the global scene semantics, thereby comprehensively integrating the captured local temperature-flow velocity correlation semantics and further deriving the global temperature-flow velocity semantics.
  • the ability of the temperature-flow velocity correlation feature vector to express the global temperature-flow velocity correlation of the temperature-flow velocity correlation matrix can be strengthened, thereby improving classification accuracy.
  • Figure 3 illustrates a block diagram of a calibration module in an automatic batching system for lithium hexafluorophosphate preparation according to an embodiment of the present application.
  • the correction module 250 includes: a transposition unit 251, used to calculate the transpose matrix of the temperature-flow rate correlation matrix; a first convolution unit 252, used to calculate the temperature-flow rate correlation The matrix is subjected to convolution processing to obtain the first convolution feature matrix; the fusion unit 253 is used to calculate the transpose matrix of the temperature-flow velocity correlation matrix and the position-wise points of the first convolution feature matrix to obtain preliminary fusion features.
  • the second convolution unit 254 is used to perform convolution processing on the preliminary fusion feature matrix to obtain the second convolution feature matrix
  • the position information extraction unit 255 is used to convert each position in the temperature-flow rate correlation matrix The two-dimensional position coordinates are mapped to one-dimensional values to obtain the position information matrix
  • the re-fusion unit 256 is used to calculate the second convolution feature matrix and the position information matrix by position points to obtain the corrected Temperature-flow rate correlation matrix.
  • the feature extraction module 260 is used to pass the corrected temperature-flow rate correlation matrix through the first convolutional neural network as a filter to obtain a temperature-flow rate correlation feature vector. That is, in the technical solution of the present application, after obtaining the corrected temperature-flow rate correlation matrix, the corrected temperature-flow rate correlation matrix is further processed through the first convolutional neural network as a filter, To extract the global high-dimensional implicit correlation feature between the gas flow rate value corresponding to the phosphorus pentafluoride gas and the reaction temperature value in the corrected temperature-flow rate correlation matrix, thereby obtaining the temperature-flow rate correlation Feature vector.
  • the feature extraction module is further configured to perform: each layer of the first convolutional neural network as a filter on the input data in the forward pass of the layer:
  • the input data is subjected to convolution processing to obtain a convolution feature map;
  • the convolution feature map is subjected to mean pooling based on a local feature matrix to obtain a pooled feature map; and, the pooled feature map is subjected to nonlinear activation to obtain Obtain activation feature map;
  • the output of the last layer of the first convolutional neural network as a filter is the temperature-flow rate correlation feature vector
  • the first layer of the first convolutional neural network as a filter is The input of the layer is the corrected temperature-flow rate correlation matrix.
  • the product data encoding module 270 is used to pass the liquid chromatograms of the reaction liquid at multiple predetermined time points within the predetermined time period through the second volume using a three-dimensional convolution kernel. Accumulate the neural network to obtain the product feature vector.
  • the liquid chromatograms of the reaction liquid at multiple predetermined time points within the predetermined time period are processed by using three-dimensional convolution
  • the second convolutional neural network of the kernel is used to obtain the product feature vector.
  • the product data encoding module is further configured to: the second convolutional neural network using a three-dimensional convolution kernel separately performs on the input data in the forward pass of the layer: based on The three-dimensional convolution kernel performs three-dimensional convolution processing on the input data to obtain a convolution feature map; performs mean pooling processing based on a local feature matrix on the convolution feature map to obtain a pooling feature map; and The pooled feature map performs nonlinear activation to obtain an activation feature map; wherein, the output of the last layer of the second convolutional neural network is the product feature vector, and the first layer of the second convolutional neural network
  • the input is the liquid chromatogram of the reaction solution at multiple predetermined time points within the predetermined time period.
  • the responsiveness estimation module 280 and the batching control result generation module 290 are used to calculate the responsiveness estimate of the product feature vector relative to the temperature-flow rate correlation feature vector to A classification feature matrix is obtained, and the classification feature matrix is passed through a classifier to obtain a classification result.
  • the classification result is used to indicate that the gas flow rate value of the phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • the classification feature matrix can be passed through the classifier to obtain a classification result indicating that the gas flow rate value of the phosphorus pentafluoride gas at the current time point should be increased or decreased. That is, in a specific example, the classifier processes the classification feature matrix with the following formula to generate a classification result, where the formula is: softmax ⁇ (W n ,B n ):...:(W 1 ,B 1 )
  • the responsiveness estimation module is further configured to: calculate the responsiveness estimate of the product feature vector relative to the temperature-flow rate correlation feature vector using the following formula to obtain the classification Characteristic matrix; where the formula is:
  • F represents the product feature vector
  • T represents the classification feature matrix
  • R represents the temperature-flow rate correlation feature vector
  • the automatic batching system 200 for the preparation of lithium hexafluorophosphate based on the embodiment of the present application has been clarified, which uses an artificial intelligence control method and a convolutional neural network model to control phosphorus pentafluoride at multiple predetermined time points.
  • the correlation matrix between the gas flow rate value and the reaction temperature value of the gas and the liquid chromatogram of the reaction liquid are used for implicit feature extraction, so as to introduce the phosphorus pentafluoride gas during the reaction process during the preparation process of the lithium hexafluorophosphate product.
  • the flow rate is controlled in real-time and dynamically with the reaction temperature change, thereby improving the preparation efficiency of the lithium hexafluorophosphate product.
  • the automatic batching system 200 for the preparation of lithium hexafluorophosphate according to the embodiment of the present application can be implemented in various terminal devices, such as servers for automatic batching algorithms for the preparation of lithium hexafluorophosphate, etc.
  • the automatic batching system 200 for preparing lithium hexafluorophosphate according to an embodiment of the present application can be integrated into a terminal device as a software module and/or a hardware module.
  • the automatic batching system 200 for the preparation of lithium hexafluorophosphate can be a software module in the operating system of the terminal equipment, or can be an application developed for the terminal equipment; of course, the automatic batching system 200 for the preparation of lithium hexafluorophosphate
  • the batching system 200 can also be one of many hardware modules of the terminal device.
  • the automatic batching system 200 for the preparation of lithium hexafluorophosphate and the terminal device can also be separate devices, and the automatic batching system 200 for the preparation of lithium hexafluorophosphate can be connected through wired and/or wireless networks. to the terminal device and transmit interactive information according to the agreed data format.
  • Figure 4 illustrates a flow chart of a batching method of an automatic batching system for the preparation of lithium hexafluorophosphate.
  • the batching method of the automatic batching system for preparing lithium hexafluorophosphate according to the embodiment of the present application includes the steps: S110, obtaining the phosphorus pentafluoride gas at multiple predetermined time points within a predetermined time period and passing it into the container.
  • the gas flow rate values and reaction temperature values of the phosphorus pentafluoride gas at a predetermined time point are arranged into a temperature input vector and a flow rate input vector according to the time dimension respectively;
  • S140 calculate the transpose vector of the temperature input vector and the flow rate input vector to obtain the temperature-flow velocity correlation matrix;
  • S170 pass the temperature-flow rate correlation matrix of the reaction liquid at
  • the liquid chromatogram is passed through a second convolutional neural network using a three-dimensional convolution kernel to obtain a product feature vector; S180, calculate the response estimate of the product feature vector relative to the temperature-flow rate correlation feature vector to obtain a classification feature matrix ; and, S190, pass the classification feature matrix through a classifier to obtain a classification result, which is used to indicate that the gas flow rate value of phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • Figure 5 illustrates a schematic diagram of the architecture of a batching method of an automatic batching system for preparing lithium hexafluorophosphate according to an embodiment of the present application.
  • the gas flow rates of the phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period are obtained Values (e.g., P1 as shown in Figure 5) and reaction temperature values (e.g., P2 as shown in Figure 5) are respectively arranged according to the time dimension into temperature input vectors (e.g., V1 as shown in Figure 5) and the flow rate input vector (for example, V2 as illustrated in Figure 5); then, the product between the transpose vector of the temperature input vector and the flow rate input vector is calculated to obtain the temperature-flow rate correlation matrix (for example, as M1 shown in Figure 5); then, based on the position information of the temperature-flow velocity correlation matrix, the temperature
  • the temperature-flow rate correlation matrix for example, as M1 shown in Figure 5
  • steps S110 and S120 the gas flow rate values of the phosphorus pentafluoride gas flowing into the anhydrous hydrogen fluoride solution containing lithium fluoride at multiple predetermined time points within the predetermined time period are obtained, and the gas flow rate values are obtained.
  • Reaction temperature values and liquid chromatograms of the reaction liquid at multiple predetermined time points within a predetermined time period are obtained. It should be understood that during the preparation process of the lithium hexafluorophosphate product, the synergy of the flow rate of the phosphorus pentafluoride gas introduced and the control strategy of the reaction temperature can effectively improve the preparation efficiency of the lithium hexafluorophosphate product.
  • the flow rate of the phosphorus pentafluoride gas introduced during the reaction process can be controlled in real time and dynamically with the change of the reaction temperature, thereby improving the efficiency of the lithium hexafluorophosphate product.
  • the flow rate sensor is used to obtain the gas of phosphorus pentafluoride gas flowing into the anhydrous hydrogen fluoride solution containing lithium fluoride at multiple predetermined time points within a predetermined time period.
  • Flow rate value It should be understood that if you want to control the flow rate of the phosphorus pentafluoride gas introduced in real-time and dynamically with the change of the reaction temperature during the preparation process of the lithium hexafluorophosphate product, you also need to obtain the temperature data within the predetermined time period through a temperature sensor.
  • the reaction temperature value at a predetermined time point is used, and the liquid chromatogram of the reaction liquid at multiple predetermined time points within the predetermined time period is collected by a liquid chromatograph as the result characteristic of the reaction. Furthermore, in this way, the gas flow rate value of the phosphorus pentafluoride gas can be controlled based on the change characteristic information of the reaction temperature and the dynamic change characteristics of the liquid chromatogram of the reaction liquid.
  • the gas flow rate values and reaction temperature values of the phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period are arranged into a temperature input vector and a flow rate input according to the time dimension, respectively. vector, and calculate the product between the transposed vector of the temperature input vector and the flow rate input vector to obtain the temperature-flow rate correlation matrix. It should be understood that since there is an implicit correlation between the gas flow rate value of the phosphorus pentafluoride gas and the reaction temperature value, the pentafluorination is performed in order to extract this implicit correlation feature.
  • the gas flow rate values and reaction temperature values of phosphorus pentafluoride gas at multiple predetermined time points within the predetermined time period are arranged respectively according to the time dimension as
  • the temperature input vector and the flow rate input vector are used to respectively integrate the gas flow rate value and reaction temperature value information of the phosphorus pentafluoride gas at each time point.
  • calculate the transpose vector between the temperature input vector and the flow rate input vector are calculated. Multiply to obtain the temperature-flow rate correlation matrix.
  • the temperature-flow velocity correlation matrix is corrected to obtain a corrected temperature-flow velocity correlation matrix-flow velocity correlation matrix, and
  • the corrected temperature-flow rate correlation matrix is passed through the first convolutional neural network as a filter to obtain the temperature-flow rate correlation feature vector.
  • the first convolutional neural network as a filter extracts local correlation features, the temperature-flow velocity correlation feature vector can have a good expression for the local temperature-flow velocity correlation of the temperature-flow velocity correlation matrix However, it is expected to further enhance the ability of the temperature-flow rate correlation feature vector to express the global temperature-flow rate correlation of the temperature-flow rate correlation matrix.
  • the temperature-flow rate correlation matrix M is further subjected to position proposal local reasoning transformation. Then, the corrected temperature-flow rate correlation matrix is processed through the first convolutional neural network as a filter to extract the corrected temperature-flow rate correlation matrix corresponding to the phosphorus pentafluoride gas. The global high-dimensional implicit correlation feature between the gas flow rate value and the reaction temperature value is obtained, thereby obtaining a temperature-flow rate correlation feature vector.
  • the position information can be used as a proposal to pass
  • the local perceptual field and the bias of the transposed structure of the convolutional layer are used to reason about the global scene semantics, thereby comprehensively integrating the captured local temperature-flow velocity correlation semantics and further deriving the global temperature-flow velocity semantics.
  • the ability of the temperature-flow velocity correlation feature vector to express the global temperature-flow velocity correlation of the temperature-flow velocity correlation matrix can be strengthened, thereby improving classification accuracy.
  • step S170 the liquid chromatograms of the reaction solution at multiple predetermined time points within the predetermined time period are passed through a second convolutional neural network using a three-dimensional convolution kernel to obtain a product feature vector.
  • a second convolutional neural network of the kernel is used to obtain the product feature vector.
  • the responsiveness estimate of the product feature vector relative to the temperature-flow rate correlation feature vector is calculated to obtain a classification feature matrix, and the classification feature matrix is passed through a classifier to obtain Classification results, the classification results are used to indicate that the gas flow rate value of phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • the classification feature matrix is passed through a classifier to obtain Classification results, the classification results are used to indicate that the gas flow rate value of phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • the classification feature matrix can be passed through the classifier to obtain a classification result indicating that the gas flow rate value of the phosphorus pentafluoride gas at the current time point should be increased or decreased.
  • the batching method of the automatic batching system for the preparation of lithium hexafluorophosphate based on the embodiments of the present application has been clarified, which uses an artificial intelligence control method and a convolutional neural network model to control the concentration of pentafluoride at multiple predetermined time points.
  • the correlation matrix between the gas flow rate value of the phosphorus gas and the reaction temperature value and the liquid chromatogram of the reaction liquid are implicitly extracted to determine the phosphorus pentafluoride in the reaction process during the preparation process of the lithium hexafluorophosphate product.
  • the flow rate of gas introduction is controlled in real-time and dynamically with the reaction temperature change, thereby improving the preparation efficiency of the lithium hexafluorophosphate product.
  • each component or each step can be decomposed and/or recombined. These decompositions and/or recombinations shall be considered equivalent versions of this application.

Landscapes

  • Chemical & Material Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Computational Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Organic Chemistry (AREA)
  • Data Mining & Analysis (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Materials Engineering (AREA)
  • Inorganic Chemistry (AREA)
  • Algebra (AREA)
  • Secondary Cells (AREA)

Abstract

本申请涉及智能制造的领域，其具体地公开了一种用于六氟磷酸锂制备的自动配料系统及其配料方法，其通过使用人工智能的控制方法，以卷积神经网络模型来对于多个预定时间点的五氟化磷气体的气体流速值与反应温度值的关联矩阵以及反应液的液相色谱图进行隐含的特征提取，以在对于所述六氟磷酸锂产品的制备过程中，对于反应过程中的五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，进而提高所述六氟磷酸锂产品的制备效率。

Description

用于六氟磷酸锂制备的自动配料系统及其配料方法 技术领域
本发明涉及智能制造的领域，且更为具体地，涉及一种用于六氟磷酸锂制备的自动配料系统及其配料方法。
背景技术
近年来，六氟磷酸锂(LiPF6)被选定为锂离子二次电池的电解质，六氟磷酸锂被溶于某些非水有机溶剂中形成锂离子二次电池的电解液。锂离子二次电池是当今国际公认的理想化学能源，其以体积小、电容量大、能反复充放电500次、电容量只降低3％的优势，被广泛用于移动电话、手提电脑和手提摄像机等，还被用于电子工业制作晶片的掺杂剂和有机合成的催化剂。
当锂离子二次电池被充电时，锂离子进入并附着在阴极的空穴处；而锂离子二次电池被放电时，附着在阴极上的锂离子重新回到阳极，在这种情况下，锂离子通过电解质溶液运动，结果为了维持电池的性能，例如使用寿命，应对在电解液中电解质的纯度进行严格控制。目前，由于制备六氟磷酸锂(LiPF6)的生产环节多、工艺复杂、设备要求较高，导致其生产成本居高不下。
因此，期待一种用于六氟磷酸锂制备的自动配料系统，以在简化六氟磷酸锂的制备工艺的同时，提高六氟磷酸锂产品的制备效率。
发明内容
为了解决上述技术问题，提出了本申请。本申请的实施例提供了一种用于六氟磷酸锂制备的自动配料系统及其配料方法，其通过使用人工智能的控制方法，以卷积神经网络模型来对于多个预定时间点的五氟化磷气体的气体流速值与反应温度值的关联矩阵以及反应液的液相色谱图进行隐含的特征提取，以在对于所述六氟磷酸锂产品的制备过程中，对于反应过程中的五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，进而提高所述六氟磷酸锂产品的制备效率。
根据本申请的一个方面，提供了一种用于六氟磷酸锂制备的自动配料系统，其包括：
气体配料数据采集模块，用于获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值；反应数据采集模块，用于获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图；离散数据向量化模块，用于将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量；
数据级关联模块，用于计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵；校正模块，用于基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵；特征提取模块，用于将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量；产物数据编码模块，用于将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量；响应性估计模块，用于计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵；以及
配料控制结果生成模块，用于将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。
在上述用于六氟磷酸锂制备的自动配料系统中，所述校正模块，包括：转置单元，用于计算所述温度-流速关联矩阵的转置矩阵；第一卷积单元，用于对所述温度-流速关联矩阵进行卷积处理以得到第一卷积特征矩阵；融合单元，用于计算所述温度-流速关联矩阵的转置矩阵和所述第一卷积特征矩阵的按位置点加以得到初步融合特征矩阵；第二卷积单元，用于对所述初步融合特征矩阵进行卷积处理以得到第二卷积特征矩阵；位置信息提取单元，用于将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到位置信息矩阵；以及，再融合单元，用于计算所述第二卷积特征矩阵和所述位置信息矩阵的按位置点加以得到所述校正后温度-流速关联矩阵。
在上述用于六氟磷酸锂制备的自动配料系统中，所述位置信息提取单元，进一步用于：以如下公式将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到所述位置信息矩阵；其中，所述公式为：M 1＝Φ(P M)
其中，M表示所述温度-流速关联矩阵，M 1表示所述位置信息矩阵，P M表示所述温度-流速关联矩阵的(x,y)坐标矩阵，
Figure PCTCN2022120829-appb-000001
用于将二维位置坐标映射为一维数值。
在上述用于六氟磷酸锂制备的自动配料系统中，所述特征提取模块，进一步用于：所述作为过滤器的第一卷积神经网络的各层在层的正向传递中对输入数据分别进行：对输入数据进行卷积处理以得到卷积特征图；对所述卷积特征图进行基于局部特征矩阵的均值池化以得到池化特征图；以及，对所 述池化特征图进行非线性激活以得到激活特征图；其中，所述作为过滤器的第一卷积神经网络的最后一层的输出为所述温度-流速关联特征向量，所述作为过滤器的第一卷积神经网络的第一层的输入为所述校正后温度-流速关联矩阵。
在上述用于六氟磷酸锂制备的自动配料系统中，所述产物数据编码模块，进一步用于：所述使用三维卷积核的第二卷积神经网络在层的正向传递中对输入数据分别进行：基于所述三维卷积核对所述输入数据进行三维卷积处理以得到卷积特征图；对所述卷积特征图进行基于局部特征矩阵的均值池化处理以得到池化特征图；以及，对所述池化特征图进行非线性激活以得到激活特征图；其中，所述第二卷积神经网络的最后一层的输出为所述产物特征向量，所述第二卷积神经网络的第一层的输入为所述预定时间段内多个预定时间点的反应液的液相色谱图。
在上述用于六氟磷酸锂制备的自动配料系统中，所述响应性估计模块，进一步用于：以如下公式计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到所述分类特征矩阵；其中，所述公式为：
F＝T*R
其中F表示所述产物特征向量，T表示所述分类特征矩阵，R表示所述温度-流速关联特征向量。
在上述用于六氟磷酸锂制备的自动配料系统中，所述配料控制结果生成模块，进一步用于：所述分类器以如下公式对所述分类特征矩阵进行处理以生成分类结果，其中，所述公式为：softmax{(W n,B n):…:(W 1,B 1)|Project(F)}，其中Project(F)表示将所述分类特征矩阵投影为向量，W 1至W n为各层全连接层的权重矩阵，B 1至B n表示各层全连接层的偏置矩阵。
根据本申请的另一方面，一种用于六氟磷酸锂制备的自动配料系统的配料方法，其包括：
获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值；获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图；将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量；计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵；基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵；将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量；将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量；计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵；以及
将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。
在上述用于六氟磷酸锂制备的自动配料系统的配料方法中，基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵，包括：计算所述温度-流速关联矩阵的转置矩阵；对所述温度-流速关联矩阵进行卷积处理以得到第一卷积特征矩阵；计算所述温度-流速关联矩阵的转置矩阵和所述第一卷积特征矩阵的按位置点加以得到初步融合特征矩阵；对所述初步融合特征矩阵进行卷积处理以得到第二卷积特征矩阵；将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到位置信息矩阵；以及，计算所述第二卷积特征矩阵和所述位置信息矩阵的按位置点加以得到所述校正后温度-流速关联矩阵。
在上述用于六氟磷酸锂制备的自动配料系统的配料方法中，将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到位置信息矩阵，包括：以如下公式将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到所述位置信息矩阵；
其中，所述公式为：M 1＝Φ(P M)
其中，M表示所述温度-流速关联矩阵，M 1表示所述位置信息矩阵，P M表示所述温度-流速关联矩阵的(x,y)坐标矩阵，
Figure PCTCN2022120829-appb-000002
用于将二维位置坐标映射为一维数值。
在上述用于六氟磷酸锂制备的自动配料系统的配料方法中，将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量，包括：所述作为过滤器的第一卷积神经网络的各层在层的正向传递中对输入数据分别进行：对输入数据进行卷积处理以得到卷积特征图；对所述卷积特征图进行基于局部特征矩阵的均值池化以得到池化特征图；以及，对所述池化特征图进行非线性激活以得到激活特征图；其中，所述作为过滤器的第一卷积神经网络的最后一层的输出为所述温度-流速关联特征向量，所述作为过滤器的第一卷积神经网络的第一层的输入为所述校正后温度-流速关联矩阵。
在上述用于六氟磷酸锂制备的自动配料系统的配料方法中，将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量，包括：所述使 用三维卷积核的第二卷积神经网络在层的正向传递中对输入数据分别进行：基于所述三维卷积核对所述输入数据进行三维卷积处理以得到卷积特征图；对所述卷积特征图进行基于局部特征矩阵的均值池化处理以得到池化特征图；以及，对所述池化特征图进行非线性激活以得到激活特征图；其中，所述第二卷积神经网络的最后一层的输出为所述产物特征向量，所述第二卷积神经网络的第一层的输入为所述预定时间段内多个预定时间点的反应液的液相色谱图。
在上述用于六氟磷酸锂制备的自动配料系统的配料方法中，计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵，包括：以如下公式计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到所述分类特征矩阵；其中，所述公式为：
F＝T*R
其中F表示所述产物特征向量，T表示所述分类特征矩阵，R表示所述温度-流速关联特征向量。
在上述用于六氟磷酸锂制备的自动配料系统的配料方法中，将所述分类特征矩阵通过分类器以得到分类结果，包括：使用所述分类器以如下公式对所述分类特征矩阵进行处理以生成分类结果，其中，所述公式为：softmaxP(W n,B n):…:(W 1,B 1)|Project(F)}，其中Project(F)表示将所述分类特征矩阵投影为向量，W 1至W n为各层全连接层的权重矩阵，B 1至B n表示各层全连接层的偏置矩阵。
与现有技术相比，本申请提供的用于六氟磷酸锂制备的自动配料系统及其配料方法，其通过使用人工智能的控制方法，以卷积神经网络模型来对于多个预定时间点的五氟化磷气体的气体流速值与反应温度值的关联矩阵以及反应液的液相色谱图进行隐含的特征提取，以在对于所述六氟磷酸锂产品的制备过程中，对于反应过程中的五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，进而提高所述六氟磷酸锂产品的制备效率。
附图说明
通过结合附图对本申请实施例进行更详细的描述，本申请的上述以及其他目的、特征和优势将变得更加明显。附图用来提供对本申请实施例的进一步理解，并且构成说明书的一部分，与本申请实施例一起用于解释本申请，并不构成对本申请的限制。在附图中，相同的参考标号通常代表相同部件或步骤。
图1为根据本申请实施例的用于六氟磷酸锂制备的自动配料系统的应用场景图。
图2为根据本申请实施例的用于六氟磷酸锂制备的自动配料系统的框图。
图3为根据本申请实施例的用于六氟磷酸锂制备的自动配料系统中校正模块的框图。
图4为根据本申请实施例的用于六氟磷酸锂制备的自动配料系统的配料方法的流程图。
图5为根据本申请实施例的用于六氟磷酸锂制备的自动配料系统的配料方法的架构示意图。
具体实施方式
下面，将参考附图详细地描述根据本申请的示例实施例。显然，所描述的实施例仅仅是本申请的一部分实施例，而不是本申请的全部实施例，应理解，本申请不受这里描述的示例实施例的限制。
场景概述
如前所述，近年来，六氟磷酸锂(LiPF6)被选定为锂离子二次电池的电解质，六氟磷酸锂被溶于某些非水有机溶剂中形成锂离子二次电池的电解液。锂离子二次电池是当今国际公认的理想化学能源，其以体积小、电容量大、能反复充放电500次、电容量只降低3％的优势，被广泛用于移动电话、手提电脑和手提摄像机等，还被用于电子工业制作晶片的掺杂剂和有机合成的催化剂。
当锂离子二次电池被充电时，锂离子进入并附着在阴极的空穴处；而锂离子二次电池被放电时，附着在阴极上的锂离子重新回到阳极，在这种情况下，锂离子通过电解质溶液运动，结果为了维持电池的性能，例如使用寿命，应对在电解液中电解质的纯度进行严格控制。目前，由于制备六氟磷酸锂(LiPF6)的生产环节多、工艺复杂、设备要求较高，导致其生产成本居高不下。
因此，期待一种用于六氟磷酸锂制备的自动配料系统，以在简化六氟磷酸锂的制备工艺的同时，提高六氟磷酸锂产品的制备效率。
具体地，在本申请的技术方案中，六氟磷酸锂产品的制备过程如下：
(1)在惰性气体保护下，将无水氟化氢与浓磷酸反应制得六氟磷酸；(2)在冷却搅拌下，向步骤(1)制备的六氟磷酸中加入发烟硫酸，制得五氟化磷气体，特别地，这里，所述六氟磷酸与发烟硫酸的反应温度为5℃-32℃，反应时间为2-4小时；(3)将高纯氟化锂溶于无水氟化氢溶液中，形成氟化锂的无水氟化氢溶液；(4)将五氟化磷气体经过-40℃冷却之后，再导入到盛有氟化锂的无水氟化氢溶液中，经反应，结晶、分离、干燥得到纯净的六氟磷酸锂产品；
(5)将未反应的冷却后的五氟化磷气体继续通入到盛有氟化锂的无水氟化氢溶液中，继续反应，结晶、分离、干燥得到六氟磷酸锂成。
相应地，本申请发明人发现在上述对于所述六氟磷酸锂产品的制备过程中，五氟化磷气体导入的流速与反应温度的控制策略的协同可有效地提升所述六氟磷酸锂产品的制备效率。因此，在本申请的 技术方案中，期望在对于所述六氟磷酸锂产品的制备步骤中，对于反应过程中的五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，进而提高六氟磷酸锂产品的制备效率。
基于此，在本申请的技术方案中，首先，通过流速传感器获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值。应可以理解，若想在对于所述六氟磷酸锂产品的制备过程中对于五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，还需要通过温度传感器获取所述预定时间段内多个预定时间点的反应温度值，并且通过液相色谱仪采集所述预定时间段内多个预定时间点的反应液的液相色谱图来作为反应的结果依据特征。进而，这样就能够通过所述反应温度的变化特征信息以及所述反应液的液相色谱图的动态变化特征来进行所述五氟化磷气体的气体流速值调控。
然后，由于所述五氟化磷气体的气体流速值和所述反应温度值之间具有着隐含的关联关系，因此为了提取出这种隐含的关联特征来进行所述五氟化磷气体气体流速值的准确调控，进一步将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量，以分别整合所述各个时间点的五氟化磷气体的气体流速值和反应温度值信息。进一步地，为了构造所述五氟化磷气体的气体流速值和所述反应温度值之间的关联来进行特征挖掘，计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵。
应可以理解，由于所述作为过滤器的第一卷积神经网络提取局部关联特征，因此所述温度-流速关联特征向量可以对所述温度-流速关联矩阵的局部温度-流速关联具有良好的表达能力，但是，期望能够进一步强化所述温度-流速关联特征向量对所述温度-流速关联矩阵的全局温度-流速关联的表达能力。
因此，在本申请的技术方案中，对所述温度-流速关联矩阵M进行位置提议局部推理转化：
Figure PCTCN2022120829-appb-000003
Figure PCTCN2022120829-appb-000004
其中M表示所述温度-流速关联矩阵，Cov 1()和Cov 2()均为单个卷积层，
Figure PCTCN2022120829-appb-000005
用于将二维位置坐标映射为一维数值，P M表示矩阵M的(x,y)坐标矩阵，
Figure PCTCN2022120829-appb-000006
表示特征矩阵的按位置加法，⊙表示按位置点乘。
由于所述温度-流速关联矩阵的每个位置的特征值都表示特定的温度-流速数值对之间的关联，通过以上位置提议局部推理转化，可以使用位置信息作为提议，来通过卷积层的局部感知场和转置结构的偏置来对全局场景语义进行推理，从而全面融合所捕获的局部温度-流速关联语义并进一步衍生全局温度-流速语义。这样，再通过作为过滤器的第一卷积神经网络时，就可以强化所述温度-流速关联特征向量对所述温度-流速关联矩阵的全局温度-流速关联的表达能力，进而提高分类的准确性。
这样，就可以进一步将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络中进行处理，以提取出所述校正后温度-流速关联矩阵中对应于所述五氟化磷气体的气体流速值和所述反应温度值之间的高维隐含关联特征，从而得到温度-流速关联特征向量。
进一步地，对于所述预定时间段内多个预定时间点的反应液的液相色谱图，考虑到所述反应液的液相色谱图在时序上具有着动态性地关联变化特征，因此，为了充分地提取出这种动态变化特征信息来用于分类处理，将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量。
由于所述反应液的液相色谱图的动态特征数据与所述温度和流速关联的隐含特征数据之间的特征尺度不同，并且所述反应液的液相色谱图的动态特征在高维特征空间中可以看作是对于所述温度流速关联变化的响应性特征，因此为了更好地融合所述产物特征向量和所述温度-流速关联特征向量，进一步计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵。这样，就可以将所述分类特征矩阵通过分类器以得到用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小的分类结果。
基于此，本申请提出了一种用于六氟磷酸锂制备的自动配料系统，其包括：气体配料数据采集模块，用于获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值；反应数据采集模块，用于获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图；离散数据向量化模块，用于将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量；数据级关联模块，用于计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵；校正模块，用于基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵；特征提取模块，用于将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量；产物数据编码模块，用于将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产 物特征向量；响应性估计模块，用于计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵；以及，配料控制结果生成模块，用于将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。
图1图示了根据本申请实施例的用于六氟磷酸锂制备的自动配料系统的应用场景图。如图1所示，在该应用场景中，首先，通过各个传感器(例如，如图1中所示意的流速传感器T1和温度传感器T2)分别获取预定时间段内多个预定时间点的五氟化磷气体(例如，如图1中所示意的P)通入到盛有氟化锂的无水氟化氢溶液(例如，如图1中所示意的N)的气体流速值和反应温度值，并且通过液相色谱仪(例如，如图1中所示意的L)采集所述预定时间段内多个预定时间点的反应液的液相色谱图。然后，将获取的所述预定时间段内多个预定时间点的气体流速值和反应温度值以及所述反应液的液相色谱图输入至部署有用于六氟磷酸锂制备的自动配料算法的服务器中(例如，如图1中所示意的云服务器S)，其中，所述服务器能够以用于六氟磷酸锂制备的自动配料算法对所述预定时间段内多个预定时间点的气体流速值和反应温度值以及所述反应液的液相色谱图进行处理，以生成用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小的分类结果。
在介绍了本申请的基本原理之后，下面将参考附图来具体介绍本申请的各种非限制性实施例。
示例性系统
图2图示了根据本申请实施例的用于六氟磷酸锂制备的自动配料系统的框图。如图2所示，根据本申请实施例的用于六氟磷酸锂制备的自动配料系统200，包括：气体配料数据采集模块210，用于获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值；反应数据采集模块220，用于获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图；离散数据向量化模块230，用于将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量；数据级关联模块240，用于计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵；校正模块250，用于基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵；特征提取模块260，用于将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量；产物数据编码模块270，用于将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量；响应性估计模块280，用于计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵；以及，配料控制结果生成模块290，用于将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。
具体地，在本申请实施例中，所述气体配料数据采集模块210和所述反应数据采集模块220，用于获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值，并获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图。如前所述，由于在对于所述六氟磷酸锂产品的制备过程中，五氟化磷气体导入的流速与反应温度的控制策略的协同可有效地提升所述六氟磷酸锂产品的制备效率。因此，在本申请的技术方案中，期望在对于所述六氟磷酸锂产品的制备步骤中，对于反应过程中的五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，进而提高六氟磷酸锂产品的制备效率。
也就是，具体地，在本申请的技术方案中，首先，通过流速传感器获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值。应可以理解，若想在对于所述六氟磷酸锂产品的制备过程中对于五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，还需要通过温度传感器获取所述预定时间段内多个预定时间点的反应温度值，并且通过液相色谱仪采集所述预定时间段内多个预定时间点的反应液的液相色谱图来作为反应的结果依据特征。进而，这样就能够通过所述反应温度的变化特征信息以及所述反应液的液相色谱图的动态变化特征来进行所述五氟化磷气体的气体流速值调控。
具体地，在本申请实施例中，所述离散数据向量化模块230和所述数据级关联模块240，用于将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量，并计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵。应可以理解，由于所述五氟化磷气体的气体流速值和所述反应温度值之间具有着隐含的关联关系，因此为了提取出这种隐含的关联特征来进行所述五氟化磷气体气体流速值的准确调控，在本申请的技术方案中，进一步将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量，以分别整合所述各个时间点的五氟化磷气体的气体流速值和反应温度值信息。进一步地，为了构造所述五氟化磷气体的气体流速值和所述反应温度值之间的关联来进行特征挖掘，计算所述温度输入向量的转置向量 与所述流速输入向量之间的乘积以得到温度-流速关联矩阵。
具体地，在本申请实施例中，所述校正模块250，用于基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵。应可以理解，由于所述作为过滤器的第一卷积神经网络提取局部关联特征，因此所述温度-流速关联特征向量可以对所述温度-流速关联矩阵的局部温度-流速关联具有良好的表达能力，但是，期望能够进一步强化所述温度-流速关联特征向量对所述温度-流速关联矩阵的全局温度-流速关联的表达能力。因此，在本申请的技术方案中，进一步对所述温度-流速关联矩阵M进行位置提议局部推理转化。
更具体地，在本申请实施例中，所述校正模块，包括：首先，计算所述温度-流速关联矩阵的转置矩阵。接着，对所述温度-流速关联矩阵进行卷积处理以得到第一卷积特征矩阵。然后，计算所述温度-流速关联矩阵的转置矩阵和所述第一卷积特征矩阵的按位置点加以得到初步融合特征矩阵。接着，对所述初步融合特征矩阵进行卷积处理以得到第二卷积特征矩阵。然后，将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到位置信息矩阵。最后，计算所述第二卷积特征矩阵和所述位置信息矩阵的按位置点加以得到所述校正后温度-流速关联矩阵。也就是，在一个具体示例中，基于所述温度-流速关联矩阵的位置信息，以如下公式对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵
Figure PCTCN2022120829-appb-000007
其中，M表示所述温度-流速关联矩阵，Cov 1()和Cov 2()均为单个卷积层，P M表示所述温度-流速关联矩阵的(x,y)坐标矩阵，
Figure PCTCN2022120829-appb-000008
用于将二维位置坐标映射为一维数值，
Figure PCTCN2022120829-appb-000009
表示特征矩阵的按位置加法，⊙表示按位置点乘。应可以理解，由于所述温度-流速关联矩阵的每个位置的特征值都表示特定的温度-流速数值对之间的关联，通过以上位置提议局部推理转化，可以使用位置信息作为提议，来通过卷积层的局部感知场和转置结构的偏置来对全局场景语义进行推理，从而全面融合所捕获的局部温度-流速关联语义并进一步衍生全局温度-流速语义。这样，再通过所述作为过滤器的第一卷积神经网络时，就可以强化所述温度-流速关联特征向量对所述温度-流速关联矩阵的全局温度-流速关联的表达能力，进而提高分类的准确性。
图3图示了根据本申请实施例的用于六氟磷酸锂制备的自动配料系统中校正模块的框图。如图3所示，所述校正模块250，包括：转置单元251，用于计算所述温度-流速关联矩阵的转置矩阵；第一卷积单元252，用于对所述温度-流速关联矩阵进行卷积处理以得到第一卷积特征矩阵；融合单元253，用于计算所述温度-流速关联矩阵的转置矩阵和所述第一卷积特征矩阵的按位置点加以得到初步融合特征矩阵；第二卷积单元254，用于对所述初步融合特征矩阵进行卷积处理以得到第二卷积特征矩阵；位置信息提取单元255，用于将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到位置信息矩阵；以及，再融合单元256，用于计算所述第二卷积特征矩阵和所述位置信息矩阵的按位置点加以得到所述校正后温度-流速关联矩阵。
具体地，在本申请实施例中，所述特征提取模块260，用于将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量。也就是，在本申请的技术方案中，在得到所述校正后温度-流速关联矩阵后，进一步将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络中进行处理，以提取出所述校正后温度-流速关联矩阵中对应于所述五氟化磷气体的气体流速值和所述反应温度值之间的对应于全局的高维隐含关联特征，从而得到温度-流速关联特征向量。
更具体地，在本申请实施例中，所述特征提取模块，进一步用于：所述作为过滤器的第一卷积神经网络的各层在层的正向传递中对输入数据分别进行：对输入数据进行卷积处理以得到卷积特征图；对所述卷积特征图进行基于局部特征矩阵的均值池化以得到池化特征图；以及，对所述池化特征图进行非线性激活以得到激活特征图；其中，所述作为过滤器的第一卷积神经网络的最后一层的输出为所述温度-流速关联特征向量，所述作为过滤器的第一卷积神经网络的第一层的输入为所述校正后温度-流速关联矩阵。
具体地，在本申请实施例中，所述产物数据编码模块270，用于将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量。应可以理解，对于所述预定时间段内多个预定时间点的反应液的液相色谱图，考虑到所述反应液的液相色谱图在时序上具有着动态性地关联变化特征，因此，在本申请的技术方案中，为了充分地提取出这种动态变化特征信息来用于分类处理，将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量。
更具体地，在本申请实施例中，所述产物数据编码模块，进一步用于：所述使用三维卷积核的第二卷积神经网络在层的正向传递中对输入数据分别进行：基于所述三维卷积核对所述输入数据进行 三维卷积处理以得到卷积特征图；对所述卷积特征图进行基于局部特征矩阵的均值池化处理以得到池化特征图；以及，对所述池化特征图进行非线性激活以得到激活特征图；其中，所述第二卷积神经网络的最后一层的输出为所述产物特征向量，所述第二卷积神经网络的第一层的输入为所述预定时间段内多个预定时间点的反应液的液相色谱图。
具体地，在本申请实施例中，所述响应性估计模块280和所述配料控制结果生成模块290，用于计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵，并将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。应可以理解，由于所述反应液的液相色谱图的动态特征数据与所述温度和流速关联的隐含特征数据之间的特征尺度不同，并且所述反应液的液相色谱图的动态特征在高维特征空间中可以看作是对于所述温度流速关联变化的响应性特征，因此为了更好地融合所述产物特征向量和所述温度-流速关联特征向量，在本申请的技术方案中，进一步计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵。这样，就可以将所述分类特征矩阵通过分类器以得到用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小的分类结果。也就是，在一个具体示例中，所述分类器以如下公式对所述分类特征矩阵进行处理以生成分类结果，其中，所述公式为：softmax{(W n,B n):…:(W 1,B 1)|Project(F)}，其中Project(F)表示将所述分类特征矩阵投影为向量，W 1至W n为各层全连接层的权重矩阵，B 1至B n表示各层全连接层的偏置矩阵。
更具体地，在本申请实施例中，所述响应性估计模块，进一步用于：以如下公式计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到所述分类特征矩阵；其中，所述公式为：
F＝T*R
其中F表示所述产物特征向量，T表示所述分类特征矩阵，R表示所述温度-流速关联特征向量。
综上，基于本申请实施例的所述用于六氟磷酸锂制备的自动配料系统200被阐明，其通过使用人工智能的控制方法，以卷积神经网络模型来对于多个预定时间点的五氟化磷气体的气体流速值与反应温度值的关联矩阵以及反应液的液相色谱图进行隐含的特征提取，以在对于所述六氟磷酸锂产品的制备过程中，对于反应过程中的五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，进而提高所述六氟磷酸锂产品的制备效率。
如上所述，根据本申请实施例的用于六氟磷酸锂制备的自动配料系统200可以实现在各种终端设备中，例如用于六氟磷酸锂制备的自动配料算法的服务器等。在一个示例中，根据本申请实施例的用于六氟磷酸锂制备的自动配料系统200可以作为一个软件模块和/或硬件模块而集成到终端设备中。例如，该用于六氟磷酸锂制备的自动配料系统200可以是该终端设备的操作系统中的一个软件模块，或者可以是针对于该终端设备所开发的一个应用程序；当然，该用于六氟磷酸锂制备的自动配料系统200同样可以是该终端设备的众多硬件模块之一。
替换地，在另一示例中，该用于六氟磷酸锂制备的自动配料系统200与该终端设备也可以是分立的设备，并且该用于六氟磷酸锂制备的自动配料系统200可以通过有线和/或无线网络连接到该终端设备，并且按照约定的数据格式来传输交互信息。
示例性方法
图4图示了用于六氟磷酸锂制备的自动配料系统的配料方法的流程图。如图4所示，根据本申请实施例的用于六氟磷酸锂制备的自动配料系统的配料方法，包括步骤：S110，获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值；S120，获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图；S130，将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量；S140，计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵；S150，基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵；S160，将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量；S170，将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量；S180，计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵；以及，S190，将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。
图5图示了根据本申请实施例的用于六氟磷酸锂制备的自动配料系统的配料方法的架构示意图。如图5所示，在所述用于六氟磷酸锂制备的自动配料系统的配料方法的网络架构中，首先，将获得的所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值(例如，如图5中所示意的P1)和反应温度值(例如，如图5中所示意的P2)分别按照时间维度排列为温度输入向量(例如，如图5 中所示意的V1)和流速输入向量(例如，如图5中所示意的V2)；接着，计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵(例如，如图5中所示意的M1)；然后，基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵(例如，如图5中所示意的M2)；接着，将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络(例如，如图5中所示意的CNN1)以得到温度-流速关联特征向量(例如，如图5中所示意的VF1)；然后，将获得的所述预定时间段内多个预定时间点的反应液的液相色谱图(例如，如图5中所示意的P3)通过使用三维卷积核的第二卷积神经网络(例如，如图5中所示意的CNN2)以得到产物特征向量(例如，如图5中所示意的VF2)；接着，计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵(例如，如图5中所示意的MF)；以及，最后，将所述分类特征矩阵通过分类器(例如，如图5中所示意的分类器)以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。
更具体地，在步骤S110和步骤S120中，获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值，并获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图。应可以理解，由于在对于所述六氟磷酸锂产品的制备过程中，五氟化磷气体导入的流速与反应温度的控制策略的协同可有效地提升所述六氟磷酸锂产品的制备效率。因此，在本申请的技术方案中，期望在对于所述六氟磷酸锂产品的制备步骤中，对于反应过程中的五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，进而提高六氟磷酸锂产品的制备效率。
也就是，具体地，在本申请的技术方案中，首先，通过流速传感器获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值。应可以理解，若想在对于所述六氟磷酸锂产品的制备过程中对于五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，还需要通过温度传感器获取所述预定时间段内多个预定时间点的反应温度值，并且通过液相色谱仪采集所述预定时间段内多个预定时间点的反应液的液相色谱图来作为反应的结果依据特征。进而，这样就能够通过所述反应温度的变化特征信息以及所述反应液的液相色谱图的动态变化特征来进行所述五氟化磷气体的气体流速值调控。
更具体地，在步骤S130和步骤S140中，将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量，并计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵。应可以理解，由于所述五氟化磷气体的气体流速值和所述反应温度值之间具有着隐含的关联关系，因此为了提取出这种隐含的关联特征来进行所述五氟化磷气体气体流速值的准确调控，在本申请的技术方案中，进一步将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量，以分别整合所述各个时间点的五氟化磷气体的气体流速值和反应温度值信息。进一步地，为了构造所述五氟化磷气体的气体流速值和所述反应温度值之间的关联来进行特征挖掘，计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵。
更具体地，在步骤S150和步骤S160中，基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵，并将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量。应可以理解，由于所述作为过滤器的第一卷积神经网络提取局部关联特征，因此所述温度-流速关联特征向量可以对所述温度-流速关联矩阵的局部温度-流速关联具有良好的表达能力，但是，期望能够进一步强化所述温度-流速关联特征向量对所述温度-流速关联矩阵的全局温度-流速关联的表达能力。因此，在本申请的技术方案中，进一步对所述温度-流速关联矩阵M进行位置提议局部推理转化。然后，将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络中进行处理，以提取出所述校正后温度-流速关联矩阵中对应于所述五氟化磷气体的气体流速值和所述反应温度值之间的对应于全局的高维隐含关联特征，从而得到温度-流速关联特征向量。
应可以理解，由于所述温度-流速关联矩阵的每个位置的特征值都表示特定的温度-流速数值对之间的关联，通过以上位置提议局部推理转化，可以使用位置信息作为提议，来通过卷积层的局部感知场和转置结构的偏置来对全局场景语义进行推理，从而全面融合所捕获的局部温度-流速关联语义并进一步衍生全局温度-流速语义。这样，再通过所述作为过滤器的第一卷积神经网络时，就可以强化所述温度-流速关联特征向量对所述温度-流速关联矩阵的全局温度-流速关联的表达能力，进而提高分类的准确性。
更具体地，在步骤S170中，将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量。应可以理解，对于所述预定时间段内多个预定时间点的反应液的液相色谱图，考虑到所述反应液的液相色谱图在时序上具有着动态性地关联变化 特征，因此，在本申请的技术方案中，为了充分地提取出这种动态变化特征信息来用于分类处理，将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量。
更具体地，在步骤S180和步骤S190中，计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵，并将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。应可以理解，由于所述反应液的液相色谱图的动态特征数据与所述温度和流速关联的隐含特征数据之间的特征尺度不同，并且所述反应液的液相色谱图的动态特征在高维特征空间中可以看作是对于所述温度流速关联变化的响应性特征，因此为了更好地融合所述产物特征向量和所述温度-流速关联特征向量，在本申请的技术方案中，进一步计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵。这样，就可以将所述分类特征矩阵通过分类器以得到用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小的分类结果。
综上，基于本申请实施例的所述用于六氟磷酸锂制备的自动配料系统的配料方法被阐明，其通过使用人工智能的控制方法，以卷积神经网络模型来对于多个预定时间点的五氟化磷气体的气体流速值与反应温度值的关联矩阵以及反应液的液相色谱图进行隐含的特征提取，以在对于所述六氟磷酸锂产品的制备过程中，对于反应过程中的五氟化磷气体导入的流速进行与反应温度变化的实时动态地协同控制，进而提高所述六氟磷酸锂产品的制备效率。
以上结合具体实施例描述了本申请的基本原理，但是，需要指出的是，在本申请中提及的优点、优势、效果等仅是示例而非限制，不能认为这些优点、优势、效果等是本申请的各个实施例必须具备的。另外，上述公开的具体细节仅是为了示例的作用和便于理解的作用，而非限制，上述细节并不限制本申请为必须采用上述具体的细节来实现。
本申请中涉及的器件、装置、设备、系统的方框图仅作为例示性的例子并且不意图要求或暗示必须按照方框图示出的方式进行连接、布置、配置。如本领域技术人员将认识到的，可以按任意方式连接、布置、配置这些器件、装置、设备、系统。诸如“包括”、“包含”、“具有”等等的词语是开放性词汇，指“包括但不限于”，且可与其互换使用。这里所使用的词汇“或”和“和”指词汇“和/或”，且可与其互换使用，除非上下文明确指示不是如此。这里所使用的词汇“诸如”指词组“诸如但不限于”，且可与其互换使用。
还需要指出的是，在本申请的装置、设备和方法中，各部件或各步骤是可以分解和/或重新组合的。这些分解和/或重新组合应视为本申请的等效方案。
提供所公开的方面的以上描述以使本领域的任何技术人员能够做出或者使用本申请。对这些方面的各种修改对于本领域技术人员而言是非常显而易见的，并且在此定义的一般原理可以应用于其他方面而不脱离本申请的范围。因此，本申请不意图被限制到在此示出的方面，而是按照与在此公开的原理和新颖的特征一致的最宽范围。
为了例示和描述的目的已经给出了以上描述。此外，此描述不意图将本申请的实施例限制到在此公开的形式。尽管以上已经讨论了多个示例方面和实施例，但是本领域技术人员将认识到其某些变型、修改、改变、添加和子组合。

Claims (10)

  1. 一种用于六氟磷酸锂制备的自动配料系统，其特征在于，包括：气体配料数据采集模块，用于获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值；反应数据采集模块，用于获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图；离散数据向量化模块，用于将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量；数据级关联模块，用于计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵；校正模块，用于基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵；特征提取模块，用于将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量；产物数据编码模块，用于将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量；响应性估计模块，用于计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵；以及配料控制结果生成模块，用于将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。
  2. 根据权利要求1所述的用于六氟磷酸锂制备的自动配料系统，其特征在于，所述校正模块，包括：转置单元，用于计算所述温度-流速关联矩阵的转置矩阵；第一卷积单元，用于对所述温度-流速关联矩阵进行卷积处理以得到第一卷积特征矩阵；融合单元，用于计算所述温度-流速关联矩阵的转置矩阵和所述第一卷积特征矩阵的按位置点加以得到初步融合特征矩阵；第二卷积单元，用于对所述初步融合特征矩阵进行卷积处理以得到第二卷积特征矩阵；位置信息提取单元，用于将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到位置信息矩阵；以及再融合单元，用于计算所述第二卷积特征矩阵和所述位置信息矩阵的按位置点加以得到所述校正后温度-流速关联矩阵。
  3. 根据权利要求2所述的用于六氟磷酸锂制备的自动配料系统，其特征在于，所述位置信息提取单元，进一步用于：以如下公式将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到所述位置信息矩阵；其中，所述公式为：M 1＝Φ(P M)
    其中，M表示所述温度-流速关联矩阵，M 1表示所述位置信息矩阵，P M表示所述温度-流速关联矩阵的(x,y)坐标矩阵，
    Figure PCTCN2022120829-appb-100001
    用于将二维位置坐标映射为一维数值。
  4. 根据权利要求3所述的用于六氟磷酸锂制备的自动配料系统，其特征在于，所述特征提取模块，进一步用于：所述作为过滤器的第一卷积神经网络的各层在层的正向传递中对输入数据分别进行：对输入数据进行卷积处理以得到卷积特征图；对所述卷积特征图进行基于局部特征矩阵的均值池化以得到池化特征图；以及对所述池化特征图进行非线性激活以得到激活特征图；其中，所述作为过滤器的第一卷积神经网络的最后一层的输出为所述温度-流速关联特征向量，所述作为过滤器的第一卷积神经网络的第一层的输入为所述校正后温度-流速关联矩阵。
  5. 根据权利要求4所述的用于六氟磷酸锂制备的自动配料系统，其特征在于，所述产物数据编码模块，进一步用于：所述使用三维卷积核的第二卷积神经网络在层的正向传递中对输入数据分别进行：基于所述三维卷积核对所述输入数据进行三维卷积处理以得到卷积特征图；对所述卷积特征图进行基于局部特征矩阵的均值池化处理以得到池化特征图；以及
    对所述池化特征图进行非线性激活以得到激活特征图；其中，所述第二卷积神经网络的最后一层的输出为所述产物特征向量，所述第二卷积神经网络的第一层的输入为所述预定时间段内多个预定时间点的反应液的液相色谱图。
  6. 根据权利要求5所述的用于六氟磷酸锂制备的自动配料系统，其特征在于，所述响应性估计模块，进一步用于：以如下公式计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到所述分类特征矩阵；其中，所述公式为：
    F＝T*R
    其中F表示所述产物特征向量，T表示所述分类特征矩阵，R表示所述温度-流速关联特征向量。
  7. 根据权利要求6所述的用于六氟磷酸锂制备的自动配料系统，其特征在于，所述配料控制结果生成模块，进一步用于：所述分类器以如下公式对所述分类特征矩阵进行处理以生成分类结果，其中，所述公式为：softmax{(W n,B n):…:(W 1,B 1)|Project(F)}，其中Project(F)表示将所述分类特征矩阵投影为向量，W 1至W n为各层全连接层的权重矩阵，B 1至B n表示各层全连接层的偏置矩阵。
  8. 一种用于六氟磷酸锂制备的自动配料系统的配料方法，其特征在于，包括：获取预定时间段内多个预定时间点的五氟化磷气体通入到盛有氟化锂的无水氟化氢溶液的气体流速值；获取所述预定时间段内多个预定时间点的反应温度值和反应液的液相色谱图；将所述预定时间段内多个预定时间点的五氟化磷气体的气体流速值和反应温度值分别按照时间维度排列为温度输入向量和流速输入向量； 计算所述温度输入向量的转置向量与所述流速输入向量之间的乘积以得到温度-流速关联矩阵；基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵；将所述校正后温度-流速关联矩阵通过作为过滤器的第一卷积神经网络以得到温度-流速关联特征向量；将所述预定时间段内多个预定时间点的反应液的液相色谱图通过使用三维卷积核的第二卷积神经网络以得到产物特征向量；计算所述产物特征向量相对于所述温度-流速关联特征向量的响应性估计以得到分类特征矩阵；以及将所述分类特征矩阵通过分类器以得到分类结果，所述分类结果用于表示当前时间点的五氟化磷气体的气体流速值应增大或应减小。
  9. 根据权利要求8所述的用于六氟磷酸锂制备的自动配料系统的配料方法，其特征在于，所述基于所述温度-流速关联矩阵的位置信息，对所述温度-流速关联矩阵进行校正以得到校正后温度-流速关联矩阵-流速关联矩阵，包括：计算所述温度-流速关联矩阵的转置矩阵；对所述温度-流速关联矩阵进行卷积处理以得到第一卷积特征矩阵；计算所述温度-流速关联矩阵的转置矩阵和所述第一卷积特征矩阵的按位置点加以得到初步融合特征矩阵；对所述初步融合特征矩阵进行卷积处理以得到第二卷积特征矩阵；将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到位置信息矩阵；以及计算所述第二卷积特征矩阵和所述位置信息矩阵的按位置点加以得到所述校正后温度-流速关联矩阵。
  10. 根据权利要求9所述的用于六氟磷酸锂制备的自动配料系统的配料方法，其特征在于，所述将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到位置信息矩阵，包括：以如下公式将所述温度-流速关联矩阵中各个位置的二维位置坐标映射为一维数值以得到所述位置信息矩阵；其中，所述公式为：M 1＝Φ(P M)其中，M表示所述温度-流速关联矩阵，M 1表示所述位置信息矩阵，P M表示所述温度-流速关联矩阵的(x,y)坐标矩阵，
    Figure PCTCN2022120829-appb-100002
    用于将二维位置坐标映射为一维数值。
PCT/CN2022/120829 2022-08-31 2022-09-23 用于六氟磷酸锂制备的自动配料系统及其配料方法 WO2024045243A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211062451.9 2022-08-31
CN202211062451.9A CN115430344B (zh) 2022-08-31 2022-08-31 用于六氟磷酸锂制备的自动配料系统及其配料方法

Publications (1)

Publication Number Publication Date
WO2024045243A1 true WO2024045243A1 (zh) 2024-03-07

Family

ID=84244462

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/120829 WO2024045243A1 (zh) 2022-08-31 2022-09-23 用于六氟磷酸锂制备的自动配料系统及其配料方法

Country Status (2)

Country Link
CN (1) CN115430344B (zh)
WO (1) WO2024045243A1 (zh)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115936682B (zh) * 2022-12-21 2023-11-07 江西有源工业废物回收处理有限公司 印制线路板的废料回收系统及其方法
CN116036639B (zh) * 2023-02-28 2024-03-15 福建省龙德新能源有限公司 超声波诱导六氟磷酸盐结晶的控制系统及其方法
CN116825215B (zh) * 2023-02-28 2024-04-16 福建省龙德新能源有限公司 用于六氟磷酸锂制备的流体循环反应控制系统及其方法
CN116825217B (zh) * 2023-03-15 2024-05-14 福建省德旭新材料有限公司 制备高纯五氟化磷的方法
CN116859830B (zh) * 2023-03-27 2024-01-26 福建天甫电子材料有限公司 用于电子级氟化铵生产的生产管理控制系统
CN116785967B (zh) * 2023-04-28 2024-03-08 福建省龙德新能源有限公司 用于电子级六氟磷酸锂制备的自动配料系统
CN116605891B (zh) * 2023-05-23 2024-03-19 福建省龙德新能源有限公司 电子级六氟磷酸锂的智能生产方法及其系统
CN117018858A (zh) * 2023-08-11 2023-11-10 滁州锡安环保科技有限责任公司 工业废气净化设备及其控制方法

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102380220A (zh) * 2011-10-10 2012-03-21 上海师范大学 一种乙烯精馏塔塔釜乙烯浓度的测量方法
CN104965967A (zh) * 2015-03-26 2015-10-07 华东理工大学 一种常减压蒸馏装置的收率实时预测方法
JP5960907B2 (ja) * 2013-03-25 2016-08-02 三井化学株式会社 五フッ化リンの製造方法及び六フッ化リン酸リチウムの製造方法
CN110052050A (zh) * 2019-04-29 2019-07-26 中国石油大学(华东) 基于塔板温度的内部热耦合精馏塔全局状态观测器及方法
CN111468060A (zh) * 2020-05-22 2020-07-31 福建省龙德新能源股份有限公司 一种用于合成六氟磷酸盐的智能合成装置
CN111871357A (zh) * 2020-08-24 2020-11-03 福建省龙德新能源股份有限公司 一种用于合成六氟磷酸锂的高效合成装置
CN112295255A (zh) * 2020-10-24 2021-02-02 四川泸天化创新研究院有限公司 一种甲醇精馏装置智能控制系统及控制方法
CN112365045A (zh) * 2020-11-09 2021-02-12 上海明华电力科技有限公司 一种基于大数据的主蒸汽温度智能预测方法
CN112486686A (zh) * 2020-11-30 2021-03-12 之江实验室 基于云边协同的定制化深度神经网络模型压缩方法及系统
CN113783527A (zh) * 2021-09-26 2021-12-10 中机国能电力工程有限公司 一种基于光伏组件温度和辐照度校验的数据挖掘方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060080041A1 (en) * 2004-07-08 2006-04-13 Anderson Gary R Chemical mixing apparatus, system and method
CN106709258B (zh) * 2017-01-06 2019-07-02 张君 一种智能自动化中草药配药系统
CN110614062B (zh) * 2019-09-20 2021-11-12 北京诚益通控制工程科技股份有限公司 一种基于自动折干折纯配料方法的配料系统
CN112919504B (zh) * 2021-03-18 2023-02-17 湖北迈可凯科技有限公司 一种采用连续流反应器合成六氟磷酸锂的方法及装置
CN113899822A (zh) * 2021-06-08 2022-01-07 森田新能源材料（张家港）有限公司 一种基于离子色谱技术六氟磷酸锂纯度的测定方法
CN114358412A (zh) * 2021-12-31 2022-04-15 上海正诚机电制造有限公司 一种优化饲料变频配料速度和精度的方法

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102380220A (zh) * 2011-10-10 2012-03-21 上海师范大学 一种乙烯精馏塔塔釜乙烯浓度的测量方法
JP5960907B2 (ja) * 2013-03-25 2016-08-02 三井化学株式会社 五フッ化リンの製造方法及び六フッ化リン酸リチウムの製造方法
CN104965967A (zh) * 2015-03-26 2015-10-07 华东理工大学 一种常减压蒸馏装置的收率实时预测方法
CN110052050A (zh) * 2019-04-29 2019-07-26 中国石油大学(华东) 基于塔板温度的内部热耦合精馏塔全局状态观测器及方法
CN111468060A (zh) * 2020-05-22 2020-07-31 福建省龙德新能源股份有限公司 一种用于合成六氟磷酸盐的智能合成装置
CN111871357A (zh) * 2020-08-24 2020-11-03 福建省龙德新能源股份有限公司 一种用于合成六氟磷酸锂的高效合成装置
CN112295255A (zh) * 2020-10-24 2021-02-02 四川泸天化创新研究院有限公司 一种甲醇精馏装置智能控制系统及控制方法
CN112365045A (zh) * 2020-11-09 2021-02-12 上海明华电力科技有限公司 一种基于大数据的主蒸汽温度智能预测方法
CN112486686A (zh) * 2020-11-30 2021-03-12 之江实验室 基于云边协同的定制化深度神经网络模型压缩方法及系统
CN113783527A (zh) * 2021-09-26 2021-12-10 中机国能电力工程有限公司 一种基于光伏组件温度和辐照度校验的数据挖掘方法

Also Published As

Publication number Publication date
CN115430344B (zh) 2023-04-28
CN115430344A (zh) 2022-12-06

Similar Documents

Publication Publication Date Title
WO2024045243A1 (zh) 用于六氟磷酸锂制备的自动配料系统及其配料方法
WO2024045247A1 (zh) 用于氟化铵生产的生产管理控制系统及其控制方法
CN115079572B (zh) 用于六氟磷酸锂制备的能源管理控制系统及其控制方法
CN115586755B (zh) 用于六氟磷酸锂制备的生产管理控制系统及其方法
CN115309215B (zh) 氟化铵制备用的自动配料控制系统及其控制方法
Alipour et al. A review on temperature-dependent electrochemical properties, aging, and performance of lithium-ion cells
CN115599049B (zh) 用于无水氟化氢生产的能源管理控制系统及其控制方法
WO2024021254A1 (zh) 电子级三氟化氯的智能分离纯化系统
CN115202265A (zh) 电子级氢氧化钾的智慧产线的控制系统及其控制方法
WO2023103150A1 (zh) 电解液设计方法、装置、设备、介质及程序产品
CN115903705B (zh) 用于电子级六氟丁二烯制备的生产管理控制系统
CN111829688A (zh) 一种锂离子电池工作过程中的温度场分布预测方法
CN113538535A (zh) 一种基于双分支特征交互的3d点云配准网络模型及配准方法
CN106430257A (zh) 一种多孔高活性氟化锂的制备方法及双(氟磺酰)亚胺锂的制备方法
CN112906899A (zh) 一种基于量子计算的最小均方误差检测方法
WO2024082475A1 (zh) 基于毫米波通感一体化无人设备的自主导航系统及方法
WO2023226236A1 (zh) 用于电子级氢氟酸制备的能源管理控制系统及其控制方法
CN115090200B (zh) 用于电子级氢氟酸制备的自动配料系统及其配料方法
WO2024036690A1 (zh) 用于剥膜液生产的自动配料系统及其配料方法
CN116067188A (zh) 一种六氟磷酸锂制备用的余热回收系统
CN116785967A (zh) 用于电子级六氟磷酸锂制备的自动配料系统
CN115048982A (zh) 一种数据中心用智能电网监测系统及其控制方法
CN117293975B (zh) 一种锂电池的充放电调节方法、装置、设备及存储介质
CN116736139B (zh) 一种家用储能系统的soc估算方法
CN113447818B (zh) 电池等效电路模型的辨识方法及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22957062

Country of ref document: EP

Kind code of ref document: A1