Continuous freezing crystallization kettle size design method and system

Abstract

The present invention belongs to the technical field of crystallization kettles, and in particular, relates to a size design method and system for a continuous freezing crystallization kettle, including: determining the production target information of materials in each partition in the continuous freezing crystallization kettle, the production target information is based on the set material state at the inlet and outlet of the continuous freezing crystallization kettle; establishing the association relationship between each partition; establishing a size association model for each partition according to the association relationship, training and optimizing the size association model; applying the size association model to the size scheme prediction of the continuous freezing crystallization kettle. In the present invention, the size of each partition is reasonably designed, which is helpful to control the growth rate and quality of the crystal, avoid the problem of uneven crystal growth, ensure that each functional partition can work in coordination, improve the heat exchange efficiency through optimized size design, comprehensively consider factors such as material properties, operating conditions and expected output, design the optimal kettle size, and improve the overall performance and economy of the equipment.

Description

Technical Field

[0001] The invention belongs to the technical field of crystallization kettles, and in particular relates to a size design method and system of a continuous freezing crystallization kettle. Background Art

[0002] For continuous freezing crystallization kettles, a continuous operation mode is often adopted. The material continuously enters the kettle and crystallizes through different partitions to achieve the purpose of continuous discharge. This mode improves production efficiency and stability.

[0003] The continuous freezing crystallization kettle partitions include:

[0004] Heat exchange crystallization area: located at the top of the kettle, its main function is to achieve high-efficiency heat exchange.

[0005] Crystal growth zone: Located in the middle of the kettle, it provides a stable temperature and flow conditions for crystallization, which is conducive to the growth and maturation of crystals. The diameter of this area is usually smaller than the heat exchange crystallization zone, which helps to control the speed and quality of crystal growth.

[0006] Crystal discharge area: Located at the bottom of the kettle, crystals further mature in this area and are continuously discharged through the discharge port.

[0007] In view of the different functional partitions of the continuous freezing crystallization kettle, the size design of the kettle body is indeed crucial. Through the reasonable design of the diameter and height of each partition, the crystallization efficiency, crystal quality and the stability of the overall operation can be optimized. When designing the size of the kettle body, it is necessary to comprehensively consider the properties of the material (such as viscosity, thermal conductivity), operating conditions (such as temperature, pressure), expected output, and the economy and operability of the equipment. The choice of size is not only related to the crystallization efficiency, but also affects the energy consumption and the overall performance of the equipment. Summary of the invention

[0008] The present invention provides a method and system for designing the size of a continuous freezing crystallization kettle, which can effectively solve the problems in the background technology.

[0009] In order to achieve the above object, the technical solution adopted by the present invention is:

[0010] The size design method of continuous freezing crystallization kettle includes:

[0011] Determine the production target information of the material in each partition in the continuous freezing crystallization kettle, wherein the production target information is based on the set material state at the inlet and outlet of the continuous freezing crystallization kettle;

[0012] Establish the association relationship between each partition;

[0013] Establishing a size association model for each partition according to the association relationship, and training and optimizing the size association model;

[0014] The size correlation model is applied to the size scheme prediction of the continuous freezing crystallization kettle.

[0015] Furthermore, the association relationship between the partitions is established, including:

[0016] The production target information data of the heat exchange crystallization zone, crystal growth zone and crystal outlet zone under different operating conditions, as well as the historical operation data and experimental data of the equipment are collected, and each data is preprocessed to obtain a data set to be used; and the correlation between each partition is analyzed based on the data set to be used.

[0017] Further, analyzing the association relationship between the partitions according to the data set to be used includes:

[0018] Dividing the data set to be used into three independent partition data sets according to each partition;

[0019] The Pearson correlation coefficient was used to calculate the correlation of variables within each partitioned data set;

[0020]

[0021] In each partition, x i is the ith observation value of variable x, is the average value of each variable x, y i is the i-th observation value of variable y, is the average value of each variable y, -1≤r≤1;

[0022] Constructing a correlation matrix according to the Pearson correlation coefficient obtained for each partition;

[0023] A comprehensive relationship diagram across partitions is constructed based on each of the correlation matrices, and the association relationship between the partitions is identified based on the comprehensive relationship diagram.

[0024] Furthermore, a comprehensive relationship diagram across partitions is constructed, including:

[0025] Determine the graph structure, define each variable as a node in the graph structure, and the correlation between two variables as the edge of the graph structure;

[0026] According to the correlation matrix of each partition, sub-graphs of heat exchange crystallization zone, crystal growth zone and crystal out zone are generated respectively;

[0027] adding edges across partitions according to the physical and process significance of the variables in each of the correlation matrices;

[0028] The subgraphs of each partition and the edges across partitions are merged to construct the comprehensive relationship graph.

[0029] Furthermore, the association relationship between the partitions is identified based on the comprehensive relationship graph, including key path analysis, important node and edge analysis, and cross-partition association analysis.

[0030] Furthermore, it also includes processing the key path analysis, important node and edge analysis, and cross-partition association analysis, and obtaining at least the following contents:

[0031] The starting and ending partitions of the key paths, the variables and weights in the paths, and the length of the paths;

[0032] The partitions, types, and importance measures of important nodes and edges;

[0033] The starting and ending partitions of the cross-partition association, the associated variables and weights, and the type of association;

[0034] A feature matrix is ​​created, and each of the contents is embedded in the feature matrix as an input of the size association model.

[0035] Furthermore, the size association model is one of a multilayer perceptron, a convolutional neural network and a recurrent neural network.

[0036] Furthermore, a size association model of each partition is established according to the association relationship, including:

[0037] Standardizing the feature matrix so that the value range of each feature is similar, thereby obtaining a standardized matrix;

[0038] Through the standardized matrix, a deep learning model is selected to train the size association model.

[0039] Furthermore, cross-partition edges are added according to the physical and process meanings of the variables in each of the correlation matrices, including: determining the key variables between different partitions according to the physical and process meanings of the variables, and defining the relationship between the key variables; determining the cross-partition edges that need to be added to the graph structure according to the key variables and the relationship.

[0040] The dimension design system of continuous freezing crystallization kettle includes:

[0041] An information collection module determines the production target information of the material in each partition in the continuous freezing crystallization kettle, and the production target information is based on the set material status at the inlet and outlet of the continuous freezing crystallization kettle;

[0042] An association relationship establishment module is used to establish association relationships among partitions;

[0043] A model training module, which establishes a size association model for each partition according to the association relationship, and trains and optimizes the size association model;

[0044] The scheme output module applies the size association model to realize the size scheme prediction of the continuous freezing crystallization kettle.

[0045] The technical solution of the present invention can achieve the following technical effects:

[0046] In the present invention, by optimizing the size design of each partition, the material can more efficiently perform heat exchange and crystallization during the entire crystallization process, thereby improving the crystallization efficiency; the reasonable design of the size of each partition helps to control the growth rate and quality of the crystal, avoid the problem of uneven crystal growth, ensure that each functional partition can work in coordination, and ensure the continuity and stability of the operation. After implementation, the optimized size design can improve the heat exchange efficiency, reduce unnecessary energy consumption, thereby reducing the overall energy consumption, and comprehensively consider factors such as material properties, operating conditions and expected output to design the optimal kettle size, thereby improving the overall performance and economy of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 A flow chart of the method for designing the size of a continuous freezing crystallization kettle;

[0049] Figure 2 A flowchart for analyzing the association relationship between partitions based on the data set to be used;

[0050] Figure 3 Flowchart for building a comprehensive relationship graph across partitions. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0052] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0053] Embodiment 1

[0054] like Figure 1 As shown, the size design method of the continuous freezing crystallization kettle includes:

[0055] S1: Determine the production target information of the material in each partition in the continuous freezing crystallization kettle, and the production target information is based on the set material state at the inlet and outlet of the continuous freezing crystallization kettle;

[0056] In this step, the production target information refers to the specific parameters and conditions set to achieve efficient and stable operation of the continuous freezing crystallization kettle; the set material state at the inlet refers to the initial state of the material when entering the continuous freezing crystallization kettle, including parameters such as temperature, pressure, flow rate and viscosity, which are set according to the upstream process; the set material state at the outlet refers to the final state of the material when leaving the continuous freezing crystallization kettle, including target crystallinity, discharge temperature, pressure and flow rate, etc., which are set according to downstream process requirements and product quality requirements.

[0057] The production target information of each partition may be determined as follows. Of course, the following contents are only for showing a specific production target information determination result. Other production target information combination methods are also applicable to the present invention:

[0058] The production target information of the heat exchange and crystallization area includes:

[0059] (1) Target temperature: The main function of the heat exchange crystallization zone is to achieve efficient heat exchange so that the material can be quickly cooled to a temperature suitable for crystal nucleation. The target temperature needs to be set according to the physical properties of the material and the thermodynamic characteristics of crystallization.

[0060] (2) Target pressure: Considering the possible gas precipitation or absorption during the crystallization process, the target pressure must ensure the stability of the material in the liquid phase.

[0061] (3) Target flow rate: The flow rate in the heat exchange and crystallization zone should ensure uniform heat exchange and crystal nucleation.

[0062] The production target information of the crystal growth area includes:

[0063] (1) Target temperature: The crystal growth zone provides a stable temperature environment, which is conducive to the growth and maturation of crystals. The target temperature needs to be controlled within a narrow range to avoid the crystals growing too fast or too slow.

[0064] (2) Target flow rate: The flow rate should be kept stable to avoid turbulence and sedimentation, which is conducive to uniform growth of crystals.

[0065] (3) Viscosity and shear force: The viscosity and flow state of the material in the crystal growth zone affect the growth rate and morphology of the crystal. The target viscosity needs to be set according to the material properties and crystallization dynamics.

[0066] The production target information of the crystal-out area includes:

[0067] (1) Target temperature: The temperature in the crystallization zone should ensure further crystal maturation while avoiding redissolution or adhesion.

[0068] (2) Target flow rate: The flow rate needs to be properly controlled to ensure smooth discharge of crystals and avoid blockage.

[0069] (3) Pressure and guidance: Considering the discharge requirements of the crystal discharge area, the target pressure should ensure smooth discharge of the material, and the design of the guide structure should ensure smooth crystal discharge.

[0070] In the specific implementation process, the set state parameters of the material at the inlet and outlet can be collected first, and then the production target information of each partition can be set according to the process requirements and equipment capabilities. By determining the production target information of each partition of the material in the continuous freezing crystallization kettle in detail, it can be ensured that the heat exchange crystallization zone is cooled rapidly to improve the crystal nucleus formation rate, and the crystal growth zone provides stable temperature and flow conditions to ensure uniform growth and maturity of the crystals and improve the crystal quality. The target temperature and flow rate of the crystal outlet zone are reasonably designed to ensure smooth discharge of the crystals and avoid blockage and re-dissolution.

[0071] During the implementation process, the production target information of each partition not only affects the respective operation results independently, but also affects the efficiency and quality of the overall crystallization process. Therefore, the following steps are required:

[0072] S2: Establish the association relationship between each partition;

[0073] S3: Establishing a size association model for each partition based on the association relationship, and training and optimizing the size association model through historical data;

[0074] S4: Apply the size correlation model to predict the size solution of continuous freezing crystallization reactor.

[0075] In the present invention, by optimizing the size design of each partition, the material can more efficiently perform heat exchange and crystallization during the entire crystallization process, thereby improving the crystallization efficiency; the reasonable design of the size of each partition helps to control the growth rate and quality of the crystal, avoid the problem of uneven crystal growth, ensure that each functional partition can work in coordination, and ensure the continuity and stability of the operation. After the implementation is completed, the optimized size design can improve the heat exchange efficiency, reduce unnecessary energy consumption, thereby reducing the overall energy consumption, and comprehensively consider factors such as material properties, operating conditions and expected output to design the optimal kettle size, thereby improving the overall performance and economy of the equipment.

[0076] During the implementation process, by establishing the association relationship between each partition and using historical data to train and optimize the size association model, this data-driven method can ensure that the designed size is more in line with actual production needs and avoid the uncertainty in traditional empirical design; the established size association model can be dynamically adjusted and optimized according to the actual production situation. For example, by real-time monitoring of data in the production process, the model parameters can be adjusted and continuously optimized, so that the size design of each partition is always maintained in the optimal state; using the size association model for design can greatly reduce the design cycle. Traditional size design often requires repeated experiments and adjustments, while model prediction can obtain a more ideal design solution at one time, thereby improving design efficiency and saving time and cost.

[0077] As a preferred embodiment of the above, establishing the association relationship between each partition includes: collecting production target information data of the heat exchange crystallization zone, crystal growth zone and crystal out zone under different operating conditions, as well as the historical operation data and experimental data of the equipment, and preprocessing each data to obtain the data set to be used; wherein, the goal of preprocessing is to clean and standardize the data to ensure data quality for subsequent analysis and modeling, and specific operations include: missing data processing, using interpolation, mean filling or deleting incomplete data records; data standardization, normalizing data of different dimensions to the same range, commonly used methods include Z-score normalization and Min-Max normalization; data denoising, applying filters or other denoising techniques to remove noise in the data to ensure that the data is clear and accurate. Establishing the association relationship between each partition also includes analyzing the association relationship between each partition according to the data set to be used.

[0078] In the above preferred solution, by collecting production target information data under different operating conditions, the performance of each partition under various working conditions can be obtained. These data include important parameters such as temperature, pressure, flow rate, etc., which can help understand the optimal operating state of each partition under different conditions. The historical operation data of the equipment records the long-term operating performance of the equipment in actual production, including successful operation data and failure data. These historical data are an important reference for optimizing equipment design and operating parameters, and help to understand the performance change trend and potential problems of the equipment. For example, historical data can reveal the performance trend and change law of the equipment in long-term operation, help to discover and prevent potential problems, or, by analyzing failure data, the causes and influencing factors of common failures can be identified, so as to formulate effective preventive measures and improve the reliability and stability of the equipment. As for experimental data, it is usually obtained under controlled conditions and has high accuracy and repeatability. These data can verify and supplement the actual operation data to ensure the comprehensiveness and reliability of the data. Experimental data can be used to verify the accuracy of actual operation data to ensure the authenticity and reliability of the data. Under certain specific operating conditions, data may be lacking in actual production. These data can be supplemented through experiments to ensure the integrity of the data set.

[0079] As a preferred embodiment of the above, Figure 2 As shown, the association relationship between the partitions is analyzed according to the data set to be used, including:

[0080] A1: Divide the dataset to be used into three independent partition datasets according to each partition. This step can be divided by data labeling. In the dataset to be used, mark the partition to which each data point belongs, including production target information data, equipment historical operation data and experimental data, and divide according to the labeling results;

[0081] A2: Use the Pearson correlation coefficient to calculate the correlation of variables within each partitioned data set;

[0082]

[0083] In each partition, x i is the ith observation value of variable x, is the average value of each variable x, y i is the i-th observation value of variable y, is the average value of each variable y, -1≤r≤1;

[0084] A3: Construct a correlation matrix based on the Pearson correlation coefficients obtained for each partition;

[0085] A4: A comprehensive relationship diagram across partitions is constructed based on each correlation matrix, and the correlation relationship between the partitions is identified based on the comprehensive relationship diagram.

[0086] In the above preferred scheme, dividing the data set into independent partitions helps to more finely analyze the variable relationship within each partition and improve the accuracy of the analysis; using the Pearson correlation coefficient to calculate the variable correlation can accurately quantify the linear relationship between the variables. The Pearson correlation coefficient can identify strongly correlated, weakly correlated or unrelated variables, help understand the mutual influence between different variables in each partition, and facilitate the optimization of design and operating parameters.

[0087] During the implementation process, by constructing a correlation matrix, the correlation of variables within the partition is clearly displayed, which helps to discover potential influencing factors. The comprehensive relationship diagram intuitively displays the relationship between partitions, making the correlation across partitions clear at a glance, which facilitates further optimization of the design.

[0088] As a preferred embodiment of the above, Figure 3 As shown, a comprehensive relationship diagram across partitions is constructed, including:

[0089] B1: Determine the graph structure, define each variable as a node in the graph structure, and define the correlation between two variables as the edge of the graph structure; for each partition (heat exchange crystallization zone, crystal growth zone, crystal out zone), the node can be a variable such as temperature, pressure, flow rate, viscosity, etc. The edge weight between variables with strong correlation is larger, and vice versa;

[0090] B2: Generate subgraphs of the heat exchange crystallization zone, crystal growth zone and crystal output zone respectively according to the correlation matrix of each partition; specifically, generate a subgraph containing nodes of all relevant variables and edges between them according to the correlation matrix of the heat exchange crystallization zone, generate a subgraph containing nodes of all relevant variables and edges between them according to the correlation matrix of the crystal growth zone, and generate a subgraph containing nodes of all relevant variables and edges between them according to the correlation matrix of the crystal output zone;

[0091] B3: Add cross-zone edges according to the physical and process significance of the variables in each correlation matrix. For example, the temperature of the heat exchange crystallization zone may affect the flow rate of the crystal growth zone, or the viscosity of the crystal growth zone may affect the pressure of the crystal outlet zone. Add edges between cross-zone variables and use the correlation coefficient as the weight of the edge. These edges represent the physical and process associations between zones.

[0092] B4: Merge the subgraphs of each partition and the edges across partitions to construct a comprehensive relationship graph.

[0093] In the above optimization scheme, the relationship between variables is intuitively displayed through the graph structure, including the correlation within and between partitions. The Pearson correlation coefficient is used to quantify the linear relationship between variables, providing accurate correlation measurement. The subgraphs and comprehensive relationship graphs constructed based on the correlation matrix help identify key variables and important associations, providing a scientific basis for optimization design.

[0094] As a further preference, cross-partition edges are added according to the physical and process meanings of the variables in each correlation matrix, including: determining the key variables between different partitions according to the physical and process meanings of the variables, and defining the relationship between the key variables; determining the cross-partition edges that need to be added to the graph structure according to the key variables and the relationship.

[0095] Screening key variables has important advantages in optimizing the design and operation of continuous freezing crystallization kettles, including improving model performance, optimizing resource allocation, improving production efficiency and product quality, supporting scientific decision-making and continuous optimization, and enhancing system stability and robustness. During the implementation process, the screening of key variables can reduce data dimensions and enhance relevance. The screening of key variables makes model training faster and more accurate, while focusing on key factors, achieving precise control and effective monitoring, reducing production costs and improving overall efficiency and product quality. In addition, this process also provides a clear basis for decision-making and the ability to adjust dynamically, ensuring that the system always maintains the best state in actual operation, reducing the occurrence of failures, extending equipment life, and ultimately achieving efficient and stable production.

[0096] Among them, the association relationship between each partition is identified according to the comprehensive relationship diagram, and the implementation process may specifically include key path analysis, important node and edge analysis, and cross-partition association analysis; further, it also includes processing the key path analysis, important node and edge analysis, and cross-partition association analysis, and at least obtains the following content: the starting partition and the ending partition of the key path, the variables and weights in the path, and the length of the path; the partitions, types, and importance measures of important nodes and edges; the starting partition and the ending partition of the cross-partition association, the associated variables and weights, and the associated type.

[0097] After obtaining the above contents, a feature matrix is ​​created, and each content is embedded in the feature matrix as the input of the size association model.

[0098] During the implementation process, the key path is a path from the starting partition to the ending partition in the comprehensive relationship diagram. The variables on this path are highly correlated and are a concentrated reflection of the key operating procedures and influencing factors. Specifically: Starting partition and ending partition: determine the starting and ending points of the path, for example, the heat exchange crystallization zone (starting point) to the crystal growth zone (end point). Path variables and weights: record all variables involved in the path and their correlation coefficients (weights), for example, temperature (T1) -> flow rate (V2) -> viscosity (Vis2). Path length: calculate the path length, that is, the number of steps between variables in the path. The path length reflects the complexity and scope of influence of the path.

[0099] Important nodes are the variables that have the greatest impact on the overall operation in the comprehensive relationship diagram. These nodes usually have a higher degree (number of connected edges) or a larger edge weight. Specifically: Partition: Record the partition where each important node is located (heat exchange crystallization zone, crystal growth zone or crystal out zone). Type: Distinguish the type of node, such as temperature, pressure, flow rate, etc. Importance measurement: Calculate the degree (number of connected edges) and edge weight of the node. A high degree and a large edge weight indicate that the node has an important impact in the production process.

[0100] Important edges are the edges that have the greatest impact on the association of variables in the comprehensive relationship diagram. These edges indicate that there is a strong correlation or significant impact between two variables. Specifically: Partition: Record the partition where each important edge is located. Type: Distinguish the type of edge, such as the relationship between temperature and flow rate, the relationship between pressure and flow rate, etc. Importance measurement: Measure the importance of the edge based on its weight (correlation coefficient). The larger the weight, the stronger the association between the two variables.

[0101] Cross-partition association refers to the association of variables between different partitions in the comprehensive relationship diagram. These associations reflect the mutual influence and synergistic relationship between different partitions. Specifically: Starting and ending partitions: Determine the starting and ending partitions of the cross-partition association, for example, from the heat exchange crystallization zone to the crystal growth zone. Associated variables and weights: Record the specific variables and their correlation coefficients (weights) of the cross-partition association, for example, the temperature (T1) of the heat exchange crystallization zone and the flow rate (V2) of the crystal growth zone. Association type: Define the type of association based on physical and process significance, such as temperature influence, pressure influence, etc.

[0102] Through the identification and analysis of key paths, important nodes, important edges and cross-partition associations, the operation process of the continuous freezing crystallization kettle can be fully understood and optimized. These analysis results can help formulate more scientific and efficient design and operation plans, and improve production efficiency, product quality and system stability.

[0103] As a preferred embodiment of the above, the size association model is one of a multi-layer perceptron, a convolutional neural network and a recurrent neural network. Various neural network models have different advantages in different application scenarios:

[0104] Multilayer Perceptron (MLP): Suitable for processing structured data, with strong flexibility and easy implementation.

[0105] Convolutional Neural Network (CNN): Suitable for processing data with spatial structure, with strong feature extraction capability, parameter sharing, and high computational efficiency.

[0106] Recurrent Neural Network (RNN): Suitable for processing sequence data, can capture time dependencies, has strong memory, and is suitable for data with dynamic input length.

[0107] Selecting a suitable neural network model according to specific tasks and data characteristics can give full play to its advantages and improve the prediction accuracy and adaptability of the size association model.

[0108] Among them, a more specific approach is to establish a size association model for each partition based on the association relationship, including: standardizing the feature matrix so that the value range of each feature is similar to obtain a standardized matrix; through the standardized matrix, a deep learning model is selected to train the size association model. By standardizing the feature matrix and selecting a suitable deep learning model to train the size association model, the prediction accuracy and adaptability of the model can be significantly improved. This solution can comprehensively analyze and optimize the operation process of the continuous freezing crystallization kettle, provide a scientific basis for equipment design and operation, and ensure that each partition works collaboratively and efficiently.

[0109] Embodiment 2

[0110] The dimension design system of continuous freezing crystallization kettle includes:

[0111] The information collection module determines the production target information of the material in each partition of the continuous freezing crystallization kettle. The production target information is based on the set material status at the inlet and outlet of the continuous freezing crystallization kettle.

[0112] An association relationship establishment module is used to establish association relationships among partitions;

[0113] Model training module, which establishes the size association model of each partition according to the association relationship, and trains and optimizes the size association model;

[0114] The scheme output module uses the size correlation model to realize the size scheme prediction of the continuous freezing crystallization kettle.

[0115] The technical effects achieved by this embodiment are the same as those of the above embodiments and will not be described again here.

[0116] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for designing the size of a continuous freezing crystallization kettle, characterized in that: include: Determine the production target information of the material in each partition in the continuous freezing crystallization kettle, wherein the production target information is based on the set material state at the inlet and outlet of the continuous freezing crystallization kettle; Establishing the association relationship between each partition includes: collecting production target information data of the heat exchange crystallization zone, the crystal growth zone and the crystal output zone under different operating conditions, as well as the historical operation data and experimental data of the equipment, and pre-processing each data to obtain a data set to be used; analyzing the association relationship between each partition according to the data set to be used; Establishing a size association model for each partition according to the association relationship, and training and optimizing the size association model; identifying the association relationship between the partitions according to the comprehensive relationship diagram, and creating a feature matrix as an input of the size association model; The size correlation model is applied to the size scheme prediction of the continuous freezing crystallization kettle.

2. The method for designing the size of a continuous freezing crystallization kettle according to claim 1, wherein: Analyzing the association relationship between the partitions according to the data set to be used includes: Dividing the data set to be used into three independent partition data sets according to each partition; The Pearson correlation coefficient was used to calculate the correlation of variables within each partitioned data set; In each partition, x i is the ith observation value of variable x, is the average value of each variable x, y i is the ith observation value of variable y, is the average value of each variable y, -1≤r≤1; Constructing a correlation matrix according to the Pearson correlation coefficient obtained for each partition; A comprehensive relationship diagram across partitions is constructed based on each of the correlation matrices, and the association relationship between the partitions is identified based on the comprehensive relationship diagram.

3. The method for designing the size of a continuous freezing crystallization kettle according to claim 2, wherein: Build a comprehensive relationship graph across partitions, including: Determine the graph structure, define each variable as a node in the graph structure, and the correlation between two variables as the edge of the graph structure; According to the correlation matrix of each partition, sub-graphs of heat exchange crystallization zone, crystal growth zone and crystal out zone are generated respectively; adding edges across partitions according to the physical and process significance of the variables in each of the correlation matrices; The subgraphs of each partition and the edges across partitions are merged to construct the comprehensive relationship graph.

4. The method for designing the size of a continuous freezing crystallization kettle according to claim 3, wherein: The association relationship between the partitions is identified based on the comprehensive relationship graph, including key path analysis, important node and edge analysis, and cross-partition association analysis.

5. The method for designing the size of a continuous freezing crystallization kettle according to claim 4, wherein: It also includes processing the key path analysis, important node and edge analysis, and cross-partition association analysis, and obtaining at least the following contents: The starting and ending partitions of the key paths, the variables and weights in the paths, and the length of the paths; The partitions, types, and importance measures of important nodes and edges; The starting and ending partitions of the cross-partition association, the associated variables and weights, and the type of association; A feature matrix is ​​created, and each of the contents is embedded in the feature matrix as an input of the size association model.

6. The method for designing the size of a continuous freezing crystallization kettle according to claim 5, wherein: The size association model is one of a multi-layer perceptron, a convolutional neural network and a recurrent neural network.

7. The method for designing the size of a continuous freezing crystallization kettle according to claim 5, wherein: A size association model of each partition is established according to the association relationship, including: Standardizing the feature matrix so that the value range of each feature is similar, thereby obtaining a standardized matrix; Through the standardized matrix, a deep learning model is selected to train the size association model.

8. The method for designing the size of a continuous freezing crystallization kettle according to claim 3, wherein: Adding cross-partition edges according to the physical and process meanings of the variables in each of the correlation matrices includes: determining key variables between different partitions according to the physical and process meanings of the variables, and defining the relationships between the key variables; determining cross-partition edges that need to be added to the graph structure according to the key variables and the relationships.

9. The dimension design system of continuous freezing crystallization kettle is characterized by: include: An information collection module determines the production target information of the material in each partition in the continuous freezing crystallization kettle, and the production target information is based on the set material status at the inlet and outlet of the continuous freezing crystallization kettle; The association relationship establishment module establishes the association relationship between each partition, including: collecting the production target information data of the heat exchange crystallization zone, the crystal growth zone and the crystal output zone under different operating conditions, as well as the historical operation data and experimental data of the equipment, and pre-processing each data to obtain the data set to be used; analyzing the association relationship between each partition according to the data set to be used; A model training module is used to establish a size association model for each partition according to the association relationship, and to train and optimize the size association model; to identify the association relationship between the partitions according to the comprehensive relationship diagram, and to create a feature matrix as an input of the size association model; The scheme output module applies the size association model to realize the size scheme prediction of the continuous freezing crystallization kettle.

Images