A method and system for continuous casting adjustment by mold assembly simulation

Abstract

The application discloses a continuous casting adjustment method and system for mold assembly simulation, and relates to the technical field of continuous casting process optimization.The method comprises the following steps: collecting network data according to the equipment ID and the liquid steel component information of the mold assembly to obtain continuous casting historical data; expanding the working conditions of the continuous casting historical data to obtain continuous casting coverage data; establishing an initial finite element model; calibrating boundary conditions to obtain a continuous casting finite element model; performing clustering analysis to locate the starting point of the continuous casting optimization difference; starting parallel local iterative optimization to obtain a continuous casting multi-objective balanced solution set; and performing global iterative optimization to obtain a continuous casting adjustment control benchmark.The application solves the technical problem that the shell uniformity and thickness are difficult to accurately control in the prior art, which leads to unstable continuous casting quality, and achieves the technical effect of indirectly optimizing the control of the shell uniformity and thickness in the continuous casting process and improving the continuous casting quality.

Description

Technical Field

[0001] The present invention relates to the technical field of continuous casting process optimization, and in particular to a continuous casting adjustment method and system for crystallizer assembly simulation. Background Art

[0002] In modern steel production, the continuous casting process, as a key link, has a crucial impact on billet quality, production efficiency, and energy consumption. However, the current continuous casting process faces numerous challenges. First, there is a lack of precise and effective means to control the uniformity and thickness of the embryo shell. Traditional methods are often difficult to adjust to complex and changing operating conditions, resulting in inconsistent embryo shell quality, which affects subsequent steel processing performance and product quality. Second, it is difficult to balance continuous casting efficiency with heat dissipation energy consumption. Increasing casting speeds in pursuit of efficiency often results in insufficient heat dissipation, increasing energy consumption while also reducing product quality.

[0003] The existing technology has a technical problem that the uniformity and thickness of the embryo shell are difficult to accurately control during the continuous casting process, resulting in unstable continuous casting quality. Summary of the Invention

[0004] The present application provides a continuous casting adjustment method and system for crystallizer assembly simulation, which is used to solve the technical problem in the prior art that the uniformity and thickness of the embryo shell during the continuous casting process are difficult to accurately control, resulting in unstable continuous casting quality.

[0005] In view of the above problems, the present application provides a continuous casting adjustment method and system for crystallizer assembly simulation.

[0006] A first aspect of the present application provides a continuous casting adjustment method for a mold assembly simulation, the method comprising:

[0007] Networked data collection is performed based on the equipment ID and molten steel component information of the crystallizer assembly to obtain continuous casting historical data; linear interpolation is used to expand the working conditions of the continuous casting historical data to obtain continuous casting coverage data; component structure information is locally called according to the equipment ID to establish an initial finite element model; the boundary conditions of the initial finite element model are calibrated using the continuous casting coverage data to obtain a continuous casting finite element model; cluster analysis is performed on the continuous casting historical data based on the embryo shell quality index to locate W continuous casting optimization difference starting points; the continuous casting finite element model is copied to W continuous casting thermal simulation containers; after the W continuous casting optimization difference starting points are input into the W continuous casting thermal simulation containers, parallel local iterative optimization is initiated to obtain a continuous casting multi-objective equilibrium solution set; the solution with the lowest energy consumption is selected from the continuous casting multi-objective equilibrium solution set as a benchmark solution, and the benchmark solution is input into the continuous casting finite element model for global iterative optimization to obtain a continuous casting adjustment control benchmark.

[0008] In a second aspect of the present application, a meteorological data-based atmospheric pollution diffusion path tracing system is provided, the system comprising:

[0009] A continuous casting historical data acquisition module is configured to acquire network data based on the equipment ID of the crystallizer assembly and the molten steel component information, and obtain continuous casting historical data; a working condition expansion module is configured to expand the working condition of the continuous casting historical data by linear interpolation, and obtain continuous casting coverage data; a model establishment module is configured to locally call component structure information based on the equipment ID, and establish an initial finite element model; a continuous casting finite element model acquisition module is configured to calibrate the boundary conditions of the initial finite element model by using the continuous casting coverage data, and obtain a continuous casting finite element model; a cluster analysis module is configured to perform cluster analysis on the continuous casting historical data based on the shell quality index, and locate W continuous casting optimization difference starting points; a model replication module is configured to replicate the continuous casting finite element model to W continuous casting thermal simulation containers; a local iterative optimization module is configured to input the W continuous casting optimization difference starting points into the W continuous casting thermal simulation containers, and start parallel local iterative optimization to obtain a continuous casting multi-objective balanced solution set; and a global iterative optimization module is configured to select the solution with the lowest energy consumption from the continuous casting multi-objective balanced solution set as a benchmark solution, and input the benchmark solution into the continuous casting finite element model for global iterative optimization to obtain a continuous casting adjustment control benchmark.

[0010] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0011] The equipment ID of the crystallizer assembly and the molten steel component information are used to acquire network data, and continuous casting historical data is obtained; linear interpolation is used to expand the working condition of the continuous casting historical data, and continuous casting coverage data is obtained; an initial finite element model is established; the boundary conditions of the initial finite element model are calibrated, and a continuous casting finite element model is obtained; cluster analysis is performed to locate W continuous casting optimization difference starting points; the continuous casting finite element model is replicated to W continuous casting thermal simulation containers; after the W continuous casting optimization difference starting points are input into the W continuous casting thermal simulation containers, parallel local iterative optimization is started to obtain a continuous casting multi-objective balanced solution set; the solution with the lowest energy consumption is selected from the continuous casting multi-objective balanced solution set as a benchmark solution, and the benchmark solution is input into the continuous casting finite element model for global iterative optimization to obtain a continuous casting adjustment control benchmark. The technical effect of indirectly optimizing the shell uniformity and thickness during the continuous casting process and improving the continuous casting quality is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort based on these drawings.

[0013] Figure 1 A flow chart of a continuous casting adjustment method provided by the embodiment of the present application for a crystallizer assembly simulation is shown in the figure.

[0014] Figure 2 A structure diagram of a continuous casting adjustment system provided by the embodiment of the present application for a crystallizer assembly simulation is shown in the figure.

[0015] Legend: continuous casting historical data acquisition module 10, working condition expansion module 20, model establishment module 30, continuous casting finite element model acquisition module 40, clustering analysis module 50, model replication module 60, local iterative optimization module 70, and global iterative optimization module 80. DETAILED DESCRIPTION

[0016] The present application provides a continuous casting adjustment method and system for a crystallizer assembly simulation, which is used to solve the technical problem that the shell uniformity and thickness are difficult to be accurately controlled in the continuous casting process in the prior art, resulting in unstable continuous casting quality.

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort fall within the scope of protection of the present application.

[0018] Embodiment one, as shown in the figure, the present application provides a continuous casting adjustment method for a crystallizer assembly simulation, which comprises: Figure 1 Step S100: network data acquisition is performed according to the equipment ID of the crystallizer assembly and the steel liquid component information, and continuous casting historical data is obtained.

[0019]

[0020] ​Specifically, the mold assembly's device ID is first identified. This ID serves as the device's identifier, enabling precise location of the target mold. Simultaneously, information on the composition of the molten steel involved in the continuous casting process is obtained, as the chemical composition of the molten steel affects its solidification characteristics and the casting results. Based on this information, networked data collection is initiated. Leveraging network connections with various sensors, monitoring equipment, and databases on the production line, historical data generated during the continuous casting process is extensively collected. This data covers several key parameters, including multiple sets of water injection volume data, which is directly related to the cooling effect of the crystallizer. Different water injection volumes can affect the solidification rate of the molten steel; scale wiping frequency data, which reflects the cleaning and maintenance of the crystallizer. The appropriate scale wiping frequency ensures the cleanliness of the crystallizer interior and prevents scale from affecting cooling efficiency; water refinement component angle data, which changes the angle and dispersion of the water flow, thereby affecting the uniformity of the molten steel cooling; injection rate data, which controls the rate at which the molten steel enters the crystallizer and plays a significant role in the forming quality of the ingot; and shell uniformity and shell thickness data. These two parameters directly reflect the quality of the continuous casting product. A uniform shell and appropriate thickness are important indicators of high-quality ingots. By comprehensively collecting this data, historical continuous casting data was obtained.

[0021] Step S200: using linear interpolation to expand the working conditions of the continuous casting history data to obtain continuous casting coverage data.

[0022] Specifically, continuous casting adjustment constraints are determined based on component structural information. These constraints include water spray volume ranges, scale wiping frequency ranges, and water refinement component angle ranges. After presetting the multidimensional interpolation scale, multidimensional grid interpolation is performed within the continuous casting adjustment constraints to generate gridded interpolated data. Next, the gridded interpolated data is traversed using the continuous casting history data to locate multiple missing data points. Using these missing data points as a guide, the continuous casting history data is expanded for adjacent operating conditions to obtain multiple sets of updated operating condition data. Based on the adjacent operating condition relationships between the continuous casting history data and the multiple sets of updated operating condition data, the embryo shell quality is predicted using a linear calculation of intermediate values ​​to obtain multiple sets of updated embryo shell qualities. The multiple sets of updated operating condition data and the multiple sets of updated embryo shell qualities are associated and stored to obtain multiple sets of operating condition prediction data. These operating condition prediction data, together with the continuous casting history data, constitute the continuous casting coverage data. This continuous casting coverage data can more comprehensively reflect the various parameters under different operating conditions during the continuous casting process, providing strong support for the subsequent establishment of a more accurate continuous casting model and optimization of the continuous casting process.

[0023] Step S300: Locally call component structure information according to the device ID to establish an initial finite element model.

[0024] Specifically, precise device ID positioning enables the retrieval of component structural information corresponding to the mold assembly from local storage, providing a key basis for building the finite element model. This component structural information details the geometry, dimensions, and interconnections of each mold component. Using this data, an initial finite element model is constructed, encompassing multiple functional modules: a cooling water injection dynamics module, a self-cleaning mechanics module, a droplet splash optimization module, and a steel solidification thermodynamics module. The cooling water injection dynamics module simulates the cooling water injection process within the mold, analyzing the effects of flow velocity, pressure distribution, and other factors on cooling efficiency. The self-cleaning mechanics module focuses on the mold cleaning mechanism during continuous casting, studying the effects of operations such as wiping on equipment operation. The droplet splash optimization module simulates and optimizes the splashing of cooling water droplets impacting the mold and molten steel. The steel solidification thermodynamics module focuses on simulating the solidification process within the mold, including key parameters such as temperature changes and solidification rate. By integrating these modules, an initial finite element model is established that comprehensively reflects the physical phenomena of the continuous casting process.

[0025] Step S400: using the continuous casting coverage data to calibrate the boundary conditions of the initial finite element model to obtain a continuous casting finite element model.

[0026] Specifically, the initial finite element model established based on the component structure information includes a cooling water injection dynamics module, a self-cleaning mechanics module, a droplet sputtering optimization module, and a molten steel solidification thermodynamics module. The continuous casting coverage data is then decomposed according to the requirements of each module to obtain a cooling water injection data set, a self-cleaning mechanics data set, a droplet sputtering data set, and a molten steel solidification data set. These data sets are mapped and loaded into the corresponding modules for individual module calibration. For example, the cooling water injection data set is loaded into the cooling water injection dynamics module to calibrate boundary conditions such as water flow velocity and injection angle; the self-cleaning mechanics data set is used to calibrate boundary conditions for related parameters such as wiping frequency and wiping force in the self-cleaning mechanics module. After completing the individual module calibration, the initial finite element model is collaboratively verified using multi-physics field coupling. This process takes into account the interactions and influences between various physical fields, such as the coupling of the cooling water flow field and the molten steel solidification temperature field, and the mutual influence between the self-cleaning process and the droplet sputtering process. After rigorous calibration and verification, the output is a continuous casting finite element model that can more accurately reflect the actual continuous casting conditions, providing a reliable model basis for subsequent continuous casting process analysis and optimization.

[0027] Step S500: performing cluster analysis on the continuous casting history data according to the embryo shell quality index to locate W continuous casting optimization difference starting points.

[0028] Specifically, historical continuous casting data is analyzed in depth based on the shell quality index to identify the starting point for continuous casting optimization. Shell quality indicators primarily address shell uniformity and shell thickness requirements, both of which directly impact the quality of continuous casting products. During cluster analysis, the Euclidean distance between the shell quality index and multiple shell quality records (including historical shell uniformity and shell thickness) in the continuous casting data is calculated. Based on this distance, H continuous casting condition records are extracted from a large set of continuous casting condition records (consisting of water injection volume parameters, scale wiping frequency parameters, water refinement component angle parameters, and injection speed parameters) in descending order of distance, with H ≥ 20W (where H and W are positive integers). These H continuous casting condition records are then combined and enumerated to obtain multiple sets of continuous casting condition record combinations. These H continuous casting condition records are used as topological nodes, and the Euclidean distance between each set of continuous casting condition record combinations is used as the topological length to construct a continuous casting condition topology. Next, a topological line removal operation is performed on the continuous casting condition topology using a preset operating condition topological distance. After screening, W isolated nodes are obtained. The W continuous casting condition records corresponding to these isolated nodes are the starting points for continuous casting optimization differences. Identifying these starting points facilitates subsequent optimization analysis for different operating conditions, thereby improving the overall quality of the continuous casting process and achieving a balanced optimization of continuous casting efficiency and product quality.

[0029] Step S600: copying the continuous casting finite element model to W continuous casting thermal simulation containers.

[0030] Specifically, after the continuous casting finite element model is successfully constructed through the previous steps, in order to realize multi-working condition parallel optimization analysis, the model needs to be copied to W continuous casting thermal simulation containers. These W continuous casting thermal simulation containers are the basic carriers for subsequent parallel local iterative optimization. Since there are many different working conditions in the continuous casting production process, each continuous casting thermal simulation container can independently simulate a working condition scenario. The continuous casting finite element model is completely copied to each container to ensure that each container has a comprehensive and consistent model foundation, including the cooling water injection dynamics module, self-cleaning mechanics module, droplet sputtering optimization module and molten steel solidification thermodynamics module in the model. In this way, each continuous casting thermal simulation container can simulate and analyze the continuous casting process under its own working condition settings, providing a parallel computing environment foundation for the subsequent input of continuous casting optimization difference starting points and parallel local iterative optimization, which helps to improve the efficiency of optimization analysis, explore the optimization direction of the continuous casting process from multiple angles at the same time, and ultimately achieve multi-objective balanced optimization of the continuous casting process.

[0031] Step S700: After the W continuous casting optimization difference starting points are input into the W continuous casting thermal simulation containers, parallel local iterative optimization is started to obtain a continuous casting multi-objective equilibrium solution set.

[0032] Specifically, after completing the collection and processing of continuous casting historical data and the positioning of the continuous casting optimization difference starting point, the parallel local iterative optimization stage is entered. First, in order to ensure efficient calculation, independent computing resources are allocated to each continuous casting thermal simulation container, including CPU / GPU cores and memory. These independent computing resources enable each continuous casting thermal simulation container to carry out calculations simultaneously and without interfering with each other, greatly improving the overall computing efficiency. The W continuous casting optimization difference starting points previously located are input into the corresponding W continuous casting thermal simulation containers respectively. Each continuous casting optimization difference starting point represents a continuous casting working condition combination with potential optimization value. After entering the container, taking one of the continuous casting thermal simulation containers as an example, the first continuous casting optimization difference starting point is input into the first continuous casting thermal simulation container for unit continuous casting production simulation. During the simulation process, the pre-constructed simulation solution evaluation weights are used to comprehensively consider factors such as energy consumption, embryo shell uniformity deviation, embryo shell thickness deviation, scale residue and water consumption, and the continuous casting performance is quantified to output the first simulation solution coefficient. Afterwards, the preset continuous casting condition search scale is used to update the continuous casting optimization difference starting point to obtain the first updated continuous casting condition, and the continuous casting performance of the first continuous casting thermal simulation container is quantified again, and the second simulation solution coefficient is output. If the second simulation solution coefficient is better than the first simulation coefficient, the continuous casting condition search scale is updated according to the deviation between the two to obtain the first updated search scale, and then the continuous casting condition is adjusted and simulated again, and this process is continued. This cycle is repeated, and the continuous casting condition search scale is continuously updated and the continuous casting performance is quantitatively evaluated according to the simulation solution coefficients obtained by the adjacent optimization, until the deviation of the obtained adjacent optimization simulation solution coefficients is less than the preset threshold, and the first local iterative optimal solution is output. For the remaining W-1 continuous casting thermal simulation containers, the iterative optimization work is also carried out in parallel in the same way. Finally, the W local iterative optimal solutions output by each of the W continuous casting thermal simulation containers together constitute the continuous casting multi-objective equilibrium solution set. This solution set comprehensively considers multiple objectives such as continuous casting quality, efficiency, and heat dissipation energy consumption. It provides a rich and diverse optimization scheme for subsequent screening out the solution with the lowest energy consumption from the solution set, and further inputting it into the continuous casting finite element model for global iterative optimization, and determining the continuous casting adjustment control benchmark, which effectively promotes the overall optimization process of the continuous casting process.

[0033] Step S800: selecting a solution with the lowest energy consumption from the continuous casting multi-objective equilibrium solution set as a benchmark solution, and inputting the benchmark solution into the continuous casting finite element model for global iterative optimization to obtain a continuous casting adjustment control benchmark.

[0034] Specifically, from the continuous casting multi-objective balanced solution set, the continuous casting multi-objective balanced solution set contains multiple balanced optimization solutions in energy consumption, shell quality and other aspects, and energy consumption is an important consideration factor in continuous casting production. By comparing the energy consumption corresponding to each solution in the solution set, the solution with the lowest energy consumption is selected as the reference solution. This reference solution represents the best continuous casting condition setting combination in terms of energy consumption in the previous local iterative optimization. Subsequently, the reference solution is input into the continuous casting finite element model, which integrates cooling water injection dynamics, self-cleaning mechanics, droplet splashing optimization and steel liquid solidification thermodynamics and other modules, and can comprehensively simulate the continuous casting process. With the reference solution as the initial condition, global iterative optimization is carried out in the model. During the iteration process, various parameters in the model are adjusted, such as water injection amount, scale removal frequency, water refining component angle, injection speed, etc., and the influence of these parameter changes on continuous casting quality, efficiency and energy consumption is comprehensively considered. After multiple iterations, the model gradually converges to a more optimal state, and finally the continuous casting adjustment control reference is obtained. This reference provides accurate adjustment basis for actual continuous casting production, helping operators to control the water injection system and intermediate ladle injection speed, etc., to realize the balance of continuous casting efficiency and heat dissipation energy consumption while ensuring the quality of continuous casting, and optimize the continuous casting production process.

[0035] In one possible implementation manner, the step S200 further includes:

[0036] Step S210: defining a continuous casting adjustment constraint according to the component structure information, wherein the continuous casting adjustment constraint includes a water injection amount interval, a scale removal frequency interval and a water refining component angle interval.

[0037] Step S220: presetting a multi-dimensional interpolation scale, and performing multi-dimensional grid interpolation within the continuous casting adjustment constraint range with the multi-dimensional interpolation scale as a constraint to generate grid interpolation data.

[0038] Step S230: traversing the grid interpolation data using the continuous casting historical data to locate a plurality of missing data points.

[0039] Step S240: covering and expanding the continuous casting historical data according to the plurality of missing data points to obtain continuous casting prediction data, wherein the continuous casting prediction data includes a plurality of working condition prediction data, and the continuous casting prediction data and the continuous casting historical data constitute the continuous casting coverage data.

[0040] Specifically, the structural information of the mold assembly components is read. This information covers key aspects such as the design specifications, material properties, and interconnected connections of each part of the mold, and determines the adjustment constraint range during the continuous casting process. The water spray volume range is limited based on the cooling requirements of the mold and the impact of water flow on the solidification process of the molten steel. Excessive or insufficient water spray volume may lead to uneven cooling, affecting the quality of the embryo shell and continuous casting efficiency. The scale wiping frequency range is related to the cleaning and maintenance of the interior of the mold. A reasonable scale wiping frequency can prevent scale accumulation from affecting the cooling effect, while also preventing unnecessary wear on the equipment due to excessive scale wiping. The angle range of the water refinement component is determined because this angle changes the spray direction and dispersion of the water flow, thereby affecting the cooling uniformity of the molten steel and the quality of the ingot. By defining these three ranges, clear boundary conditions are provided for subsequent continuous casting process adjustments and optimizations, ensuring that the continuous casting process is carried out within a reasonable parameter range.

[0041] A pre-set multidimensional interpolation scale is established based on the complexity of the continuous casting process and the required data accuracy. It comprehensively considers multiple key factors influencing the continuous casting process, such as water injection volume, scrubbing frequency, and the angle of the water refinement component, and serves as an important basis for multidimensional grid interpolation. After the multidimensional interpolation scale is set, operations are carried out within the defined continuous casting adjustment constraints. This constraint range includes the water injection volume range, the scrubbing frequency range, and the water refinement component angle range, which limits the reasonable range of data variation. Within this range, multidimensional grid interpolation is performed using the pre-set multidimensional interpolation scale as a constraint. This means that the system interpolates the data in multiple dimensions, establishing connections between dimensions such as water injection volume, scrubbing frequency, and the angle of the water refinement component. Through these calculations, gridded interpolation data is generated.

[0042] Using historical continuous casting data as a reference, a comprehensive and detailed traversal analysis of the gridded interpolated data was performed. This historical data, which contains multiple sets of key information such as water injection volume, scale wiping frequency, water refinement component angle, injection speed, embryo shell uniformity, and embryo shell thickness, serves as a record of the actual continuous casting production process. During the traversal, each data point in the historical continuous casting data was compared one by one with the corresponding position data in the gridded interpolated data. This point-by-point comparison identified data points in the gridded interpolated data that had no corresponding actual record in the historical continuous casting data; these data points are referred to as missing data points.

[0043] Guided by these missing data points, the continuous casting historical data is augmented with adjacent operating conditions. This involves referencing the operating condition information near the missing data point and, based on the historical data, fine-tuning parameters such as water spray volume, scrubbing frequency, and water refinement component angle to generate multiple sets of updated operating condition data. For example, if the water spray volume in the historical operating condition near a missing data point is X, then during augmentation, a new water spray volume value can be obtained by adjusting the value around X by a certain percentage. This is repeated to determine other parameters, forming new sets of operating condition data. Next, based on the adjacent operating condition relationships between the continuous casting historical data and the multiple sets of updated operating condition data, a linear calculation of intermediate values ​​is used to predict the shell quality. For each set of updated operating condition data, the two closest operating condition data points in the historical data are found. Based on the shell uniformity and shell thickness corresponding to these two operating condition data points, the predicted shell uniformity and shell thickness corresponding to the updated operating condition data are calculated using a linear relationship, thereby generating multiple sets of updated shell quality data. Finally, the multiple sets of updated operating condition data and the corresponding sets of updated shell quality are stored in association. Each set of updated operating condition data and the corresponding updated shell quality constitute a complete information unit. These information units are multiple operating condition prediction data, which together form the continuous casting prediction data. The continuous casting prediction data, combined with the existing continuous casting history data, constitutes the continuous casting coverage data, providing more comprehensive and rich data support for subsequent more accurate analysis and optimization of the continuous casting process.

[0044] In one possible implementation, step S240 further includes:

[0045] Step S241: using the multiple missing data points as a guide, expanding the continuous casting historical data by adjacent working conditions to obtain multiple groups of updated working condition data.

[0046] Step S242: According to the adjacent working condition relationship between the continuous casting historical data and the multiple sets of updated working condition data, the embryo shell quality is predicted by using the linear calculation intermediate value to obtain multiple sets of updated embryo shell qualities.

[0047] Step S243: associating the stored multiple sets of updated working condition data with the multiple sets of updated embryo shell qualities to obtain the multiple working condition prediction data as the continuous casting prediction data.

[0048] Specifically, missing data points are used as clues to conduct an in-depth analysis of the continuous casting history data. For each missing data point, the data point that most closely matches the operating condition is searched for in the continuous casting history data. For example, if the missing data point involves a water spray volume of X, a scale wiping frequency of Y, and a water refinement component angle of Z, the continuous casting history data is searched for data records with similar water spray volume, scale wiping frequency, and water refinement component angle. After finding similar data points, these similar data points are used as a basis to make minor adjustments to various parameters, such as appropriately increasing or decreasing the water spray volume, changing the scale wiping frequency, and fine-tuning the water refinement component angle. This generates multiple sets of updated operating condition data, making the operating condition data richer and more diverse.

[0049] The process of predicting the shell quality begins by fully utilizing the adjacent operating condition relationships between the continuous casting history data and multiple sets of updated operating condition data. For each set of updated operating condition data, the two most similar operating condition data are found in the continuous casting history data. The principle of linear calculation of intermediate values ​​is based on the assumption that the change in shell quality (shell uniformity and thickness) between adjacent operating conditions is linear. In other words, the change in shell quality is assumed to be uniform from a water spray rate of 900 L / min to 1100 L / min. The specific calculation process is as follows: assuming that the shell uniformity at a water spray rate of 900 L / min is U1 and the shell thickness is T1; and that the shell uniformity at a water spray rate of 1100 L / min is U2 and the shell thickness is T2. For the inserted update condition of 1000 L / min water spray rate, the shell uniformity U can be calculated using the formula U = U1 + (1000 - 900) / (1100 - 900) × (U2 - U1); the shell thickness T can be calculated using the formula T = T1 + (1000 - 900) / (1100 - 900) × (T2 - T1). This linear calculation method can be used to determine the shell uniformity and thickness under the update condition of 1000 L / min water spray rate, thereby obtaining a set of updated shell masses. The same method is used to calculate data points for other similar update conditions, resulting in multiple sets of updated shell masses, providing more comprehensive data support for subsequent continuous casting process analysis and optimization.

[0050] After obtaining multiple sets of updated operating condition data and updated shell quality data through the initial steps, these two data types are integrated and correlated, precisely matching each set of updated operating condition data with the corresponding updated shell quality data. For example, updated operating condition data for a specific water spray volume, a certain range of scrubbing frequency, and a specific angle of the water refinement assembly are mapped to updated shell quality data for the corresponding shell uniformity and thickness, derived from linearly calculated intermediate values. Through this one-to-one mapping, each set of updated operating condition data and updated shell quality data is combined into a complete information unit, which in turn becomes multiple sets of operating condition prediction data. These operating condition prediction data are then stored in an orderly manner according to specific storage rules, ultimately forming continuous casting prediction data.

[0051] In one possible implementation, step S241 further includes:

[0052] The continuous casting historical data includes multiple continuous casting condition records and multiple embryo shell quality records. The continuous casting condition records are composed of water spraying volume parameters, scale scraping frequency parameters, water refinement component angle parameters, and injection speed parameters. The embryo shell quality records are composed of historical embryo shell uniformity and historical embryo shell thickness.

[0053] Specifically, continuous casting historical data, as an important basis for continuous casting process optimization analysis, includes multiple continuous casting condition records and multiple embryo shell quality records. Among them, the continuous casting condition record records the key parameters of the continuous casting process in detail. The water spray volume parameter directly affects the cooling rate of the molten steel in the crystallizer. The appropriate water spray volume can ensure uniform cooling of the molten steel and form a high-quality embryo shell. The scale wiping frequency parameter is related to the cleanliness of the interior of the crystallizer. Regular scale wiping can prevent scale accumulation from affecting the cooling effect. Different scale wiping frequencies have different degrees of impact on the stability of the continuous casting process and product quality. The water refinement component angle parameter determines the spray direction and dispersion of the cooling water. A reasonable angle can make the cooling water more evenly act on the molten steel and promote the uniform growth of the embryo shell. The injection rate parameter controls the rate at which the molten steel enters the crystallizer. It works together with other parameters to jointly affect the molding quality of the ingot. The shell quality record primarily consists of historical shell uniformity and shell thickness. Shell uniformity reflects the uniformity of the shell during its growth, which helps improve the quality and performance of the ingot. Shell thickness reflects the thickness of the shell after solidification of the molten steel. Appropriate shell thickness is a key factor in ensuring the strength and stability of the ingot. This historical continuous casting data comprehensively records various information about the past continuous casting process.

[0054] In one possible implementation, step S500 further includes:

[0055] Step S510: extracting H continuous casting condition records in descending order of distance from the multiple continuous casting condition records according to the Euclidean distance between the embryo shell quality index and the multiple embryo shell quality records, where H≥20W, and H and W are positive integers.

[0056] Step S520: Enumerate H continuous casting condition records in combination and obtain After the continuous casting condition records are grouped, the H continuous casting condition records are used as topological nodes. The Euclidean distance of the continuous casting condition records is used as the topological length to construct the continuous casting condition topology.

[0057] Step S530: using a preset working condition topology distance to remove topological connections of the continuous casting working condition topology to obtain W isolated nodes.

[0058] Step S540: taking the W continuous casting operating condition records of the W isolated nodes as the W continuous casting optimization difference starting points.

[0059] Specifically, the embryo shell quality index is defined. This index is a key criterion for measuring embryo shell quality during the continuous casting process and primarily addresses standard requirements for key parameters such as embryo shell uniformity and thickness. Furthermore, the continuous casting historical data contains multiple embryo shell quality records, detailing the actual embryo shell uniformity and thickness data generated under different continuous casting conditions. Using the Euclidean distance algorithm, the distance between the embryo shell quality index and each embryo shell quality record is calculated. Euclidean distance accurately measures the degree of difference between two data points in multidimensional space. In this scenario, it measures the difference between the actual embryo shell quality and the ideal embryo shell quality index. This calculation assigns a corresponding Euclidean distance value to each embryo shell quality record. Based on these Euclidean distance values, multiple continuous casting condition records are then screened. These continuous casting condition records include parameters such as water injection volume, scale wiping frequency, water refinement component angle, and injection speed, which collectively determine the actual continuous casting conditions. From these numerous continuous casting condition records, H continuous casting condition records are extracted in descending order of Euclidean distance. It is important to emphasize here that H and W have a quantitative relationship: H ≥ 20W, where both H and W are positive integers. This means that the number of extracted continuous casting condition records is at least 20 times W. A large H value ensures that the selected condition records are sufficiently diverse and representative, covering a variety of different continuous casting conditions. This provides a rich data foundation for subsequently accurately constructing the continuous casting condition topology and locating the starting point of continuous casting optimization differences, facilitating more comprehensive and in-depth analysis and optimization of the continuous casting process.

[0060] The H continuous casting condition records are combined and enumerated. Each continuous casting condition record is a unique condition combination consisting of water spraying parameters, scale scraping frequency parameters, water refinement component angle parameters and injection speed parameters. Through combination enumeration, the H continuous casting condition records are combined in pairs, and finally the result is Combine the continuous casting condition records. Based on these H continuous casting condition records, they are used as topological nodes. Each node represents a specific continuous casting condition, which occupies a unique position in the entire topological structure. Then, for the previously enumerated combination The continuous casting condition records are grouped together, and the Euclidean distance between each group of records is calculated. The Euclidean distance can quantify the degree of difference between different continuous casting conditions. In the continuous casting process scenario, it is a comprehensive measure of the differences in parameters such as water spray volume, scale wiping frequency, water refinement component angle, and injection speed. These calculated Euclidean distances are used as topological lengths. Finally, based on the topological nodes and topological lengths determined above, a continuous casting condition topology is constructed. In this topological structure, each topological node is connected to each other through a topological length, and the topological length reflects the size of the difference between different conditions. This continuous casting condition topology intuitively shows the association and difference between different continuous casting conditions, and provides a clear structural framework for the subsequent use of preset condition topological distances to eliminate topological connections and locate the starting point of continuous casting optimization differences, which helps to deeply analyze the continuous casting process and find the optimization direction.

[0061] Based on the constructed continuous casting condition topology, it is further processed to determine the starting point for continuous casting optimization differences. After the continuous casting condition topology is constructed, it contains H topological nodes (representing H continuous casting condition records) and topological links connecting these nodes (their lengths are determined by the Euclidean distances between the corresponding groups of continuous casting condition records). A preset condition topological distance is introduced. This preset condition topological distance is pre-determined based on practical experience, historical data, and a deep understanding of the continuous casting process. It is used to measure whether the differences between different continuous casting conditions are meaningful. The preset condition topological distance is then compared with the length (i.e., the Euclidean distance) of each topological link in the continuous casting condition topology. Links whose lengths exceed the preset condition topological distance are considered to be too different between the two continuous casting conditions and not closely related for the current analysis. Therefore, these links are removed from the topological structure. This removal process alters the previously interconnected topological structure, with some nodes no longer connected to other nodes, resulting in isolated nodes. W isolated nodes are selected from these isolated nodes. The continuous casting condition records corresponding to these W isolated nodes are of special significance in the subsequent process optimization. They will be used as the starting point for continuous casting optimization differences, providing a key starting point for subsequent exploration of the optimization direction of the continuous casting process, and helping to achieve balanced optimization of continuous casting quality, efficiency and energy consumption.

[0062] The continuous casting condition records corresponding to the W isolated nodes obtained through preliminary processing were identified as W continuous casting optimization difference starting points. Prior to this, a continuous casting condition topology was constructed through cluster analysis of historical continuous casting data. A topological line removal operation was then performed using a preset condition topology distance to screen out these W isolated nodes. Each isolated node represents a unique continuous casting condition record, containing key information such as water injection volume parameters, scale wiping frequency parameters, water refinement component angle parameters, and injection speed parameters. The continuous casting condition records of these isolated nodes were selected as continuous casting optimization difference starting points because they exhibit significant differences compared to other condition records. This difference may indicate that certain aspects of the continuous casting process under these conditions are unique. For example, under specific combinations of parameters such as water injection volume and scale wiping frequency, the quality of the embryo shell will differ from that under conventional conditions. Using these differences as a starting point, targeted simulation and optimization exploration can be conducted in the continuous casting thermal simulation vessel. Through in-depth research and adjustment of these different starting points for continuous casting optimization, it is expected that a better combination of continuous casting process parameters can be discovered, thereby improving the continuous casting quality, while balancing the continuous casting efficiency and heat dissipation energy consumption, and providing strong support for the improvement of the continuous casting process.

[0063] In one possible implementation, step S700 further includes:

[0064] Step S710: inputting the first continuous casting optimization difference starting point into the first continuous casting thermal simulation container to perform unit continuous casting production simulation, then quantifying the continuous casting performance, and outputting the first simulation solution coefficient.

[0065] Step S720: updating the first continuous casting optimization difference starting point using a preset continuous casting condition search scale, outputting a first updated continuous casting condition, quantifying the continuous casting performance of the first continuous casting thermal simulation container, and outputting a second simulation solution coefficient.

[0066] Step S730: If the second simulation solution coefficient is better than the first simulation solution coefficient, the continuous casting condition search scale is updated according to the deviation between the first simulation solution coefficient and the second simulation solution coefficient, and a first updated search scale is output.

[0067] Step S740: Similarly, the continuous casting condition search scale is updated and the continuous casting performance is quantitatively evaluated based on the simulation solution coefficients obtained by the adjacent optimization search, until the deviation of the obtained adjacent optimization simulation solution coefficients is less than a preset threshold, and the first local iterative optimal solution is output.

[0068] Step S750: Similarly, through the parallel local iterative optimization of the W continuous casting thermal simulation containers, W local iterative optimal solutions are obtained to constitute the continuous casting multi-objective equilibrium solution set.

[0069] Specifically, to ensure computational efficiency, independent computing resources, including CPU / GPU cores and memory, are allocated to each continuous casting thermal simulation container, and then the starting point of the first continuous casting optimization difference is input into the first continuous casting thermal simulation container. A unit continuous casting production simulation is performed in the container, and after the simulation, the continuous casting performance is quantitatively evaluated. In the quantification process, simulation solution evaluation weights are pre-constructed, covering energy consumption weights, embryo shell uniformity deviation weights, embryo shell thickness deviation weights, scale residue weights, and water consumption weights. Based on these weights, combined with the simulated output of the first continuous casting energy consumption, the first simulated uniformity, the first simulated thickness, the first scale residue, and the first water consumption, the first uniformity deviation and the first thickness deviation are calculated based on the embryo shell quality index, and finally these data are fused to output the first simulation solution coefficient.

[0070] After completing a preliminary simulation of the first continuous casting optimization differential starting point in the first continuous casting thermal simulation vessel and obtaining the first simulation solution coefficients, the first continuous casting optimization differential starting point is adjusted based on a pre-set continuous casting condition search scale. This search scale, set based on the characteristics of the continuous casting process, past experience, and desired optimization accuracy, dictates the magnitude of each adjustment to the continuous casting condition parameters (such as water injection volume, scale frequency, water refinement component angle, injection speed, etc.). Based on this search scale, the various operating parameters included in the first continuous casting optimization differential starting point are modified accordingly to obtain the first updated continuous casting condition. For example, if the water injection volume is originally X, the search scale setting may adjust it to X±ΔX (ΔX is determined by the search scale), and similar adjustments are made to other parameters. The first updated continuous casting condition is then input into the first continuous casting thermal simulation vessel, and the unit continuous casting production simulation is repeated. After the simulation is complete, similar to the process of calculating the coefficients for the first simulation solution, the pre-established simulation solution evaluation weights are combined with the output data for continuous casting energy consumption, simulated uniformity, simulated thickness, scale residue, and water consumption. The corresponding uniformity deviation and thickness deviation are calculated based on the shell quality index. Finally, these data are integrated and calculated to output the coefficients for the second simulation solution. This second simulation solution coefficient is used to compare with the coefficients for the first simulation solution to determine whether the adjustment of the continuous casting conditions is moving in a more optimal direction, thereby providing a basis for subsequent continuous casting condition search scale updates and continuous casting process optimization.

[0071] When the simulation of the first updated continuous casting condition in the first continuous casting thermal simulation container is completed and the second simulation solution coefficient is obtained, the first and second simulation solution coefficients are compared. If the second simulation solution coefficient is better, it means that the current adjustment direction of the continuous casting condition is correct, and it is advancing towards the goal of improving the continuous casting performance, balancing the continuous casting quality, efficiency and heat dissipation energy consumption. At this time, the search scale of the continuous casting condition will be updated according to the deviation between the first simulation solution coefficient and the second simulation solution coefficient. This deviation is converted into the adjustment amount of the search scale of the continuous casting condition, thereby obtaining the first updated search scale. This updated search scale will be used to adjust the first updated continuous casting condition again. For example, if the water spraying amount, the scale cleaning frequency and other continuous casting condition parameters have specific set values in the first updated continuous casting condition, according to the first updated search scale, these parameters are adjusted accordingly to increase or decrease, and then the second updated continuous casting condition is output. After obtaining the second updated continuous casting condition, it is input into the first continuous casting thermal simulation container to simulate the unit continuous casting production again. After the simulation is completed, the simulation results are quantitatively analyzed according to the pre-constructed simulation solution evaluation weight. These weights include energy consumption weight, shell uniformity deviation weight, shell thickness deviation weight, scale residue weight and water consumption weight, etc. Combined with the data of continuous casting energy consumption, simulation uniformity, simulation thickness, scale residue and water consumption output by the simulation, the corresponding uniformity deviation and thickness deviation are calculated according to the shell quality index, and these data are integrated for calculation, and finally the third simulation solution coefficient is output.

[0072] After obtaining the third simulation solution coefficient, an iterative process begins. During each iteration, the current simulation solution coefficient is compared with the previous one, and the continuous casting condition search scale is updated based on the deviation between the two. For example, if the current simulation solution coefficient indicates improved continuous casting performance, the search scale is adjusted accordingly based on the magnitude of the deviation, making subsequent adjustments to the continuous casting condition more targeted. If the deviation is positive and large, the search scale is appropriately increased to accelerate the search for a more optimal condition. If the deviation is small, the search scale is reduced to more carefully explore potential surrounding conditions that are more optimal. After updating the search scale, the current continuous casting condition is adjusted using the new search scale to generate a new continuous casting condition, and a unit continuous casting production simulation is performed again in the first continuous casting thermal simulation vessel. After the simulation completes, the simulation results are quantitatively evaluated using the simulation solution evaluation weights to obtain the new simulation solution coefficients. This process is repeated continuously, continuously comparing the simulation solution coefficients obtained from the adjacent optimal search, updating the continuous casting condition search scale, and quantitatively evaluating the continuous casting performance. This cycle continues until the deviation between the obtained adjacent optimal simulation solution coefficients is less than a preset threshold. The preset threshold is set based on the accuracy requirements of the continuous casting process and practical experience, representing the degree of optimization deemed acceptable by the system. When this threshold is reached, the current continuous casting condition is determined to be the first local iterative optimal solution and output. This optimal solution is the best condition achieved within the current local search space by comprehensively considering multiple objectives, including continuous casting quality, efficiency, and heat dissipation energy consumption. This lays the foundation for subsequent parallel optimization across multiple continuous casting thermal simulation vessels, ultimately achieving a multi-objective equilibrium solution set for continuous casting.

[0073] After completing the local iterative optimization for the first continuous casting thermal simulation container and obtaining the first local iterative optimal solution, the process is expanded to W continuous casting thermal simulation containers. The system contains W continuous casting thermal simulation containers, each of which is allocated independent computing resources (CPU / GPU cores, memory) to ensure efficient parallel computing. For each container, the local iterative optimization steps previously performed in the first container are repeated. The corresponding continuous casting optimization difference starting point is input into the corresponding continuous casting thermal simulation container for unit continuous casting production simulation. The simulation results are quantified using the simulation solution evaluation weights, and the simulation solution coefficients are output. The continuous casting condition search scale is then continuously updated based on the simulation solution coefficients obtained from the adjacent optimization search. The continuous casting condition is adjusted, and the simulation and quantization are repeated. This cycle continues until the deviation of the obtained adjacent optimization simulation solution coefficients is less than a preset threshold, resulting in the local iterative optimal solution for each container. Due to the parallel computing of each container, these operations can be performed simultaneously, significantly reducing the optimization time cost. Ultimately, the W local iterative optimal solutions output by each of the W continuous casting thermal simulation containers together constitute the continuous casting multi-objective equilibrium solution set. This solution set comprehensively considers multiple objectives such as continuous casting quality, efficiency, and heat dissipation energy consumption, and covers a variety of possible continuous casting working condition combinations. It provides a rich and comprehensive reference basis for subsequently screening out the solution with the lowest energy consumption from the solution set and inputting it into the continuous casting finite element model for global iterative optimization, thereby determining the continuous casting adjustment control benchmark, and effectively promoting the optimization process of the continuous casting process.

[0074] In one possible implementation, step S710 further includes:

[0075] Step S711: pre-constructing simulation solution evaluation weights, wherein the simulation solution evaluation weights include energy consumption weights, embryo shell uniformity deviation weights, embryo shell thickness deviation weights, scale residue weights, and water consumption weights.

[0076] Step S712: inputting the first continuous casting optimization difference starting point into the first continuous casting thermal simulation container to perform unit continuous casting production simulation, and outputting the first continuous casting energy consumption, the first simulated uniformity, the first simulated thickness, the first scale residue and the first water consumption.

[0077] Step S713: performing deviation evaluation on the first simulated uniformity and the first simulated thickness according to the embryo shell quality index, and outputting a first uniformity deviation and a first thickness deviation.

[0078] Step S714: using the simulation solution evaluation weight to fuse the first continuous casting energy consumption, the first uniformity deviation, the first thickness deviation, the first scale residue and the first water consumption, and outputting the first simulation solution coefficient.

[0079] Specifically, simulation solution evaluation weights are pre-established. Before launching the continuous casting thermal simulation, a weighting system is constructed to evaluate simulation results based on long-term accumulated actual data from continuous casting production, specific production process standards, and the company's comprehensive goals for cost control, product quality, and resource utilization. The energy consumption weight in this system is determined based on the proportion of energy consumption in the total cost of the continuous casting process, energy market price fluctuations, and the company's strategic plan for energy conservation and emission reduction. The shell uniformity deviation weight and shell thickness deviation weight are primarily determined based on core requirements for ingot quality. Shell uniformity is directly related to the stability of the ingot's internal structure and surface smoothness, while shell thickness affects the ingot's strength and subsequent processing performance. Therefore, these two weights are precisely assigned based on the company's stringent ingot quality standards and the differentiated ingot quality requirements across different industries. The scale residue weight is determined based on historical data on the accumulation rate of scale on the mold and related equipment surfaces, as well as its impact on cooling efficiency and equipment life. This aims to measure the comprehensive impact of scale on continuous casting production continuity and equipment maintenance costs. Water consumption weights are determined based on local water scarcity, the company's water conservation policies, and sustainable development goals, emphasizing the importance of rational water resource utilization in continuous casting. These weights together form the simulation solution evaluation weights, providing a unified and scientific quantitative evaluation standard for subsequent data derived from continuous casting thermal simulations.

[0080] The optimized differential starting point for the first continuous casting process is input into the first continuous casting thermal simulation vessel to simulate unit continuous casting production. This simulation fully considers various physical phenomena and process parameters during the continuous casting process, ultimately outputting key indicators: first continuous casting energy consumption, first simulated uniformity, first simulated thickness, first scale residue, and first water consumption. These indicators intuitively reflect the performance under the current continuous casting conditions.

[0081] Deviations from the first simulated uniformity and first simulated thickness are evaluated based on pre-set shell quality indicators. Shell quality indicators define the ideal uniformity and thickness of continuous casting shells. By comparing the first simulated uniformity and first simulated thickness with the ideal indicators, the first uniformity deviation and first thickness deviation are calculated. These two deviations quantify the gap between the current shell quality and the ideal state.

[0082] The first simulation solution coefficient is obtained by using the constructed simulation solution evaluation weight to comprehensively operate a plurality of results of the continuous casting thermal simulation. First, a plurality of key data including a first continuous casting energy consumption, a first uniformity deviation, a first thickness deviation, a first scale residue, and a first water consumption are obtained, which reflect the running situation of the continuous casting process under a specific working condition from different aspects. The first continuous casting energy consumption reflects the energy consumption level in the continuous casting process, the first uniformity deviation and the first thickness deviation show the gap between the shell quality and the ideal state, the first scale residue reflects the potential influence of the equipment cleaning degree on the process, and the first water consumption represents the utilization status of water resources. Then, the simulation solution evaluation weight previously constructed is called, which includes an energy consumption weight, a shell uniformity deviation weight, a shell thickness deviation weight, a scale residue weight, and a water consumption weight. The weights are multiplied by the corresponding first continuous casting energy consumption, first uniformity deviation, first thickness deviation, first scale residue, and first water consumption, respectively, to give each data a corresponding proportion in the comprehensive evaluation. For example, if the energy consumption weight is 0.3 and the first continuous casting energy consumption is 500 (unit), the contribution value of the energy consumption in the comprehensive calculation is 500*0.3. Finally, the product results are added together, and after a series of fusion calculations, a value that comprehensively reflects the overall performance of the continuous casting process under the current working condition is obtained, which is the first simulation solution coefficient.

[0083] In one possible implementation manner, the step S400 further includes:

[0084] The step S410: establishing the initial finite element model according to the component structure information, wherein the initial finite element model includes a cooling water injection dynamics module, a self-cleaning mechanics module, a liquid droplet splashing optimization module, and a liquid steel solidification thermodynamics module.

[0085] The step S420: decomposing the continuous casting coverage data according to the module composition of the initial finite element model to obtain a cooling water injection data group, a self-cleaning mechanics data group, a liquid droplet splashing data group, and a liquid steel solidification data group.

[0086] The step S430: after the cooling water injection data group, the self-cleaning mechanics data group, the liquid droplet splashing data group, and the liquid steel solidification data group are mapped and loaded to the cooling water injection dynamics module, the self-cleaning mechanics module, the liquid droplet splashing optimization module, and the liquid steel solidification thermodynamics module for single module calibration, the initial finite element model is subjected to multi-physical field coupling collaborative verification, and the continuous casting finite element model is output.

[0087] Specifically, the component structure information covers the detailed characteristics of each component of the continuous casting equipment, such as the geometry, size, material properties, etc. This information provides the necessary parameters and basis for model construction. The initial finite element model contains four key modules. The cooling water injection dynamics module focuses on simulating the cooling water injection process at the water nozzle. Through the dynamic analysis of the water flow, it verifies the uniformity of the water flow coverage, requiring its standard deviation to be ≤3%. This is crucial to ensure that the molten steel can dissipate heat evenly during the cooling process and thus form a high-quality embryo shell. The self-cleaning mechanics module focuses on simulating the expansion and scale scraping of the water supply ring and the impact process of the scale-breaking components. The impact frequency is set at 10-30 times / minute. Through such simulations, the module aims to ensure a scale removal rate of ≥90% to maintain the cleanliness of the interior of the continuous casting equipment and reduce the adverse effects of scale on the cooling effect and the quality of the ingot. The droplet splashing optimization module mainly calibrates the effect of the elastic ring angle on the splashing uniformity. This module simulates and analyzes the droplet sputtering process, requiring a temperature gradient of ≤50°C / mm to ensure uniform distribution of droplets during the sputtering process, thereby better cooling and protecting the molten steel. The molten steel solidification thermodynamics module primarily adjusts the molten steel injection rate (ranging from 0.5-2.5 m / min). By simulating and analyzing heat transfer and phase transformation during the solidification process, it ensures a shell thickness error of ≤2 mm, thereby guaranteeing the quality and dimensional accuracy of the ingot. By integrating these modules, the initial finite element model can comprehensively and meticulously simulate various phenomena in the continuous casting process from multiple physical levels, providing a powerful tool and platform for subsequent in-depth analysis and optimization of the continuous casting process.

[0088] Based on the modular structure of the initial finite element model, the continuous casting coverage data was decomposed. The continuous casting coverage data is a comprehensive dataset that includes various physical phenomena and process parameters in the continuous casting process. Through the decomposition operation, it was split into four data groups: the cooling water injection data group, which contains parameters related to cooling water injection, such as injection speed and pressure; the self-cleaning mechanics data group, which covers the mechanical data of the self-cleaning process, such as friction and impact force; the droplet splashing data group, which records information related to droplet splashing, such as splashing angle and speed; and the molten steel solidification data group, which contains thermal data of the molten steel solidification process, such as temperature change and solidification time.

[0089] The four data sets obtained by decomposition are respectively mapped and loaded into the corresponding modules of the initial finite element model. The cooling water injection data set is loaded into the cooling water injection dynamics module, the self-cleaning mechanics data set is loaded into the self-cleaning mechanics module, the droplet splashing data set is loaded into the droplet splashing optimization module, and the liquid steel solidification data set is loaded into the liquid steel solidification thermodynamics module. After completing the data loading, calibrate each single module to ensure that the simulation results of the module are consistent with the actual physical phenomenon. For example, adjust the parameters in the module so that the simulated cooling water injection effect is consistent with the actual observed situation. After completing the calibration of the single module, the initial finite element model is subjected to collaborative verification of multi-physical field coupling. In the continuous casting process, different physical phenomena interact with each other, for example, changes in water injection volume will affect the refining angle and uniformity of the water, and in turn affect the cooling effect and solidification process of the liquid steel. Therefore, the interaction between each module needs to be considered comprehensively for collaborative verification. By continuously adjusting and optimizing the model parameters, the model can accurately simulate the multi-physical field coupling phenomenon in the continuous casting process. Finally, after the above series of operations, the continuous casting finite element model is output.

[0090] In one possible implementation, step S200 further includes:

[0091] Step S250: During the continuous casting production process controlled by the mold assembly control unit according to the continuous casting adjustment control reference, real-time shell uniformity and real-time shell thickness are monitored and collected.

[0092] Step S260: Extract the shell uniformity requirement and the shell thickness requirement from the shell quality index.

[0093] Step S270: Perform PID closed-loop feedback control of the continuous casting adjustment control reference according to the first dynamic deviation between the real-time shell uniformity and the shell uniformity requirement, and the second dynamic deviation between the real-time shell thickness and the shell thickness requirement.

[0094] Specifically, when the mold assembly control unit carries out continuous casting production according to the continuous casting adjustment control reference, in order to ensure that the production process always remains in the best state, key indicators are monitored and collected in real time. Real-time shell uniformity and real-time shell thickness are closely monitored, and relevant data are continuously collected through various high-precision sensors installed on the mold assembly. These real-time data can accurately reflect the actual state of the shell in the current continuous casting process.

[0095] The shell uniformity requirement and the shell thickness requirement are extracted from the pre-set shell quality index. The shell quality index is an ideal shell state parameter determined according to the quality standards of continuous casting products, production process requirements and other factors. The shell uniformity requirement and the shell thickness requirement are key parts of it, which represent the shell quality level expected to be achieved in the continuous casting production.

[0096] Real-time data collected from the embryo shell is compared and analyzed against the ideal requirement. The first dynamic deviation between the real-time embryo shell uniformity and the required uniformity, as well as the second dynamic deviation between the real-time embryo shell thickness and the required thickness, are calculated. Based on these two dynamic deviations, a PID closed-loop feedback control mechanism is employed to adjust the continuous casting control baseline. PID control is a commonly used feedback control algorithm that adjusts the control variable based on three parameters: the proportional (P), integral (I), and differential (D) of the deviation. During this process, relevant parameters in the continuous casting process, such as water injection rate and injection speed, are automatically calculated and adjusted based on the magnitude and trend of the dynamic deviation. This allows continuous casting production to continuously approach the ideal embryo shell quality state, thereby ensuring the stability of continuous casting quality while optimizing the balance between continuous casting efficiency and heat dissipation energy consumption.

[0097] Example 2, based on the same inventive concept as the continuous casting adjustment method of a crystallizer assembly simulation in the above embodiment, Figure 2 As shown, the present application provides a continuous casting adjustment system for crystallizer assembly simulation. The system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0098] The continuous casting history data acquisition module 10 is used to collect networked data based on the device ID of the crystallizer assembly and the steel liquid component information to obtain the continuous casting history data.

[0099] The working condition expansion module 20 is used to expand the working condition of the continuous casting history data by using linear interpolation to obtain continuous casting coverage data.

[0100] The model building module 30 is used to locally call component structure information according to the device ID and build an initial finite element model.

[0101] The continuous casting finite element model acquisition module 40 is configured to calibrate the boundary conditions of the initial finite element model using the continuous casting coverage data to obtain a continuous casting finite element model.

[0102] The cluster analysis module 50 is used to perform cluster analysis on the continuous casting history data according to the embryo shell quality index, and locate W continuous casting optimization difference starting points.

[0103] The model replication module 60 is used to replicate the continuous casting finite element model to W continuous casting thermal simulation containers.

[0104] The local iterative optimization module 70 is used to input the W continuous casting optimization difference starting points into the W continuous casting thermal simulation containers, and then start parallel local iterative optimization to obtain a continuous casting multi-objective equilibrium solution set.

[0105] The global iterative optimization module 80 is used to select the solution with the lowest energy consumption from the continuous casting multi-objective equilibrium solution set as a benchmark solution, and input the benchmark solution into the continuous casting finite element model for global iterative optimization to obtain a continuous casting adjustment control benchmark.

[0106] Furthermore, the system is also used to implement the following functions:

[0107] Continuous casting adjustment constraints are defined according to the component structure information, wherein the continuous casting adjustment constraints include a water spray volume range, a scale wiping frequency range, and a water refinement component angle range; a multidimensional interpolation scale is preset, and within the continuous casting adjustment constraint range, multidimensional grid interpolation is performed with the multidimensional interpolation scale as a constraint to generate gridded interpolation data; the continuous casting history data is used to traverse the gridded interpolation data to locate multiple missing data points; based on the multiple missing data points, the continuous casting history data is covered and expanded to obtain continuous casting prediction data, wherein the continuous casting prediction data includes multiple working condition prediction data, and the continuous casting prediction data and the continuous casting history data constitute the continuous casting coverage data.

[0108] Furthermore, the system is also used to implement the following functions:

[0109] Guided by the multiple missing data points, the continuous casting historical data is expanded under adjacent working conditions to obtain multiple groups of updated working condition data; based on the adjacent working condition relationship between the continuous casting historical data and the multiple groups of updated working condition data, the embryo shell quality is predicted using a linear calculation intermediate value to obtain multiple groups of updated embryo shell qualities; the multiple groups of updated working condition data and the multiple groups of updated embryo shell qualities are associated and stored to obtain the multiple working condition prediction data as the continuous casting prediction data.

[0110] Furthermore, the system is also used to implement the following functions:

[0111] The continuous casting historical data includes multiple continuous casting condition records and multiple embryo shell quality records. The continuous casting condition records are composed of water spraying volume parameters, scale scraping frequency parameters, water refinement component angle parameters, and injection speed parameters. The embryo shell quality records are composed of historical embryo shell uniformity and historical embryo shell thickness.

[0112] Furthermore, the system is also used to implement the following functions:

[0113] According to the Euclidean distance between the embryo shell quality index and the plurality of embryo shell quality records, H continuous casting condition records in descending order of distance are extracted from the plurality of continuous casting condition records, wherein H ≥ 20W, H and W are positive integers; after enumerating the H continuous casting condition records in combination, the following is obtained: After the continuous casting condition records are grouped, the H continuous casting condition records are used as topological nodes. The Euclidean distance of the continuous casting condition records is used as the topological length to construct a continuous casting condition topology; the topological connections of the continuous casting condition topology are removed using a preset condition topological distance to obtain W isolated nodes; and the W continuous casting condition records of the W isolated nodes are used as the W continuous casting optimization difference starting points.

[0114] Furthermore, the system is also used to implement the following functions:

[0115] The first continuous casting optimization difference starting point is input into the first continuous casting thermal simulation container for unit continuous casting production simulation, and then the continuous casting performance is quantified, and the first simulation solution coefficient is output; the first continuous casting optimization difference starting point is updated using a preset continuous casting condition search scale, and after the first updated continuous casting condition is output, the continuous casting performance of the first continuous casting thermal simulation container is quantified, and the second simulation solution coefficient is output; if the second simulation solution coefficient is better than the first simulation coefficient, the continuous casting condition search scale is updated according to the deviation between the first simulation solution coefficient and the second simulation solution coefficient, and the first updated search scale is output; and so on, the continuous casting condition search scale is updated and the continuous casting performance is quantitatively evaluated according to the simulation solution coefficients obtained by adjacent optimization, until the deviation of the obtained adjacent optimization simulation solution coefficients is less than a preset threshold, and the first local iterative optimal solution is output; and so on, through the parallel local iterative optimization of the W continuous casting thermal simulation containers, W local iterative optimal solutions are obtained to constitute the continuous casting multi-objective equilibrium solution set.

[0116] Furthermore, the system is also used to implement the following functions:

[0117] Pre-construct simulation solution evaluation weights, wherein the simulation solution evaluation weights include energy consumption weight, embryo shell uniformity deviation weight, embryo shell thickness deviation weight, scale residue weight, and water consumption weight; input the first continuous casting optimization difference starting point into the first continuous casting thermal simulation container to perform unit continuous casting production simulation, and output the first continuous casting energy consumption, first simulated uniformity, first simulated thickness, first scale residue and first water consumption; perform deviation evaluation on the first simulated uniformity and the first simulated thickness according to the embryo shell quality index, and output the first uniformity deviation and the first thickness deviation; use the simulation solution evaluation weights to fuse the first continuous casting energy consumption, first uniformity deviation, first thickness deviation, first scale residue and first water consumption, and output the first simulation solution coefficient.

[0118] Furthermore, the system is also used to implement the following functions:

[0119] The initial finite element model is established according to the component structure information, wherein the initial finite element model includes a cooling water injection dynamics module, a self-cleaning mechanics module, a droplet sputtering optimization module and a molten steel solidification thermodynamics module; the continuous casting coverage data is decomposed according to the module structure of the initial finite element model to obtain a cooling water injection data group, a self-cleaning mechanics data group, a droplet sputtering data group and a molten steel solidification data group; after the cooling water injection data group, the self-cleaning mechanics data group, the droplet sputtering data group and the molten steel solidification data group are mapped and loaded into the cooling water injection dynamics module, the self-cleaning mechanics module, the droplet sputtering optimization module and the molten steel solidification thermodynamics module for single module calibration, the initial finite element model is collaboratively verified by multi-physical field coupling, and the continuous casting finite element model is output.

[0120] Furthermore, the system is also used to implement the following functions:

[0121] During the continuous casting production process, the control unit of the crystallizer assembly monitors and collects real-time embryo shell uniformity and real-time embryo shell thickness according to the continuous casting adjustment control benchmark; extracts the embryo shell uniformity requirement and embryo shell thickness requirement from the embryo shell quality index; and performs PID closed-loop feedback control of the continuous casting adjustment control benchmark based on a first dynamic deviation between the real-time embryo shell uniformity and the embryo shell uniformity requirement and a second dynamic deviation between the real-time embryo shell thickness and the embryo shell required thickness.

[0122] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0123] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0124] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A continuous casting adjustment method for crystallizer assembly simulation, characterized in that: The method comprises: Conduct network data collection based on the mold assembly's device ID and molten steel composition information to obtain continuous casting history data; Using linear interpolation to expand the working conditions of the continuous casting historical data to obtain continuous casting coverage data; Locally call component structure information based on the device ID to establish an initial finite element model; calibrating the boundary conditions of the initial finite element model using the continuous casting coverage data to obtain a continuous casting finite element model; Performing cluster analysis on the continuous casting history data according to the embryo shell quality index to locate W continuous casting optimization difference starting points; The continuous casting finite element model is copied to W continuous casting thermal simulation containers; After inputting the W continuous casting optimization difference starting points into the W continuous casting thermal simulation containers, starting parallel local iterative optimization to obtain a continuous casting multi-objective equilibrium solution set; The solution with the lowest energy consumption is selected from the continuous casting multi-objective equilibrium solution set as a benchmark solution, and the benchmark solution is input into the continuous casting finite element model for global iterative optimization to obtain a continuous casting adjustment control benchmark.

2. A continuous casting adjustment method for crystallizer assembly simulation according to claim 1, characterized in that: The continuous casting historical data is expanded using linear interpolation to obtain continuous casting coverage data. The method includes: Defining continuous casting adjustment constraints according to the component structure information, wherein the continuous casting adjustment constraints include a water spraying amount interval, a scale scraping frequency interval, and a water refinement component angle interval; Presetting a multidimensional interpolation scale, and performing multidimensional grid interpolation within the continuous casting adjustment constraint range with the multidimensional interpolation scale as a constraint to generate gridded interpolation data; Using the continuous casting historical data to traverse the gridded interpolation data, and locate multiple missing data points; Based on the multiple missing data points, the continuous casting historical data is covered and expanded to obtain continuous casting prediction data, wherein the continuous casting prediction data includes multiple working condition prediction data, and the continuous casting prediction data and the continuous casting historical data constitute the continuous casting coverage data.

3. A continuous casting adjustment method for mold assembly simulation according to claim 2, characterized in that: Based on the multiple missing data points, the continuous casting historical data is covered and expanded to obtain continuous casting prediction data, and the method includes: Using the multiple missing data points as a guide, the continuous casting historical data is expanded with adjacent working conditions to obtain multiple sets of updated working condition data; According to the adjacent working condition relationship between the continuous casting historical data and the multiple sets of updated working condition data, the embryo shell quality is predicted by using the linear calculation intermediate value to obtain multiple sets of updated embryo shell qualities; The plurality of groups of updated working condition data and the plurality of groups of updated embryo shell qualities are associated and stored to obtain the plurality of working condition prediction data as the continuous casting prediction data.

4. A continuous casting adjustment method for mold assembly simulation according to claim 3, characterized in that: The continuous casting historical data includes multiple continuous casting condition records and multiple embryo shell quality records. The continuous casting condition records are composed of water spraying volume parameters, scale scraping frequency parameters, water refinement component angle parameters, and injection speed parameters. The embryo shell quality records are composed of historical embryo shell uniformity and historical embryo shell thickness.

5. A continuous casting adjustment method for mold assembly simulation according to claim 4, characterized in that: Cluster analysis is performed on the continuous casting historical data according to the embryo shell quality index to locate W continuous casting optimization difference starting points. The method includes: Extracting H continuous casting condition records in descending order of distance from the plurality of continuous casting condition records according to the Euclidean distance between the embryo shell quality index and the plurality of embryo shell quality records, wherein H ≥ 20W, H and W are positive integers; In the combined enumeration of H continuous casting condition records, we get After the continuous casting condition records are grouped, the H continuous casting condition records are used as topological nodes. The Euclidean distance of the continuous casting condition records is used as the topological length to construct the continuous casting condition topology; Using a preset working condition topology distance to remove topological connections of the continuous casting working condition topology, and obtaining W isolated nodes; The W continuous casting operating condition records of the W isolated nodes are used as the W continuous casting optimization difference starting points.

6. A continuous casting adjustment method for crystallizer assembly simulation according to claim 5, characterized in that: After inputting the W continuous casting optimization difference starting points into the W continuous casting thermal simulation containers, parallel local iterative optimization is initiated to obtain a continuous casting multi-objective equilibrium solution set. The method includes: Inputting the first continuous casting optimization difference starting point into the first continuous casting thermal simulation container to perform unit continuous casting production simulation, then quantifying the continuous casting performance and outputting the first simulation solution coefficient; updating the first continuous casting optimization difference starting point by using a preset continuous casting condition search scale, outputting a first updated continuous casting condition, quantifying the continuous casting performance of the first continuous casting thermal simulation container, and outputting a second simulation solution coefficient; If the second simulation solution coefficient is better than the first simulation solution coefficient, updating the continuous casting condition search scale according to the deviation between the first simulation solution coefficient and the second simulation solution coefficient, and outputting a first updated search scale; Similarly, the continuous casting condition search scale is updated and the continuous casting performance is quantitatively evaluated based on the simulation solution coefficients obtained by the adjacent optimization search, until the deviation of the obtained adjacent optimization simulation solution coefficients is less than the preset threshold, and the first local iterative optimal solution is output; By analogy, through the parallel local iterative optimization of the W continuous casting thermal simulation containers, W local iterative optimal solutions are obtained to constitute the continuous casting multi-objective equilibrium solution set.

7. A continuous casting adjustment method for mold assembly simulation according to claim 6, characterized in that: Inputting the first continuous casting optimization difference starting point into the first continuous casting thermal simulation container to perform unit continuous casting production simulation, then quantifying the continuous casting performance, and outputting the first simulation solution coefficient. The method includes: Pre-constructing simulation solution evaluation weights, wherein the simulation solution evaluation weights include energy consumption weight, embryo shell uniformity deviation weight, embryo shell thickness deviation weight, scale residue weight, and water consumption weight; Inputting the first continuous casting optimization difference starting point into the first continuous casting thermal simulation container to perform unit continuous casting production simulation, and outputting the first continuous casting energy consumption, the first simulated uniformity, the first simulated thickness, the first scale residue and the first water consumption; performing deviation evaluation on the first simulated uniformity and the first simulated thickness according to the embryo shell quality index, and outputting a first uniformity deviation and a first thickness deviation; The simulation solution evaluation weight is used to fuse the first continuous casting energy consumption, the first uniformity deviation, the first thickness deviation, the first scale residue and the first water consumption, and the first simulation solution coefficient is output.

8. The continuous casting adjustment method for mold assembly simulation according to claim 1, characterized in that: The continuous casting coverage data is used to calibrate the boundary conditions of the initial finite element model to obtain a continuous casting finite element model, and the method includes: Establishing the initial finite element model according to the component structure information, wherein the initial finite element model includes a cooling water injection dynamics module, a self-cleaning mechanics module, a droplet splashing optimization module, and a molten steel solidification thermodynamics module; Decomposing the continuous casting coverage data according to the module structure of the initial finite element model to obtain a cooling water injection data group, a self-cleaning mechanics data group, a droplet splashing data group, and a molten steel solidification data group; After the cooling water injection data group, self-cleaning mechanics data group, droplet splashing data group and molten steel solidification data group are mapped and loaded into the cooling water injection dynamics module, self-cleaning mechanics module, droplet splashing optimization module and molten steel solidification thermodynamics module for single module calibration, the initial finite element model is collaboratively verified by multi-physical field coupling and the continuous casting finite element model is output.

9. The continuous casting adjustment method for mold assembly simulation according to claim 1, characterized in that: The continuous casting historical data is expanded using linear interpolation to obtain continuous casting coverage data. The method further includes: During the continuous casting production process, the control unit of the mold assembly monitors and collects the real-time uniformity and thickness of the embryo shell according to the continuous casting adjustment control benchmark; Extracting the embryo shell uniformity requirement and the embryo shell thickness requirement from the embryo shell quality index; According to a first dynamic deviation between the real-time embryo shell uniformity and the embryo shell uniformity requirement and a second dynamic deviation between the real-time embryo shell thickness and the embryo shell required thickness, PID closed-loop feedback control of the continuous casting adjustment control benchmark is performed.

10. A continuous casting adjustment system for crystallizer assembly simulation, characterized in that: The system is used to implement the continuous casting adjustment method for crystallizer assembly simulation according to any one of claims 1 to 9, and the system comprises: The continuous casting history data acquisition module is used to collect network data based on the device ID of the crystallizer assembly and the steel liquid composition information to obtain the continuous casting history data; A working condition expansion module, configured to expand the working condition of the continuous casting historical data by using linear interpolation to obtain continuous casting coverage data; The model building module is used to locally call the component structure information according to the equipment ID and build the initial finite element model; a continuous casting finite element model acquisition module, configured to calibrate the boundary conditions of the initial finite element model using the continuous casting coverage data to obtain a continuous casting finite element model; A cluster analysis module, configured to perform cluster analysis on the continuous casting history data according to the embryo shell quality index, and locate W continuous casting optimization difference starting points; A model replication module, used for replicating the continuous casting finite element model to W continuous casting thermal simulation containers; A local iterative optimization module is used to input the W continuous casting optimization difference starting points into the W continuous casting thermal simulation containers, start parallel local iterative optimization, and obtain a continuous casting multi-objective equilibrium solution set; A global iterative optimization module is used to select the solution with the lowest energy consumption from the continuous casting multi-objective equilibrium solution set as a benchmark solution, and input the benchmark solution into the continuous casting finite element model for global iterative optimization to obtain a continuous casting adjustment control benchmark.

Images