KR desulfurization stirring control optimization system based on big data
By using a KR desulfurization stirring control optimization system based on big data, stirring parameters are dynamically adjusted, solving the problem of poor adaptability to operating condition fluctuations in traditional control methods, and improving the stability and economy of the KR desulfurization process.
Patent Information
- Application Number
- CN202610364316.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-24
- Publication Date
- 2026-06-09
AI Technical Summary
The existing KR desulfurization stirring control relies on experience and cannot adapt to fluctuations in operating conditions, resulting in unstable desulfurization efficiency, energy waste and equipment wear and tear. It also lacks effective use of historical production data, making it difficult to achieve refined control.
A KR desulfurization stirring control optimization system based on big data is adopted. Through data acquisition, preprocessing, model building and real-time optimization solution, stirring parameters are dynamically adjusted to achieve adaptive control.
It improves the stability and economy of the KR desulfurization process, reduces energy consumption and equipment wear, and enhances the level of intelligence.
Smart Images
Figure CN122168815A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of steel industry technology, and in particular to a KR desulfurization stirring control optimization system based on big data. Background Technology
[0002] KR desulfurization is a commonly used hot metal desulfurization technology in the iron and steel metallurgy industry. Its core lies in stirring the molten iron with a stirring device to ensure that the desulfurizing agent is fully mixed and reacted with the molten iron, thereby achieving the purpose of desulfurization. In the KR desulfurization process, the rationality of the stirring control parameters (such as stirring speed, stirring depth, stirring time, etc.) directly affects the desulfurization efficiency, hot metal temperature loss, and equipment wear and tear.
[0003] Currently, KR desulfurization stirring control mostly adopts traditional experience-based control methods, where operators set fixed stirring control parameters based on past production experience. However, this method has significant drawbacks: Firstly, production conditions such as molten iron composition (e.g., sulfur content, silicon content, carbon content), molten iron temperature, and the type and amount of desulfurizing agent added exhibit significant fluctuations. Fixed stirring control parameters cannot adapt to different production conditions, easily leading to unstable desulfurization efficiency, or even problems such as substandard desulfurization or excessive stirring causing energy waste. Secondly, traditional control methods lack effective utilization of historical production data, failing to extract optimal stirring control patterns from large amounts of production data, making it difficult to achieve refined and intelligent control of the KR desulfurization process. Summary of the Invention
[0004] This application provides a KR desulfurization stirring control optimization system based on big data to overcome the shortcomings of existing KR desulfurization stirring control, such as reliance on experience, inability to adapt to operating condition fluctuations, and low control accuracy.
[0005] The system includes: A data acquisition module, configured to acquire several multi-dimensional data related to KR desulfurization; The data preprocessing module is configured to preprocess all the multi-dimensional data to obtain several multi-dimensional data samples; each multi-dimensional data sample corresponds to a complete KR desulfurization process. The model building module is configured to build a KR desulfurization process prediction model and train the KR desulfurization process prediction model using the multi-dimensional data samples to obtain a target KR desulfurization process prediction model. A real-time optimization solution module is configured to input the parameters to be predicted into the target KR desulfurization process prediction model for prediction and obtain the optimal solution parameters; the parameters to be predicted include known parameters and unknown parameters, and the unknown parameters correspond to the optimal solution parameters; An optimization control execution module is configured to periodically adjust the KR desulfurization parameters based on the optimal solution parameters.
[0006] Preferably, the multi-dimensional data includes on-site equipment data, molten iron sulfur data, and desulfurization process data.
[0007] Preferably, the data acquisition module includes: A field PLC data acquisition unit is configured to acquire data from the field equipment; the field equipment data includes stirring speed, stirring torque, and lifting mechanism position. A laboratory system data acquisition unit, configured to acquire the molten iron sulfur data; the molten iron sulfur data includes the initial sulfur content and the final sulfur content of the molten iron. The MES system data acquisition unit is configured to acquire the desulfurization process data, which includes the ladle number, molten iron weight, molten iron temperature, type and amount of desulfurizing agent, average particle size, average surface temperature of the stirring paddle, and Reynolds number of molten iron flow.
[0008] Preferably, the preprocessing includes: The multi-dimensional data is processed to remove outliers, resulting in the first preprocessed data; The first preprocessed data is padded with missing values to obtain the second preprocessed data; The second preprocessed data is timestamped to obtain the multi-dimensional data sample.
[0009] Preferably, the model building module includes: A feature extraction unit is configured to extract temporal features based on the multi-dimensional data samples to obtain training temporal features and validation temporal features. The model building and training unit is configured to build a KR desulfurization process prediction model with KR desulfurization parameters and the final sulfur content of the molten iron, and to train the KR desulfurization process prediction model using the training time series features to obtain the target KR desulfurization process prediction model. The model verification unit is configured to verify the prediction accuracy of the target KR desulfurization process prediction model based on the verification time series characteristics, and to encapsulate the target KR desulfurization process prediction model when the prediction accuracy meets the preset requirements.
[0010] Preferably, the feature extraction unit is further configured as follows: The multi-dimensional data samples are filtered based on the degree of influence of the data on the final sulfur content of the molten iron to obtain the target multi-dimensional data; The target multidimensional data is divided into a training dataset and a validation dataset, and features of the training dataset and the validation dataset are extracted respectively to obtain training set features and validation set features. Temporal feature extraction is performed on the training set features and the validation set features respectively to obtain the training temporal features and the validation temporal features.
[0011] Preferably, the model building and training unit is further configured as follows: The training time-series features are incorporated into the training set features to construct the KR desulfurization process prediction model; The KR desulfurization process prediction model is trained using the training time series features to obtain dynamic constraint weights, and the dynamic constraint weights are then integrated into the KR desulfurization process prediction model to obtain the target KR desulfurization process prediction model.
[0012] Preferably, the real-time optimization solution module includes: A question-and-answer definition unit, configured to provide a platform for users to input the parameters to be predicted; The parameter prediction unit is configured to input the parameters to be predicted into the target KR desulfurization process prediction model and solve it using a particle swarm optimization algorithm to obtain the optimal solution parameters.
[0013] Preferably, the optimization control execution module is further configured to send the optimal solution parameters to the KR desulfurization main PLC; the optimal solution parameters are used to adjust the speed of the frequency converter regulating motor and the positioning height of the stirring paddle in the lifting mechanism.
[0014] Preferably, the optimal solution parameters are also used to feed back real-time KR desulfurization data to the real-time optimization solution module and the optimization control execution module.
[0015] As described above, this application provides a KR desulfurization stirring control optimization system based on big data. The system includes a data acquisition module configured to acquire several multi-dimensional data related to KR desulfurization; a data preprocessing module configured to preprocess all the multi-dimensional data to obtain several multi-dimensional data samples; each multi-dimensional data sample corresponds to a complete KR desulfurization process; a model building module configured to build a KR desulfurization process prediction model and train the KR desulfurization process prediction model using the multi-dimensional data samples to obtain a target KR desulfurization process prediction model; a real-time optimization solution module configured to input the parameters to be predicted into the target KR desulfurization process prediction model for prediction to obtain optimal solution parameters; the parameters to be predicted include known parameters and unknown parameters, and the unknown parameters correspond to the optimal solution parameters; and an optimization control execution module configured to periodically adjust the KR desulfurization parameters according to the optimal solution parameters. This application overcomes the shortcomings of existing technologies in KR desulfurization stirring control, such as reliance on experience, inability to adapt to operating condition fluctuations, and low control accuracy, through the aforementioned system. Attached Figure Description
[0016] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a schematic diagram of a KR desulfurization stirring control optimization system based on big data, as described in this application. Figure 2 This is a schematic diagram of the data acquisition module in a KR desulfurization stirring control optimization system based on big data, as described in this application. Figure 3 This is a schematic diagram of the model building module in a KR desulfurization stirring control optimization system based on big data, as described in this application. Figure 4 This is a schematic diagram of the real-time optimization solution module in a KR desulfurization stirring control optimization system based on big data, as described in this application. Detailed Implementation
[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0019] It should be noted that the brief descriptions of terms in this application are only for the convenience of understanding the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise stated, these terms should be understood in their ordinary and common meaning.
[0020] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.
[0021] KR desulfurization is a commonly used hot metal pretreatment technology in the iron and steel metallurgy industry. Its core lies in the enhanced mixing of molten iron through mechanical stirring, ensuring thorough contact and reaction between the desulfurizing agent and the molten iron, thereby effectively reducing the sulfur content. Due to its superior kinetic conditions, high desulfurization efficiency, and low desulfurizing agent consumption, this process is widely used in the production of low-sulfur steel grades. In a typical KR desulfurization process, a stirring device (such as a cross-shaped stirring head) is inserted into the ladle and rotates to create a vortex, promoting uniform dispersion of the desulfurizing agent in the molten iron. The agent then undergoes a chemical reaction to generate a sulfide slag phase, and the desulfurization target is finally achieved through slag removal.
[0022] However, existing KR desulfurization processes still have significant limitations in terms of stirring control. Currently, industrial production mostly adopts traditional experience-based control methods, where operators set fixed stirring control parameters (such as stirring speed, stirring depth, and stirring time) based on historical production experience. This control method is highly dependent on human experience and is difficult to adapt to dynamic fluctuations in production conditions such as molten iron composition (such as sulfur, silicon, and carbon content), molten iron temperature, and the type and amount of desulfurizing agent added. For example, when the initial sulfur content of the molten iron is high or the reactivity of the desulfurizing agent changes, fixed parameters may not achieve the optimal mixing effect, leading to insufficient desulfurization reaction, unstable desulfurization efficiency, or even failure to meet desulfurization standards. On the other hand, excessively pursuing desulfurization effects often requires extending the stirring time or increasing the stirring intensity, which not only increases energy consumption but may also exacerbate molten iron temperature loss and equipment wear, affecting production economy and equipment lifespan.
[0023] Furthermore, traditional experience-based control methods lack systematic mining and utilization of historical production data, failing to identify optimal control patterns under different operating conditions from accumulated process data. As steel production moves towards intelligence and precision, this extensive control model is no longer sufficient to meet the demands of high-quality steel production and energy conservation and emission reduction. Although some improved processes have attempted to enhance desulfurization efficiency by optimizing desulfurizing agent composition or agitator structure in recent years, their control strategies have not fundamentally solved the problem of poor adaptability to operating conditions. Therefore, developing a data-driven intelligent agitation control method capable of adapting to changes in molten iron conditions is of great significance for improving the stability, economy, and intelligence level of the KR desulfurization process.
[0024] Based on the above problems, this application provides the following implementation method.
[0025] Figure 1 This is a schematic diagram of a KR desulfurization stirring control optimization system based on big data, as described in this application.
[0026] See Figure 1 As can be seen, this embodiment provides a KR desulfurization stirring control optimization system based on big data, the system comprising: The data acquisition module is configured to acquire several multi-dimensional data related to KR desulfurization.
[0027] Specifically, in this embodiment, the algorithm device and data acquisition device are deployed through the data acquisition module to establish communication, thereby obtaining several multi-dimensional data related to KR desulfurization, wherein the multi-dimensional data includes field equipment data, molten iron sulfur data and desulfurization process data.
[0028] Figure 2 This is a schematic diagram of the data acquisition module in a KR desulfurization stirring control optimization system based on big data, as described in this application.
[0029] See Figure 2 Furthermore, in some embodiments, the data acquisition module includes: A field PLC data acquisition unit is configured to acquire data from the field equipment; the field equipment data includes stirring speed, stirring torque, and lifting mechanism position. A laboratory system data acquisition unit, configured to acquire the molten iron sulfur data; the molten iron sulfur data includes the initial sulfur content and the final sulfur content of the molten iron. The MES system data acquisition unit is configured to acquire the desulfurization process data, which includes the ladle number, molten iron weight, molten iron temperature, type and amount of desulfurizing agent, average particle size, average surface temperature of the stirring paddle, and Reynolds number of molten iron flow.
[0030] Specifically, in this embodiment, the stirring speed, stirring torque, and lifting mechanism position (used to calculate the height from the bottom and immersion depth) are collected from the on-site PLC; the initial and final sulfur content of the molten iron are obtained from the laboratory system; and the ladle number, weight, temperature, type and amount of desulfurizing agent, average particle size (dp), average surface temperature (Tp) of the stirring paddle, and Reynolds number (Re) of the molten iron flow are obtained from the MES system. All of this data is then used to form a structured vector database. Each data record contains 18 dimensions, forming a standardized data vector.
[0031] The system also includes: The data preprocessing module is configured to preprocess all the multi-dimensional data to obtain several multi-dimensional data samples; each multi-dimensional data sample corresponds to a complete KR desulfurization process.
[0032] Specifically, in this embodiment, the data preprocessing module performs operations such as removing outliers, processing missing values, and aligning timestamps on the data to form a structured, high-quality "data sample," with each sample corresponding to a complete KR desulfurization process.
[0033] The system also includes: The model building module is configured to build a KR desulfurization process prediction model and train the KR desulfurization process prediction model using the multi-dimensional data samples to obtain the target KR desulfurization process prediction model.
[0034] Specifically, in this embodiment, the KR desulfurization process prediction model is constructed and trained through the model building module to obtain the target KR desulfurization process prediction model for subsequent parameter prediction.
[0035] Figure 3 This is a schematic diagram of the model building module in a KR desulfurization stirring control optimization system based on big data, as described in this application.
[0036] See Figure 3 Furthermore, in some embodiments, the model building module includes: The feature extraction unit is configured to extract time-series features based on the multi-dimensional data samples to obtain training time-series features and validation time-series features.
[0037] Specifically, in this embodiment, the feature extraction unit filters the multi-dimensional data samples based on the degree of influence of the data on the final sulfur content of the molten iron to obtain target multi-dimensional data; the target multi-dimensional data is divided into a training dataset and a validation dataset, and features of the training dataset and the validation dataset are extracted respectively to obtain training set features and validation set features; time-series feature extraction is performed on the training set features and the validation set features respectively to obtain training time-series features and validation time-series features.
[0038] The model building module also includes: The model building and training unit is configured to build a KR desulfurization process prediction model with KR desulfurization parameters and the final sulfur content of the molten iron, and to train the KR desulfurization process prediction model using the training time series features to obtain the target KR desulfurization process prediction model.
[0039] Specifically, in this embodiment, the training time series features are integrated into the training set features through the model construction and training unit to construct the KR desulfurization process prediction model; the KR desulfurization process prediction model is trained using the training time series features to obtain dynamic constraint weights, and the dynamic constraint weights are integrated into the KR desulfurization process prediction model to obtain the target KR desulfurization process prediction model.
[0040] The model building module also includes: The model verification unit is configured to verify the prediction accuracy of the target KR desulfurization process prediction model based on the verification time series characteristics, and to encapsulate the target KR desulfurization process prediction model when the prediction accuracy meets the preset requirements.
[0041] Specifically, in this embodiment, the model validation unit uses a test set to validate the model's prediction accuracy (e.g., R0). 2 (RMSE). Once the accuracy meets the requirements, the model is packaged and deployed in the model prediction module of the industrial server for online use.
[0042] For example: Feature engineering: Select features that significantly affect the final sulfur content from the preprocessed data, such as initial sulfur content, molten iron temperature, molten iron weight, desulfurizing agent dosage, average stirring speed, and average height from the bottom. Divide the historical data into training and testing sets proportionally. Extract time-series features of process parameters: Let the time-series sequence of a certain process parameter (such as stirring speed) in the desulfurization process be N(t) = [N_1, N_2, ..., N_k], where t is the time step and k is the number of time-series data points. Define its "time-series fluctuation feature" F. Nand "Trend Change Characteristics" T N :
[0043] ; In the formula, F N T reflects the degree of fluctuation of parameters during the desulfurization process. N Reflecting the overall trend of parameter change, the two together constitute the time-series feature vector of the parameter.
[0044] Model building and training: Using the training set data, a Deep Neural Network (DNN) algorithm was employed for training. DNNs effectively handle nonlinear relationships and provide excellent evaluation of feature importance. The time-series feature vectors were integrated into the original feature set to establish a dynamic correlation model describing the relationship between dynamic changes in process parameters and the final sulfur content. ; In the formula, To predict the final sulfur content, , These represent the temporal fluctuation characteristics and trend change characteristics of the stirring torque, respectively. This is a DNN model that integrates time-series features. By capturing the dynamic changes in process parameters, this model improves the accuracy of predicting the final sulfur content, and is particularly suitable for production scenarios with large parameter fluctuations.
[0045] Dynamic calculation of constraint weights: defining feature vectors for production scenarios G1 represents the required sulfur content level for the steel grade, G2 represents the equipment health status, and G3 represents the energy consumption control requirements. The dynamic weights of each constraint are as follows: 'm' represents the constraint type. For example, m=1 corresponds to the endpoint sulfur content constraint, and m=2 corresponds to the torque constraint. The calculation is as follows: ; By incorporating dynamic weights into the original objective function, a multi-constraint dynamic optimization model is formed: ; In the formula, P m (N, H, D) is the penalty term for the m-th type of constraint. When the constraint is violated, P... m As the value increases, it tends towards 0; conversely, as it decreases, it approaches 0. This function dynamically adjusts the constraint weights to prioritize meeting key requirements in different scenarios, thereby improving the practicality of the optimization solution.
[0046] The system also includes: The real-time optimization solution module is configured to input the parameters to be predicted into the target KR desulfurization process prediction model for prediction and obtain the optimal solution parameters; the parameters to be predicted include known parameters and unknown parameters, and the unknown parameters correspond to the optimal solution parameters.
[0047] Specifically, in this embodiment, the problem to be solved, i.e. the parameters to be predicted, is input into the target KR desulfurization process prediction model through the real-time optimization solution module for prediction, thereby obtaining the parameters required for the problem to be solved.
[0048] Figure 4 This is a schematic diagram of the real-time optimization solution module in a KR desulfurization stirring control optimization system based on big data, as described in this application.
[0049] See Figure 4 Furthermore, in some embodiments, the real-time optimization solution module includes: A question-and-answer definition unit, configured to provide a platform for users to input the parameters to be predicted; The parameter prediction unit is configured to input the parameters to be predicted into the target KR desulfurization process prediction model and solve it using a particle swarm optimization algorithm to obtain the optimal solution parameters.
[0050] Specifically, in this embodiment, the question-and-answer definition unit provides a platform for users to input the parameters to be predicted, and the parameter prediction unit solves the optimal solution parameters based on the target KR desulfurization process prediction model using the particle swarm optimization algorithm.
[0051] For example: Problem Definition: For a new batch of molten iron, its initial sulfur content S0, temperature T, and weight W are known. The optimization objective is to find a set of control parameters (rotation speed N, height from the bottom H, immersion depth D) that minimizes the desulfurizing agent consumption J.
[0052] Build an optimization model: Objective function: ; Constraints: Predict the endpoint sulfur content: Send = Model(S0, T, W, N, H, D, ...) ≤ S_target(target sulfur content).
[0053] Restrictions: N_min≤N≤N_max (Safe range of rotational speed); Torque ≤ Torque_max (torque safety limit); ... (Other process constraints).
[0054] Solution: The optimization decision module calls the deployed prediction model and uses the particle swarm optimization (PSO) algorithm to solve the above optimization problem. The PSO algorithm is highly efficient in continuous space optimization problems and can quickly find an approximate optimal solution (N, H, D).
[0055] The system also includes: An optimization control execution module is configured to periodically adjust the KR desulfurization parameters based on the optimal solution parameters.
[0056] Specifically, in this embodiment, the optimal setpoints (N, H, D) obtained by the optimization control execution module are sent to the KR desulfurization main PLC via the industrial communication protocol. According to the received setpoints, the PLC controls the frequency converter to adjust the motor speed to N and controls the lifting mechanism to position the stirring paddle to height H, thereby ensuring that the immersion depth is D. The system can periodically (e.g., every minute) repeat the execution functions of the real-time optimization solution module and the optimization control execution module throughout the entire desulfurization process, realizing adaptive fine-tuning based on real-time torque, temperature and other data, forming a closed-loop control.
[0057] This embodiment has the following advantages: Compared with traditional experience-based operations: This embodiment, through optimization, uses a lower rotation speed and less desulfurizer, saving desulfurizer while ensuring compliance with standards. At the same time, the reduced rotation speed saves electricity and reduces equipment wear.
[0058] Safety: Through dynamic adjustment, equipment safety accidents caused by overloading of stirring torque are effectively avoided.
[0059] Adaptability: The system can adjust itself based on real-time operating conditions and continuously evolve through continuous learning, exhibiting a high level of intelligence.
[0060] For ease of explanation, the above description has been provided in conjunction with specific embodiments. However, the discussion in some embodiments is not intended to be exhaustive or to limit the embodiments to the specific forms disclosed above. Various modifications and variations can be obtained based on the above teachings. The selection and description of the above embodiments are for the purpose of better explaining the contents of this disclosure, thereby enabling those skilled in the art to better utilize the embodiments.
Claims
1. A KR desulfurization stirring control optimization system based on big data, characterized in that, The system includes: A data acquisition module, configured to acquire several multi-dimensional data related to KR desulfurization; The data preprocessing module is configured to preprocess all the multi-dimensional data to obtain several multi-dimensional data samples; each multi-dimensional data sample corresponds to a complete KR desulfurization process. The model building module is configured to build a KR desulfurization process prediction model and train the KR desulfurization process prediction model using the multi-dimensional data samples to obtain a target KR desulfurization process prediction model. A real-time optimization solution module is configured to input the parameters to be predicted into the target KR desulfurization process prediction model for prediction and obtain the optimal solution parameters; the parameters to be predicted include known parameters and unknown parameters, and the unknown parameters correspond to the optimal solution parameters; An optimization control execution module is configured to periodically adjust the KR desulfurization parameters based on the optimal solution parameters.
2. The KR desulfurization stirring control optimization system based on big data according to claim 1, characterized in that, The multi-dimensional data includes on-site equipment data, molten iron sulfur data, and desulfurization process data.
3. The KR desulfurization stirring control optimization system based on big data according to claim 2, characterized in that, The data acquisition module includes: A field PLC data acquisition unit is configured to acquire data from the field equipment; the field equipment data includes stirring speed, stirring torque, and lifting mechanism position. A laboratory system data acquisition unit, configured to acquire the molten iron sulfur data; the molten iron sulfur data includes the initial sulfur content and the final sulfur content of the molten iron. The MES system data acquisition unit is configured to acquire the desulfurization process data, which includes the ladle number, molten iron weight, molten iron temperature, type and amount of desulfurizing agent, average particle size, average surface temperature of the stirring paddle, and Reynolds number of molten iron flow.
4. The KR desulfurization stirring control optimization system based on big data according to claim 1, characterized in that, The preprocessing includes: The multi-dimensional data is processed to remove outliers, resulting in the first preprocessed data; The first preprocessed data is padded with missing values to obtain the second preprocessed data; The second preprocessed data is timestamped to obtain the multi-dimensional data sample.
5. The KR desulfurization stirring control optimization system based on big data according to claim 3, characterized in that, The model building module includes: A feature extraction unit is configured to extract temporal features based on the multi-dimensional data samples to obtain training temporal features and validation temporal features. The model building and training unit is configured to build a KR desulfurization process prediction model with KR desulfurization parameters and the final sulfur content of the molten iron, and to train the KR desulfurization process prediction model using the training time series features to obtain the target KR desulfurization process prediction model. The model verification unit is configured to verify the prediction accuracy of the target KR desulfurization process prediction model based on the verification time series characteristics, and to encapsulate the target KR desulfurization process prediction model when the prediction accuracy meets the preset requirements.
6. The KR desulfurization stirring control optimization system based on big data according to claim 5, characterized in that, The feature extraction unit is further configured to: The multi-dimensional data samples are filtered based on the degree of influence of the data on the final sulfur content of the molten iron to obtain the target multi-dimensional data; The target multidimensional data is divided into a training dataset and a validation dataset, and features of the training dataset and the validation dataset are extracted respectively to obtain training set features and validation set features. Temporal feature extraction is performed on the training set features and the validation set features respectively to obtain the training temporal features and the validation temporal features.
7. The KR desulfurization stirring control optimization system based on big data according to claim 6, characterized in that, The model building and training units are also configured as follows: The training time-series features are incorporated into the training set features to construct the KR desulfurization process prediction model; The KR desulfurization process prediction model is trained using the training time series features to obtain dynamic constraint weights, and the dynamic constraint weights are then integrated into the KR desulfurization process prediction model to obtain the target KR desulfurization process prediction model.
8. The KR desulfurization stirring control optimization system based on big data according to claim 1, characterized in that, The real-time optimization solution module includes: A question-and-answer definition unit, configured to provide a platform for users to input the parameters to be predicted; The parameter prediction unit is configured to input the parameters to be predicted into the target KR desulfurization process prediction model and solve it using a particle swarm optimization algorithm to obtain the optimal solution parameters.
9. The KR desulfurization stirring control optimization system based on big data according to claim 1, characterized in that, The optimized control execution module is also configured to send the optimal solution parameters to the KR desulfurization main PLC; the optimal solution parameters are used to adjust the speed of the frequency converter regulating motor and the positioning height of the stirring paddle in the lifting mechanism.
10. The KR desulfurization stirring control optimization system based on big data according to claim 9, characterized in that, The optimal solution parameters are also used to feed back real-time KR desulfurization data to the real-time optimization solution module and the optimization control execution module.