Production method for preparing steel based on cyclic utilization of hot-state refining slag in converter

Abstract

The invention discloses a production method for preparing steel based on converter cyclic utilization of hot-state refining slag, and relates to the technical field of network security. An LF slag database is determined, an invisible box is deployed for the LF slag database, and the invisible box comprises an invisible channel, a transfer surface and a plurality of induction points; basic data of steel preparation, real-time data of the smelting process and process constraint data are collected in real time to serve as analysis data. According to the method, the LF slag database can be dynamically transferred and hidden when the unauthorized network is accessed, so that the LF slag database cannot be accessed by the unauthorized network, meanwhile, normal access of the authorized network cannot be influenced, a good data protection effect is achieved, the safe LF slag database obtains an accurate target model and an expert system, and the security of the LF slag database is improved. And an accurate decision can be given in the production process of preparing steel by cyclically utilizing the hot-state refining slag in the converter.

Description

Technical Field

[0001] This invention relates to the field of network security technology, and specifically to a production method for preparing steel based on the recycling of hot refining slag in a converter. Background Technology

[0002] The core objective of steelmaking using hot refining slag in converters is to directly return the hot slag generated in the LF refining process to the converter, replacing some of the slag-forming agents such as lime, thereby achieving energy conservation, reduced consumption, shortened smelting cycles, and improved steel quality. The key to this technology lies in the precise control of core parameters such as the composition, temperature, basicity, and inclusion content of the hot refining slag. The source and optimization basis of these parameters is the LF slag database; therefore, the LF slag database is a crucial link in the process of steelmaking using hot refining slag. In recent years, cases of cyberattacks on industrial control systems have exploded globally, with targets extending from critical infrastructure such as energy and electricity to manufacturing industries such as steel and chemicals. As a pillar industry of the national economy, steel companies' production data and process information have become key targets of cyberattacks. In particular, if the LF slag database lacks proactive protection capabilities such as dynamic transfer and stealth, it is highly vulnerable to attack. Once data is leaked or maliciously tampered with, it can not only lead to converter production process disruptions and product quality decline, but also potentially cause serious consequences such as production interruptions and equipment damage, threatening the normal operation of enterprises and the stable development of the industry. Summary of the Invention

[0003] The purpose of this invention is to provide a production method for preparing steel based on the recycling of hot refining slag in a converter, so as to solve the problems in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for producing steel based on the recycling of hot refining slag in a converter, comprising the following steps: The LF slag database is identified, and a stealth box is deployed on the LF slag database. The stealth box includes a stealth channel, a transfer surface, and multiple sensing points. The LF slag database is protected based on the stealth box. An expert system is built based on the LF slag database, and a learning model is trained to obtain a target model. The target model is integrated with the expert system to obtain a control model. Real-time data collection of basic steelmaking data, real-time smelting process data, and process constraint data is used as analysis data. The analysis data is input into the control model, and the output parameters include hot slag core usage parameters, fresh slag replenishment decision parameters, and process synergistic optimization suggestion parameters, thus obtaining the analysis results. Based on the analysis results, the on-site operation for steel preparation is carried out until the production of steel is completed by recycling hot refining slag in a converter.

[0005] In a preferred embodiment, the step of determining the LF scum database and deploying a stealth box on the LF scum database includes: Collect and store basic slag composition data, historical corresponding smelting process data, and quality and effect correlation data to obtain the LF slag database; A stealth channel is configured for the corresponding LF slag database. The stealth channel consists of multiple storage spaces, which are connected by nodes. Multiple transfer surfaces are configured for the corresponding stealth passage, and these transfer surfaces are interconnected. Multiple sensor points are set up in the stealth channel, and the multiple sensor points are interconnected. In addition, the multiple sensor points are interconnected with multiple transfer surfaces to obtain a stealth box. Protect the LF slag database using stealth boxes.

[0006] In a preferred embodiment, the step of configuring a stealth channel for the corresponding LF slag database includes: Set up multiple storage spaces, define storage ranges in each storage space, configure at least two nodes for each storage range, connect each node to the storage range in the corresponding storage space, and connect the storage ranges to each other through the nodes. Configure connection rules for multiple storage spaces, where the connection rules define the connection relationships between multiple storage ranges; Multiple storage ranges are connected through nodes according to the connection rules to obtain a stealth channel.

[0007] In a preferred embodiment, the step of configuring multiple transfer surfaces in the corresponding stealth channel and interconnecting the multiple transfer surfaces includes: Multiple transfer surfaces are configured for each storage range. A first connection port is set for each transfer surface, and multiple second connection ports are configured for each storage range. The transfer surfaces are matched one-to-one with the second connection ports of the storage ranges. The transfer surfaces are connected to the corresponding second connection ports through the first connection ports. Transfer surfaces are stored in the storage space, and multiple transfer surfaces are interconnected.

[0008] In a preferred embodiment, the step of setting multiple sensing points within the stealth channel, interconnecting the multiple sensing points, and interconnecting the multiple sensing points with multiple transfer surfaces to obtain the stealth box includes: Multiple sensing points are set up within each storage area, and the storage locations of the multiple sensing points within the storage area are determined. Sensors in adjacent storage locations are interconnected. Within different storage ranges, sensors closer to a node are selected as cross-region sensors. Cross-region sensors are connected to the nearest cross-region sensors in other storage ranges through the nearest node. Establish a management range between each sensing point and the transfer surface within its storage range. The management range is the spatial range of at least one transfer surface that the sensing point is pre-managed within the storage range. Establish a transfer mechanism between the sensing point and the transfer surface within the management range.

[0009] In a preferred embodiment, the step of protecting the LF slag database based on the stealth box includes: Store the LF slag database in any one of the transfer surfaces, which will serve as the target transfer surface. Set an open network port for each of the multiple storage ranges. When one or more unauthorized networks access one or more storage ranges through storage space, the location of the unauthorized network access is determined by the sensing point; Disconnect the storage range currently without authorized network access from the transfer plane of the connection; When the management scope of an unauthorized network access sensing point includes a target transfer surface, the transfer mechanism enables the connection between the transfer surface and other transfer surfaces in storage ranges that are not accessed by unauthorized networks, and confirms the transfer surface in other storage ranges that are not accessed by unauthorized networks closest to the communication management as a standby transfer surface; Based on the connection channel between the target transfer surface and the standby transfer surface, the LF slag database in the target transfer surface is transferred to the standby transfer surface. Then, the connection between the target transfer surface and the standby transfer surface is disconnected, thus completing the transfer of the LF slag database and protecting the LF slag database. When an authorized network accesses the storage space, the transfer surface containing the LF Slag Database is randomly connected to a secure port. The LF Slag Database is accessed through the secure port by connecting to the authorized network. Multiple secure ports are configured for the storage space. When no authorized network accesses the storage space, the secure ports are closed. After the authorized network exits the access, the enabled secure ports are turned off again.

[0010] In a preferred embodiment, the step of inputting the analytical data into the control model and outputting the hot slag core usage parameters, fresh slag replenishment decision parameters, and process synergistic optimization suggestion parameters to obtain the analytical results includes: The analysis data is based on the real-time collection of basic data, smelting process data and process constraint data during the steel preparation process using the Internet of Things. The analysis data is input into the target model and constrained by an expert system, and the analysis results are output.

[0011] In a preferred embodiment, the step of performing on-site steel preparation based on the analysis results includes: Based on the core usage parameters of hot slag, the decision parameters for adding fresh slag, and the suggested parameters for process synergy optimization, the analysis results are broken down into multiple instructions. The instructions are sent to the corresponding steel preparation equipment for control and on-site operation.

[0012] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention can dynamically transfer and hide the LF slag database when faced with unauthorized network access, ensuring that the LF slag database cannot be accessed by unauthorized networks, while not affecting the normal access of authorized networks. It has a good data protection effect, enabling the secure LF slag database to obtain accurate target models and expert systems, and to make accurate decisions in the production process of steelmaking by recycling hot refining slag in converters. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0014] Figure 1 This is a flowchart of the method of the present invention.

[0015] Figure 2 This is a system block diagram of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0017] Example 1, please refer to Figure 1 and Figure 2 As shown in the figure, the steel production method based on the recycling of hot refining slag in a converter, as described in this embodiment, includes the following steps: S1. Determine the LF slag database, deploy a stealth box on the LF slag database, wherein the stealth box includes a stealth channel, a transfer surface and multiple sensing points, protect the LF slag database based on the stealth box, build an expert system based on the LF slag database and train the learning model to obtain the target model, and integrate the target model with the expert system to obtain the control model 5. S2. Real-time acquisition of basic data for steel preparation, real-time data of smelting process, and process constraint data as analysis data. Input the analysis data into the control model and output the core usage parameters of hot slag, decision parameters for fresh slag material replenishment, and process synergistic optimization suggestion parameters to obtain the analysis results. S3. Based on the analysis results, carry out on-site operations to prepare steel until the production of steel is completed by recycling hot refining slag in a converter.

[0018] As described in steps S1-S3 above, for high-performance machine tool steel and tunneling machine tool steel: by precisely controlling the slag-steel reaction through the LF slag database, the hardness, wear resistance, and toughness of the steel can be optimized to meet the demanding working conditions of machine tool cutting tools, tunneling machine cutting teeth, and other components. Then, based on the initial decision-making framework provided by the expert system, the machine learning model accurately predicts the metallurgical effects (such as desulfurization efficiency and steel inclusion content) under different slag blending schemes by inputting real-time parameters (hot slag composition and weight, tapping temperature, etc.), and finds the optimal solution through algorithm iteration. In the LF furnace hot slag addition control model integrating the expert system and machine learning, the expert system constructs constraint boundaries in the form of rule bases and metallurgical mechanisms. The output results of the machine learning module must conform to these constraints to be used as the basis for on-site operation. The specific constraint logic is reflected in the following aspects: process rule constraints, defining the compliance range of the output. The expert system transforms mature steel smelting process rules into explicit constraint conditions, limiting the output boundary of the machine learning. For example, GCr15, as a high-purity bearing steel, has stringent requirements for slag basicity, typically needing to be maintained within the range of 2.5-3.5 to ensure desulfurization and reduce inclusions. The input hot slag has a CaO content of only 38%. If relying solely on machine learning for calculation, problems such as excessive hot slag addition leading to excessive basicity or insufficient slag replenishment resulting in substandard desulfurization might occur. However, the basicity rules preset by the expert system constrain the model, ensuring that while outputting 3.5 tons of hot slag addition, it also includes 250 kg of lime (to increase CaO content) and 100 kg of dolomite (to adjust slag fluidity), ensuring the slag system meets the refining requirements of GCr15. Furthermore, the input parameter of 1.2 tons of converter slag will trigger the expert system's rule of "excessive slag addition requires controlling the amount of hot slag added to prevent excessive oxidation," preventing the model from outputting excessively high hot slag usage. Metallurgical mechanism constraints ensure the physicochemical rationality of the output. The metallurgical thermodynamics and kinetic mechanisms integrated in the expert system constrain the output data to conform to the basic laws of smelting reactions. For example, in LF refining, the slag-steel reaction efficiency is directly related to the bottom-blown argon flow rate. Too low a flow rate leads to insufficient slag-steel mixing, while too high a flow rate may cause secondary oxidation or splashing of the molten steel. This is a core mechanism in metallurgy. The expert system sets a reasonable range for the bottom-blown argon flow rate based on this mechanism (e.g., in the initial stage of GCr15 refining, the gas volume introduced per minute per ton of molten steel is typically 0.8-1.2 m³). The learning model learns the flow rate data under different operating conditions, and the output must also fall within this range. Safety and quality baseline constraints prevent abnormal outputs: The expert system presets safety and quality baseline rules to prevent unreasonable outputs from the learning model due to extreme data or noise.On one hand, there are temperature-related constraints. The expert system sets matching rules between the amount of hot slag added and the tapping temperature. For example, when the temperature is too low, the amount of hot slag added needs to be reduced to avoid excessive cooling, while when the temperature is too high, the amount can be appropriately increased to adjust the temperature. The hot slag amount of 3.5 tons output by the constraint learning model is compatible with a tapping temperature of 1580℃. On the other hand, there are quality tolerance constraints. In response to the strict requirements of bearing steel for inclusions and sulfur content, the expert system restricts the type and amount of slag added. For example, excessive addition of high-silicon slag is prohibited. The constraint model prioritizes slag materials such as lime, which can increase basicity and have low impurity content. This is an important constraint basis for selecting lime and dolomite for addition in the output. In terms of production practice constraints, the expert system also incorporates the practical experience rules of the steel plant to ensure the executable nature of the output data. For example, the amount of different slag materials added needs to be matched with the rated capacity of the feeding equipment and the bearing capacity of the ladle. The expert system will preset practical rules such as the single addition of lime not exceeding the specified weight and the dolomite not exceeding the specified weight. At the same time, considering the requirements of the steel refining cycle, the addition of hot slag is constrained to prevent excessive addition that would prolong the refining time. The lime, dolomite weight, and other data output by the model all conform to these practical constraints, ensuring that the on-site equipment can execute smoothly, rather than just remaining at the theoretical optimal solution.

[0019] In one embodiment, step S1 of determining the LF scum database and deploying a stealth box on the LF scum database includes: S11. Collect and store basic slag composition data, historical corresponding smelting process data, and quality and effect correlation data to obtain the LF slag database. S12. Configure a stealth channel for the corresponding LF slag database. The stealth channel consists of multiple storage spaces, which are connected by nodes. S13. Multiple transfer surfaces are configured for the corresponding stealth channel, and the multiple transfer surfaces are interconnected; S14. Set up multiple sensing points in the stealth channel, connect the multiple sensing points to each other, and connect the multiple sensing points to multiple transfer surfaces to obtain a stealth box; S15. Protect the LF slag database based on the stealth box.

[0020] In one embodiment, step S12 of configuring a stealth channel for the corresponding LF slag database includes: S121. Set up multiple storage spaces, define storage ranges in each of the multiple storage spaces, configure at least two nodes for each of the multiple storage ranges, connect each node to the storage range in the corresponding storage space, and connect the storage ranges to each other through the nodes. S122. Set connection rules for multiple storage spaces, where the connection rules are the connection relationships between multiple storage ranges; S123. Connect multiple storage ranges through nodes according to the connection rules to obtain a stealth channel.

[0021] In one embodiment, the step S13, in which the corresponding stealth channel is configured with multiple transfer surfaces and the multiple transfer surfaces are interconnected, includes: S131. Multiple transfer surfaces are configured for each of the multiple storage ranges. A first connection port is set for each transfer surface, and multiple second connection ports are configured for each storage range. The transfer surfaces are matched one-to-one with the second connection ports of the storage ranges. The transfer surfaces are connected to the corresponding second connection ports through the first connection ports. S132. Transfer surfaces are stored in the storage space, and multiple transfer surfaces are interconnected.

[0022] In one embodiment, step S14, which involves setting multiple sensing points within the stealth channel, interconnecting these sensing points, and connecting these sensing points to multiple transfer surfaces to obtain the stealth box, includes: S141. Set up multiple sensing points within each storage area and determine the storage location of the multiple sensing points within the storage area; S142. Sensors in adjacent storage locations are interconnected. Within different storage ranges, sensors closer to the node are selected as cross-region sensors. Cross-region sensors are connected to the nearest cross-region sensors in other storage ranges through the nearest node (sensors and nodes are connected via the SNMPv3 application layer protocol, and the sensors and nodes are access monitoring and control interfaces, respectively). S143. Establish a management range between each sensing point and the transfer surface within its storage range, wherein the management range is the spatial range of at least one transfer surface pre-managed by the sensing point within the storage range, and establish a transfer mechanism between the sensing point and the transfer surface within the management range.

[0023] As described in steps S11-S15 above, the LF slag database is a database that stores and manages various types of slag and corresponding smelting-related data during the LF refining process. Its core data includes: Basic slag composition data: This covers the key chemical components of LF slag used in smelting different steel grades, such as the content of CaO, SiO2, Al2O3, FeO, and MnO. This data directly determines the slag's basicity, oxidizing properties, and other key metallurgical characteristics. It also records physical parameters such as slag weight and hot-state temperature, providing a basis for slag recycling and process adaptation. Corresponding smelting process data: This data is linked to the complete process information for each heat of LF slag, including the target steel grade, heat number, dissolved oxygen content of the molten steel at the converter smelting endpoint, tapping temperature, molten steel weight, and operational parameters such as the amount of slag-forming materials (e.g., lime, fluorite) and deoxidizing alloys (e.g., aluminum-manganese-iron, calcium carbide) added during the LF furnace refining process. Quality and Efficacy Correlation Data: The database also incorporates data related to slag-forming effects, such as slag grayscale values ​​obtained through image processing and their corresponding basicity and oxidizing properties evaluation results, as well as quality indicators such as the acid-soluble aluminum content and desulfurization and deoxidation effects of the refined steel, establishing a correlation between slag condition and steelmaking quality. The LF slag database supports precise batching and process decisions: After accumulating sufficient furnace data, the LF slag database can serve as a reference for determining process parameters for new furnace runs. For example, when smelting a new furnace, historical furnaces with similar steel grades and steel compositions can be searched in the LF slag database. By calculating the average values ​​of slag-forming materials and alloy additions from historical furnaces, the batching scheme for the current furnace can be determined, replacing manual experience-based judgment and reducing errors. It can also facilitate slag recycling; combined with slag classification and traceability requirements, the database can classify and label LF slag generated from different steel grades. For example, high-quality special steel LF slag can be separately categorized, its compositional stability advantages recorded, and its targeted recommendations for smelting similar or lower-grade steels can be made. This avoids compositional disorder caused by mixing slags of different qualities and ensures the compatibility of recycled slag. This can assist in process optimization and system upgrades. The large amount of historical data continuously accumulated in the LF slag database can provide training samples for intelligent decision-making systems. Through regression analysis and algorithm modeling of the data, model parameters such as slag addition and feed rate can be optimized. Simultaneously, by analyzing the correlation between slag composition and smelting effect, data support can be provided for process optimization such as addition timing and stirring mode, further improving refining efficiency and steel quality.

[0024] Therefore, the data in the LF slag database is crucial. Subsequent expert systems and training models rely on the data support of the LF slag database. Specifically, the LF slag database provides decision-making support for the expert system. It stores massive amounts of data, including the composition of refining slag for different steel grades, the weight of hot slag in each heat, the amount of slag added to the converter, the tapping temperature, slag blending schemes, and the final molten steel quality. This forms the core foundation for the operation of the expert system. The slag blending rule base in the expert system is built and improved based on historical data from the database. For example, the database contains a large amount of slag system data from high-quality steel smelting, which helps the expert system clarify core rules such as the basicity range and FeO content threshold for slag blending of this type of steel. When the expert system handles real-time slag blending needs, it retrieves matching data from the LF slag database to assist in decision-making. For example, in a heat of bearing steel smelting, the expert system can quickly retrieve historical data from the database for similar operating conditions of the same steel grade, providing a reference for the initial slag blending scheme. The expert system also uses database data to optimize decision-making accuracy. Various functional models of the expert system rely on data from the LF slag database to complete accurate calculations. The expert system feeds back data to the LF slag database to complete a closed-loop update. The decision-making and execution results of the expert system flow back to the LF slag database, enabling dynamic updates. During a heat's slag blending process, after the expert system outputs the hot slag addition amount and fresh slag replenishment plan, the on-site equipment executes the plan. The final actual data, such as the steel composition, slag composition, and smelting energy consumption, are collected by the system and fed back to the LF slag database. If the steel quality of that heat meets the standards, these data become high-quality cases and are added to the LF slag database. Through an authorized port, it can directly connect to the LF slag database, and newly added data can supplement the high-quality case data. If slag blending deviations occur, the relevant abnormal data will also be marked and stored, providing a basis for subsequent optimization. For example, if an excessive amount of hot slag is added in a heat, resulting in a lower steel temperature, this set of abnormal parameters will be entered into the database for subsequent correction of the slag blending rules.

[0025] The LF slag database can be used to train the learning model and obtain the target model. Specific training data includes: Basic data on raw materials and slag materials. This type of data is core to the model's judgment of the basic conditions for slag blending, determining the initial direction of slag blending. This includes data on steel-related raw materials, such as the weight and composition of molten iron (e.g., the content of elements like Si, P, and S), the proportion and weight of scrap steel, and the amount of various alloys added (e.g., Al-containing alloys, Si-containing alloys). It also includes basic data on slag materials, such as the composition (CaO, Al2O3, SiO2 content) and weight of hot slag, the amount and composition of slag added from the converter, the amount of slag turned over, the type and amount of initial slag added during tapping, and the composition data of fresh slag-forming materials such as lime, high-Al slag, and pre-melted slag. For example, data such as the CaO content in the converter slag of a certain heat and the Al2O3 percentage in the hot circulating slag will be included in the learning scope. During the smelting process, this type of data records dynamic operating parameters, helping the learning model to correlate operational behavior with the slag blending effect. This includes key process parameters such as tapping temperature, molten steel temperature entering the plant, heating rate during refining, bottom-blown argon flow rate and stirring mode, refining path, and refining time; it also covers equipment and operation-related data, such as the number of times the ladle is used, empty furnace time, furnace lining condition, and operational data such as tapping time, hot slag addition rate, and timing. For example, data on slag-steel mixing efficiency under different bottom-blown argon flow rates will be used by the model to optimize the stirring coordination scheme during slag blending. Smelting result feedback data is the key basis for the model to verify the effectiveness of the slag blending scheme and for reverse optimization rules, and it is divided into two core data categories: quality and energy consumption. Quality data includes the composition of the molten steel leaving the furnace (such as the final content of elements like sulfur, aluminum, and silicon), the cleanliness of the molten steel (the quantity and size of inclusions), the final quality inspection results of the target steel grade (such as whether it meets the standards for high-quality steel), and the actual composition, basicity, and oxidizing properties of the final slag. Energy consumption and cost data cover the consumption of energy media such as water, electricity, and argon during the smelting process, as well as the consumption costs of various slag materials and alloys, and the total smelting cost corresponding to different slag blending schemes. For example, if a certain slag blending scheme meets the steel quality requirements but has high energy consumption, this type of data will be used by the model to subsequently optimize the cost-quality balance rules.

[0026] The trained machine learning module, in the actual hot slag addition control scenario of steelmaking production, takes real-time collected smelting condition parameters as input and outputs precise slag blending decision parameters. Both are designed around the actual needs of the steelmaking process. Specific input and output data types and examples are as follows: The learning model, after training, is used to obtain the target model as follows: Input data consists of multi-dimensional operating parameters collected in real-time from the production site and pre-processed. These parameters must be consistent with the feature dimensions used during model training. This includes basic data for steel preparation (target steel grade and quality requirements data, as well as basic data for hot slag and furnace slag), real-time data of the smelting process, and process constraint data. Specifically: Target steel grade and quality requirements data: target steel grade, target steel composition, and required cleanliness level; Basic data for hot slag and furnace slag: real-time composition of hot LF slag (CaO, SiO2, Al2O3, FeO, MnO content, etc.), hot slag weight, and hot slag temperature; converter slag discharge amount and composition (e.g., FeO content), and current remaining slag in the furnace. Real-time data for steel and the smelting process: converter tapping temperature, tapped steel weight, initial steel composition (e.g., sulfur, phosphorus, and oxygen content); ladle number (associated with ladle usage count and lining condition), and basic stirring parameters for bottom-blown argon. Process constraint data: Types of fresh slag materials available on-site (e.g., lime, dolomite, pre-melted slag) and their inventory; maximum smelting energy consumption (e.g., electricity consumption ≤ 80 kWh / ton of steel); production rhythm requirements (e.g., refining time ≤ 30 minutes). Output data of the target model in actual use includes: Core usage parameters for hot slag: optimal addition amount of hot LF slag (e.g., 3.6 tons), timing of hot slag addition (e.g., starting addition when steel output reaches 60%), and addition rate (e.g., 0.5 tons / minute); Fresh slag material replenishment decision parameters: whether fresh slag material needs to be added; if so, output the specific type (e.g., lime, dolomite) and amount to be added; replenishment stage. Process co-optimization suggestion parameters: suitable bottom-blown argon stirring mode, mid-term medium flow rate, temperature compensation suggestions for the refining process, and whether deoxidizer needs to be added.

[0027] Because the data in the LF database is extremely important, it is necessary to protect it. Therefore, a stealth box was set up. The stealth box is a cloud server that hosts the LF database. The storage space in the cloud server is divided and isolated according to a preset data storage capacity, resulting in multiple storage ranges. Each storage range is configured with at least two nodes for interconnection between the multiple storage ranges, forming a maze of interconnected storage ranges. The specific connection method is determined by preset connection rules, and the multiple storage ranges are built according to the connection rules. This storage range is provided for unauthorized network access. Multiple transfer surfaces are then stored within the cloud server's storage space. Each storage range connects to multiple transfer surfaces, which are located in different positions within the storage space. These transfer surfaces are used to store the LF scum database. In case of danger, the LF scum database can be moved between these transfer surfaces to evade unauthorized network access, creating a "stealth" effect and thus protecting the LF scum database. To monitor unauthorized network access, multiple sensing points are set up within the storage range. These sensing points are virtual machines stored within the storage range. When an unauthorized network accesses the storage range, it will also access these sensing points. By monitoring the access activity of these sensing points, the dynamics of unauthorized network access can be understood. The adjacent sensing points are interconnected. Sensing points within the interconnected storage range are connected by clamping the nearest sensing point to the node, which serves as the cross-region sensing point. This interconnection ensures that unauthorized networks can continuously access and transfer data within the storage range, effectively deceiving them into believing they are constantly accessing and attacking. Simultaneously, it allows for monitoring the access dynamics of unauthorized networks, enabling effective access tracking and early transfer for subsequent security protection of the LF slag database. This provides strong data protection for the LF slag database, ensuring a secure target model and expert system for accurate decision-making during the steel production process using hot refining slag in converters.

[0028] In one embodiment, step S15, which involves protecting the LF slag database based on the stealth box, includes: S151. Store the LF slag database in any one of the transfer surfaces as the target transfer surface (here, the transfer surface is a virtual machine, and multiple virtual machines are connected to the storage range through the first connection port and the second connection port. The storage resources of the transfer surface can be allocated according to the storage location of the LF slag database. For example, if there are 10 transfer surfaces in the entire stealth box, first select the transfer surface that stores the LF slag database as the target transfer surface, and allocate the storage resources of 9 of the transfer surfaces to the target transfer surface. This can store the LF slag database and also save the resources used for the transfer surface. The transfer surface has no externally open ports, only the storage range has externally open ports. The storage range is the data space that is divided and isolated in the storage space). Set a network open port for each of the multiple storage ranges. S152. When one or more unauthorized networks access one or more storage ranges through the storage space (here, access to the storage range is achieved through the network open port of the storage range), the location of the unauthorized network access is determined by the sensing point. S153. Disconnect the storage range currently not authorized for network access from the transfer plane of the connection; S154. When the management range of the unauthorized network access sensing point includes the target transfer surface, the connection relationship between the transfer surface and other transfer surfaces in the storage range that are not accessed by the unauthorized network is enabled through the transfer mechanism, and the transfer surface in the other storage range that is not accessed by the unauthorized network that is closest to the communication management is identified as the transfer surface to be used. S155. Based on the connection channel between the target transfer plane and the standby transfer plane, the LF slag database in the target transfer plane is transferred to the standby transfer plane. Then, the connection between the target transfer plane and the standby transfer plane is disconnected (after the LF slag database is transferred to the standby transfer plane, the storage resources in the transfer plane will be transferred to the standby transfer plane), thus completing the transfer of the LF slag database and protecting the LF slag database. S156. When an authorized network accesses the storage space, the transfer surface storing the LF slag database is randomly connected to a security port. The LF slag database is accessed through the security port by connecting to the authorized network. Multiple security ports are set according to the storage space. When there is no authorized network access, the security port is in a closed state. After the authorized network exits the access, the enabled security port returns to a closed state.

[0029] As described in steps S151-S156 above, after the stealth box is set up, the LF slag database is stored in any transfer surface. This transfer surface serves as the target transfer surface. A network open port is set for each of the multiple storage ranges to guide unauthorized network access. When one or more unauthorized networks access one or more storage ranges through the storage space, the location of the unauthorized network access is determined by the sensing point. The connection between the currently accessed storage range and the connected transfer surface is then disconnected. The connection between the storage range and the transfer surface is through a connection control interface using the TCP communication protocol. Even if the unauthorized network accesses a storage range that is not the storage range where the LF slag database is located, the connection between the transfer surface and the storage range must still be disconnected. When the management range of the unauthorized network access sensing point includes the target transfer surface, the sensing point and the transfer surface are interconnected. This connection is triggered and does not support access or data transmission; it is only used to trigger the transfer mechanism. The transfer mechanism enables the connection between the transfer surface and other transfer surfaces in storage ranges not accessed by unauthorized networks. It identifies the transfer surface in the nearest unauthorized storage range as the standby transfer surface, transfers the LF slag database from the target transfer surface to the standby transfer surface, and then disconnects the connection between the target transfer surface and the standby transfer surface, completing the transfer of the LF slag database and protecting it. When an authorized network accesses the storage space, the transfer surface containing the LF slag database randomly connects to a secure port. Multiple secure ports are configured for different storage spaces. When there is no authorized network access, the secure ports are closed. After the authorized network exits access, the enabled secure ports return to the closed state, providing continuous protection. Regardless of which transfer surface the LF slag database is in, it can establish a connection with the secure port. The secure interface is the database access interface, using MySQL. The 8.0 protocol grants authorized network access, enabling data access, deletion, and addition. In the event of unauthorized network access, it dynamically transfers and hides the LF slag database, preventing unauthorized access while maintaining normal access for authorized networks. This provides strong data protection, allowing the secure LF slag database to obtain accurate target models and expert systems, enabling accurate decision-making in the steel production process of recycling hot refining slag in converters.

[0030] In one embodiment, step S2, which involves inputting analytical data into a control model and outputting hot slag core usage parameters, fresh slag replenishment decision parameters, and process synergistic optimization suggestion parameters to obtain analytical results, includes: S21. Based on the Internet of Things, real-time data of the steel preparation process, real-time data of the smelting process, and process constraint data are collected in real time as analysis data. S22. Input the analysis data into the target model and use the expert system as constraints to output the analysis results; As described in steps S21 and S22 above, the output data of the target model is constrained by the expert system. Here, the output data of the target model is constrained by the data in the expert system, that is, the target model and the expert system are integrated to obtain the control model.

[0031] In one embodiment, step S3, which involves the on-site operation of preparing steel based on the analysis results, includes: S31. Based on the core usage parameters of hot slag, the decision parameters for adding fresh slag, and the suggested parameters for process synergy optimization, the analysis results are broken down into multiple instructions. S32. Issue instructions to the corresponding steel preparation equipment for control and implement on-site operations.

[0032] As described in steps S31 and S32 above, the slag feeding parameters are broken down according to equipment type: for example, the core usage parameters of hot slag, the decision parameters for fresh slag replenishment, and the process co-optimization suggestion parameters output by the model will be broken down by the system according to the function of the executing equipment, and converted into multiple independent and precise operation instructions. The core usage parameters of hot slag will be broken down into instructions for hot slag feeding equipment, the decision parameters for fresh slag replenishment will be broken down into instructions for discharging fresh slag silos, and the process co-optimization suggestion parameters will be broken down into control instructions for the bottom-blown argon system. The broken-down instructions will be accompanied by constraints such as execution time, execution order, and parameter thresholds to ensure that the operation process conforms to the smelting rhythm. Instruction issuance and equipment execution: realize on-site automated operation.

[0033] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for producing steel based on the recycling of hot refining slag in a converter, characterized in that, Includes the following steps: The LF slag database is identified, and a stealth box is deployed on the LF slag database. The stealth box includes a stealth channel, a transfer surface, and multiple sensing points. The LF slag database is protected based on the stealth box. An expert system is built based on the LF slag database, and a learning model is trained to obtain a target model. The target model is integrated with the expert system to obtain a control model. Real-time data collection of basic steelmaking data, real-time smelting process data, and process constraint data is used as analysis data. The analysis data is input into the control model, and the output parameters include hot slag core usage parameters, fresh slag replenishment decision parameters, and process synergistic optimization suggestion parameters, thus obtaining the analysis results. Based on the analysis results, the on-site operation for steel preparation is carried out until the production of steel is completed by recycling hot refining slag in a converter.

2. The method for producing steel based on the recycling of hot refining slag in a converter, as described in claim 1, is characterized in that... The steps of identifying the LF scum database and deploying a stealth box on the LF scum database include: Collect and store basic slag composition data, historical corresponding smelting process data, and quality and effect correlation data to obtain the LF slag database; A stealth channel is configured for the corresponding LF slag database. The stealth channel consists of multiple storage spaces, which are connected by nodes. Multiple transfer surfaces are configured for the corresponding stealth passage, and these transfer surfaces are interconnected. Multiple sensor points are set up in the stealth channel, and the multiple sensor points are interconnected. In addition, the multiple sensor points are interconnected with multiple transfer surfaces to obtain a stealth box. Protect the LF slag database using stealth boxes.

3. The method for producing steel based on the recycling of hot refining slag in a converter, as described in claim 2, is characterized in that... The steps for configuring a stealth channel for the corresponding LF slag database include: Set up multiple storage spaces, define storage ranges in each storage space, configure at least two nodes for each storage range, connect each node to the storage range in the corresponding storage space, and connect the storage ranges to each other through the nodes. Configure connection rules for multiple storage spaces, where the connection rules define the connection relationships between multiple storage ranges; Multiple storage ranges are connected through nodes according to the connection rules to obtain a stealth channel.

4. The method for producing steel based on the recycling of hot refining slag in a converter, as described in claim 3, is characterized in that... The step of configuring multiple transfer surfaces in the corresponding stealth channel and connecting the multiple transfer surfaces to each other includes: Multiple transfer surfaces are configured for each storage range. A first connection port is set for each transfer surface, and multiple second connection ports are configured for each storage range. The transfer surfaces are matched one-to-one with the second connection ports of the storage ranges. The transfer surfaces are connected to the corresponding second connection ports through the first connection ports. Transfer surfaces are stored in the storage space, and multiple transfer surfaces are interconnected.

5. The method for producing steel based on the recycling of hot refining slag in a converter, as described in claim 4, is characterized in that... The step of setting multiple sensing points within the stealth channel, connecting these sensing points to each other, and connecting these sensing points to multiple transfer surfaces to obtain the stealth box includes: Multiple sensing points are set up within each storage area, and the storage locations of the multiple sensing points within the storage area are determined. Sensors in adjacent storage locations are interconnected. Within different storage ranges, sensors closer to a node are selected as cross-region sensors. Cross-region sensors are connected to the nearest cross-region sensors in other storage ranges through the nearest node. Establish a management range between each sensing point and the transfer surface within its storage range. The management range is the spatial range of at least one transfer surface pre-managed by the sensing point within the storage range. Establish a transfer mechanism between the sensing point and the transfer surface within the management range.

6. The method for producing steel based on the recycling of hot refining slag in a converter, as described in claim 5, is characterized in that... The steps for protecting the LF slag database based on the stealth box include: Store the LF slag database in any one of the transfer surfaces, which will serve as the target transfer surface. Set an open network port for each of the multiple storage ranges. When one or more unauthorized networks access one or more storage ranges through storage space, the location of the unauthorized network access is determined by the sensing point; Disconnect the storage range currently without authorized network access from the transfer plane of the connection; When the target transfer surface is included in the management range of the unauthorized network access sensing point, the connection relationship between the transfer surface and other transfer surfaces in the storage range that are not accessed by the unauthorized network is enabled through the transfer mechanism, and the transfer surface in the other storage range that is not accessed by the unauthorized network that is closest to the communication management is identified as the transfer surface to be used. Based on the connection channel between the target transfer surface and the standby transfer surface, the LF slag database in the target transfer surface is transferred to the standby transfer surface. Then, the connection between the target transfer surface and the standby transfer surface is disconnected, thus completing the transfer of the LF slag database and protecting the LF slag database. When an authorized network accesses the storage space, the transfer surface containing the LF Slag Database is randomly connected to a secure port. The LF Slag Database is accessed through the secure port by connecting to the authorized network. Multiple secure ports are configured for the storage space. When no authorized network accesses the storage space, the secure ports are closed. After the authorized network exits the access, the enabled secure ports are turned off again.

7. The method for producing steel based on the recycling of hot refining slag in a converter according to claim 1, characterized in that, The steps of inputting the analysis data into the control model and outputting the core usage parameters of the hot slag, the decision parameters for adding fresh slag, and the suggested parameters for process synergistic optimization to obtain the analysis results include: The analysis data is based on the real-time collection of basic data, smelting process data and process constraint data during the steel preparation process using the Internet of Things. The analysis data is input into the target model and constrained by an expert system, and the analysis results are output.

8. The method for producing steel based on the recycling of hot refining slag in a converter according to claim 1, characterized in that, The steps for on-site steel preparation based on the analysis results include: Based on the core usage parameters of hot slag, the decision parameters for adding fresh slag, and the suggested parameters for process synergy optimization, the analysis results are broken down into multiple instructions. The instructions are sent to the corresponding steel preparation equipment for control and on-site operation.

Images