Stainless steel smelting intelligent batching method and system based on multi-furnace rolling plan

Abstract

The embodiment of the application discloses a stainless steel smelting intelligent batching method and system based on multi-furnace rolling planning, which comprises the following steps: obtaining production planning data of each furnace in a unit time, obtaining material inventory data and various process feeding constraint data, and preprocessing the above data as model input data; calling a batching model, inputting the model input data into the batching model, solving by using the batching model, and outputting a steel grade batching list of each furnace; the batching model is solved under the constraint of multiple constraint conditions, and the total cost of all materials used in all processes of all furnaces is minimized as a target function; when the steel grade batching list of each furnace is audited, the corresponding material inventory is locked, and charging and smelting are carried out according to the steel grade batching list. The application achieves fine steel grade batching of a single furnace, and realizes intelligent, accurate and optimal batching process.

Description

Technical Field

[0001] This application relates to the field of intelligent manufacturing in the iron and steel metallurgy industry, specifically to an intelligent batching method and system for stainless steel smelting based on a multi-furnace rolling plan. Background Technology

[0002] In the stainless steel production sector of steel enterprises, the alloy and scrap steel batching process is a critical step that determines the final product quality, production cost, and production efficiency. However, existing batching management methods have many significant shortcomings in practical applications and are no longer suitable for the high precision, high efficiency, and low cost requirements of modern industrial production. Specifically, existing technologies mainly face the following technical bottlenecks: 1. High reliance on manual labor, low calculation efficiency, and prone to errors: The current batching process is highly dependent on manual operation, relying on the personal experience of batching technicians for material selection and proportion calculation. Automation and intelligence are insufficient, and batching plans vary significantly among different technicians, lacking standardization and consistency. Staff typically rely on static inventory data (1-2 times daily) to manually input raw material weights through Excel spreadsheets or basic system forms, and use simple formulas to estimate the steel composition. Manual batching struggles to accurately control the fluctuation range of various elemental compositions, frequently resulting in unmet AOD (Argon Oxygen Decarburization) target compositions and unstable product quality in actual production. This traditional method is not only cumbersome and time-consuming, severely slowing down production, but also highly susceptible to composition control failures due to human error in data entry or calculation deviations, increasing quality risks and labor costs.

[0003] 2. The material batching scheme is rudimentary and lacks refined control: Existing material batching sheets are mostly presented as master plans, lacking personalized customization for specific production plans. This often results in multiple plans for the same steel grade sharing a single material batching sheet, leading to poor scheme targeting. Furthermore, existing technology lacks precision in material management, only specifying the material type without accurately linking it to its specific physical location in the warehouse, such as the orientation and layer of the material pile. This lack of spatial information makes material retrieval path planning difficult, increasing on-site operation time and the probability of mismatches, thus hindering the collaborative efficiency of logistics and production flow.

[0004] 3. The system is static and rigid, lacking dynamic response and emergency adjustment mechanisms: The batching method is statically locked, unable to respond in real time to frequent changes in on-site production plans. When encountering abnormal operating conditions such as sudden equipment failures or raw material supply interruptions, the system lacks the ability to dynamically recalculate and optimize based on real-time data, and cannot quickly generate emergency response plans. The production site can only passively wait for manual recalculation, leading to production stoppages, delays, and severely weakening the robustness of the production system.

[0005] 4. Lack of a holistic perspective and insufficient cost optimization capabilities: Existing batching processes are often limited to achieving the required composition for a single batch, making it difficult to comprehensively consider the cost factors of continuous production across multiple batches from a holistic perspective. Due to the inability to integrate dynamic information such as raw material market price fluctuations and equipment energy consumption curves in real time, existing technologies cannot achieve optimal configuration of the overall cost function, leaving enterprises in a passive position regarding raw material procurement and usage strategies, and making it difficult to explore potential cost reduction opportunities.

[0006] 5. Frequent material shortages trigger a chain reaction in production, driving up overall costs: Due to the lack of accurate inventory early warning and supply-demand matching mechanisms, material shortages and disruptions occur frequently during production, becoming a key obstacle to production continuity. This not only leads to frequent start-ups and shutdowns of production lines, idle equipment, and wasted labor hours, but also disrupts the close coordination between processes. More seriously, to cope with material shortage crises, the site is often forced to use high-cost backup plans, such as purchasing expensive rush-order raw materials or using low-cost-performance substitutes, resulting in a significant increase in the cost per ton of steel.

[0007] Therefore, how to free stainless steel batching from manual labor, achieve precise material positioning, enable dynamic real-time adjustments, and optimize overall costs has become a pressing technical challenge for the steel and metallurgical industry. Summary of the Invention

[0008] In view of the above problems, this application proposes an intelligent batching method and system for stainless steel smelting based on multi-furnace rolling planning, which is used to solve the following problems: in the existing stainless steel batching process, there is a high degree of dependence on manual labor, the scheme is extensive, there is a lack of dynamic adjustment capability and global cost optimization.

[0009] According to one aspect of the embodiments of this application, a smart batching method for stainless steel smelting based on a multi-heat rolling plan is provided, comprising: The production plan data for each furnace within a unit of time is obtained, as well as the material inventory data and various process feeding constraint data. The production plan data, material inventory data, and various process feeding constraint data are preprocessed and then used as model input data. The batching model is invoked, and the model input data is input into the batching model. The batching model is then used to solve the problem and output the steel grade batching list for each heat. The batching model is solved under multiple constraints with the objective function of minimizing the total cost of all materials used in all processes of all heats. Once the steel batching list for each heat is approved, the corresponding material inventory is locked, and charging and smelting are carried out according to the steel batching list for each heat.

[0010] Furthermore, before invoking the ingredient model, the method also includes: Multiple constraints are constructed, and the total cost of all materials used in all processes across all furnaces is minimized as the objective function to construct the batching model.

[0011] Furthermore, multiple constraints include: production planning constraints, process material balance constraints, process capacity constraints, material input constraints, material inventory constraints, steel grade material restriction constraints, material batch selection constraints, charging efficiency constraints, and composition balance constraints.

[0012] Furthermore, inputting the model input data into the ingredient model further includes: For each heat, determine the steel grade to be produced based on the production plan data of the current heat, traverse the material library, remove prohibited material types in the process specifications of the steel grade, and generate a whitelist of available materials. For each type of material in the available material whitelist, sort all inventory batches of that type of material according to unit price from low to high, and select the inventory batches with the highest preset quantity as decision variables to input into the batching model.

[0013] Furthermore, by using a batching model to solve the problem, the output of the steel batching list for each heat includes: Each heat cycle is grouped according to the composite primary key formed by steel grade and process route to form multiple steel families, which are then added to the steel family list. Perform rolling temporal optimization independently on each steel race in the steel race list to obtain the target batching optimization result for each steel race; Summarize the target batching optimization results for each steel class to form a global cost optimization plan, and determine the steel batching list for each heat based on the global cost optimization plan.

[0014] Furthermore, rolling temporal optimization is performed independently for each steel race in the steel race list, and the target batching optimization results for each steel race further include: Based on the planned time in the production plan data of each heat, sort the multiple steel types in the steel type list to form a baseline sequence; For the baseline sequence, intra-group rolling optimization is performed on each steel race in the baseline sequence to obtain the material optimization results for each steel race. The cost of each steel race is calculated and accumulated to obtain the baseline total cost corresponding to the baseline sequence. Adjust the execution order among the various steel races to generate multiple candidate sequences; For each candidate sequence, perform intra-group rolling optimization for each steel race in the candidate sequence to obtain the batching optimization results for each steel race. Calculate the cost of each steel race and sum them up to obtain the total candidate cost corresponding to the candidate sequence. Based on the baseline total cost and multiple candidate total costs, a target sequence is selected from the baseline sequence and multiple candidate sequences, and the batching optimization result of each steel race corresponding to the target sequence is taken as the target batching optimization result of each steel race.

[0015] According to another aspect of the embodiments of this application, a smart batching system for stainless steel smelting based on a multi-furnace rolling plan is provided, comprising: The data acquisition module is suitable for acquiring production plan data for each furnace within a unit of time, acquiring material inventory data, and various process feeding constraint data; The data preprocessing module is suitable for preprocessing production planning data, material inventory data, and various process material feeding constraint data before using them as model input data; The model processing module is suitable for calling the batching model, inputting the model input data into the batching model, using the batching model to solve the problem, and outputting the steel grade batching list for each heat. The batching model is solved under multiple constraints with the objective function of minimizing the total cost of all materials used in all processes of all heats. The execution module is suitable for locking the corresponding material inventory after the steel grade batching list for each heat is approved, and for charging and smelting according to the steel grade batching list for each heat.

[0016] According to another aspect of the embodiments of this application, a computing device is provided, including: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other through the communication bus; The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan.

[0017] According to another aspect of the embodiments of this application, a computer storage medium is provided, wherein at least one executable instruction is stored in the storage medium, the executable instruction causing a processor to perform the operation corresponding to the above-described intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan.

[0018] According to another aspect of the embodiments of this application, a computer program product is provided, including at least one executable instruction that causes a processor to perform operations corresponding to the above-described intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan.

[0019] According to the technical solution provided in this application, a batching model is constructed. The batching model takes minimizing the total cost of all materials used in all processes across all heats as the objective function and simultaneously embeds multiple constraints. The batching model is used to solve the problem, outputting steel grade batching lists for each heat. This achieves highly available, cost-optimal, and feasible automatic generation of steel grade batching lists, realizing refined batching of steel grades for a single heat and transforming the process from manual calculation to automated real-time optimization. During the solution process, a four-layer collaborative optimization system is adopted: constraint variable pre-screening—heat decomposition—constraint local fine-tuning—global optimization. This system can efficiently solve complex batching optimizations involving multiple materials, multiple steel grades, multiple heats, and multiple process routes. A precise batching mechanism based on material inventory location is established, achieving a single batch... The refined management of this solution enables real-time rescheduling under production disruptions to cope with emergencies and ensure production continuity. It also optimizes global costs based on multi-dimensional dynamic factors such as raw material prices, inventory, and production rhythm, fundamentally preventing passive high-price substitutions due to material shortages. This enhances the scientific nature of the material batching process for technical personnel, significantly improving batching efficiency, accuracy, and overall steel plant benefits. The solution achieves intelligent batching for multiple heats of stainless steel based on multi-constraint coupled modeling and an efficient hybrid solution strategy. It effectively solves problems in existing technologies such as high reliance on manual labor, static and coarse batching schemes, inability to dynamically respond to production disruptions, and lack of global cost coordination. This achieves intelligent, precise, and optimized batching processes, significantly reducing the overall cost per ton of steel and ensuring continuous and stable production operation.

[0020] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of the embodiments of this application are described below. Attached Figure Description

[0021] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the embodiments of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A schematic flowchart of a smart batching method for stainless steel smelting based on a multi-heat rolling plan according to an embodiment of this application is shown. Figure 2 The flowchart of the grouped rolling optimization algorithm based on steel races in this application is shown. Figure 1 ; Figure 3 The flowchart of the grouped rolling optimization algorithm based on steel races in this application is shown. Figure 2 ; Figure 4 A schematic diagram illustrating the principle of a smart batching method for stainless steel smelting based on a multi-heat rolling plan according to an embodiment of this application is shown. Figure 5 A structural block diagram of an intelligent batching system for stainless steel smelting based on a multi-furnace rolling plan according to an embodiment of this application is shown. Figure 6 A schematic diagram of the structure of a computing device according to an embodiment of this application is shown. Detailed Implementation

[0022] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0023] In existing stainless steel batching processes, multiple steps are highly coupled, production disturbances are frequent, and human experience is insufficient for global optimization, easily leading to the neglect of key indicators. To facilitate understanding of this application, the relevant background of the stainless steel batching scenario will be introduced below.

[0024] Steelmaking materials are complex. Stainless steel production involves more than 100 materials, including high-carbon ferrochrome, low-carbon ferrochrome, nickel plates, ferrosilicon, and scrap steel. Each material has strict usage restrictions in different processes, and the composition content of the same material varies significantly between different batches. The use of different materials will affect the final cost per ton of steel. Existing manual batching methods are difficult to systematically handle such complex material constraints.

[0025] The stainless steel batching scheme needs to simultaneously meet three mutually restrictive objectives: accurate composition, reasonable sourcing of inventory, and cost minimization. These multi-objective optimizations present inherent conflicts. Accurate composition: The final molten steel must meet the steel composition standards for each process stage. Different steel grades have different output composition requirements, specifically related to the contract grade. Reasonable sourcing of inventory: Materials are stored dispersedly in scrap yards, low-level silos, and high-level silos. The scrap yard employs a grid management system, with scrap from different manufacturers and batches stored in different grid areas. During batching, materials need to be transferred to troughs and containers using overhead cranes, loading machines, etc. Therefore, the selection of material batches must comprehensively consider loading distance, transportation time costs, and actual inventory limitations at each storage location to ensure the reasonableness of material sourcing and production efficiency. Cost minimization: Under the premise of reasonable sourcing of inventory, meeting process output composition requirements, and meeting material usage restrictions, the total material cost should be minimized.

[0026] The process routes for stainless steel smelting are diverse. The material flow paths of various process routes differ greatly. The composition of molten steel produced in the melting process (alloy melting furnace, electric furnace, converter, three-stage stripping station, etc.) directly affects the material addition strategy in the AOD process. Traditional single-process optimization methods cannot establish a cross-process material-composition-cost hybrid optimization model, making it difficult to obtain the global optimal solution for cost, composition, etc.

[0027] Furthermore, constraints such as material input, steel output capacity, composition yield loss coefficient, and steel material loss in each process are intertwined, making multi-constraint coordination difficult. When a process adjusts its batching due to equipment limitations, material supply shortages, or composition fluctuations, it will have a cascading effect on the material requirements and composition control of subsequent processes. The existing batching method lacks an effective global constraint coordination mechanism.

[0028] To address the problems of high reliance on manual labor, crude methods, lack of dynamic adjustment capabilities, and lack of global cost optimization in existing stainless steel batching processes, this application constructs a batching model that transforms manual calculation into automated real-time optimization. It enables precise batching based on the physical location of materials, achieving intelligent and refined batching control for multiple heats, steel grades, and process routes. Based on multi-dimensional dynamic factors such as raw material prices, inventory, and production rhythm, it optimizes global costs, fundamentally eliminating passive high-price substitutions caused by material shortages, and significantly improving batching efficiency, accuracy, and overall enterprise benefits.

[0029] Figure 1 A schematic flowchart of a smart batching method for stainless steel smelting based on a multi-heat rolling plan according to an embodiment of this application is shown, as follows: Figure 1 As shown, the method includes the following steps: Step S101: Obtain production plan data for each furnace within a unit time period, obtain material inventory data and various process feeding constraint data, and use the preprocessed production plan data, material inventory data and various process feeding constraint data as model input data.

[0030] To address the core challenges of complex material coupling and superposition effects and poor model scalability and adaptability caused by the rigidity of process topology in multi-stage collaborative batching of pre-solvent solutions, this application proposes a dynamic coupling modeling and collaborative optimization mechanism for multi-stage pre-solvent solutions based on abstract set representation. Specifically, a complete abstract set of processes is defined, which can be represented as... This complete set abstractly represents all process types that may participate in the batching process, where AMF represents an alloy melting furnace, EAF represents an electric furnace, BOF represents a converter, and AOD represents an argon-oxygen decarburization furnace. A subset of melting processes is defined, which can be represented as... This subset explicitly identifies the specific process instance that performs the melting function. The refining process only contains AOD, so no subset needs to be defined; the AOD identifier is used directly. The topological connections, material flow constraints, and information interaction rules between processes are dynamically defined through set operations (such as union ∪, intersection ∩, etc.), realizing the construction of dynamic relationships based on set operations.

[0031] This application embodiment also includes a multi-process composition transfer and control algorithm. Specifically, the steel output calculation for the pre-solution process is based on the total amount of raw materials input and the yield of the pre-solution process (which can be dynamically calibrated using historical data). The elemental composition synthesis for the pre-solution process involves weighted calculation of the initial concentration of each element in the molten steel based on the element content of each raw material and its yield factor during smelting (which can be dynamically calibrated using historical data). The steel output calculation for the AOD process combines the steel output from the pre-solution process and the amount of alloy added, calculating the total amount of refined molten steel using the AOD process yield and mechanistic model. The elemental composition synthesis for the AOD process dynamically synthesizes the final refined composition by integrating the mixed composition of the molten steel after the pre-solution process (considering element retention rate) with the element content of the added materials (including alloy yield).

[0032] This application embodiment connects to the steel plant's MES system, secondary system, ironworks system, and resource system to acquire production plan data, material inventory data, and various process feeding constraint data within a unit time (e.g., daily). The acquired production plan data, material inventory data, and various process feeding constraint data undergo data governance, preprocessing them into a model input format to provide structured and standardized model input data for the batching model. This facilitates the batching model's solution and outputs multi-heat, multi-steel-grade steel batching lists. Taking daily time as an example, the daily production plan for multiple heats, multiple steel grades, and multiple process routes of stainless steel is used as the optimization object of the model to achieve the lowest total raw material cost for all heats.

[0033] This system can connect to the steel plant's intelligent scheduling system to obtain daily production plan data for 40-60 heats of stainless steel, grouping and summarizing the number of heats by steel type and process route. For example, if steel type A and process route a correspond to 4 heats, it can be represented as "Steel type A + route a: 4 heats"; if steel type A and process route b correspond to 5 heats, it can be represented as "Steel type A + route b: 5 heats". To facilitate the division of process routes, 9 process routes are defined according to actual application needs, clarifying the process sequence of each process route. These routes consist of different combinations of alloy melting furnaces (maximum 3) and electric furnaces (maximum 1), etc. The steel output from the preceding processes is summed and then added to the AOD furnace. For example, a certain process route is alloy melting furnace - electric furnace - AOD.

[0034] Table 1 shows the various process routes and their codes, where AMF represents an alloy melting furnace, EAF represents an electric furnace, BOF represents a converter, and AOD represents an argon-oxygen decarburization furnace. Table 2 shows the production plan data for a certain day.

[0035] Table 1. Explanation of each process route and its code

[0036] Table 2. Production Plan Data for a Certain Day

[0037] Material inventory data is obtained by connecting to the MES (Manufacturing Execution System) and CE (Cost and Profitability) systems. The MES system is a production information management system for the shop floor execution layer of manufacturing enterprises, while the CE system is a cost and profitability management information system. The material inventory data package contains the inventory, composition, and price of raw materials such as alloys and scrap steel. Specifically, material inventory data can be obtained from the material databases of the aforementioned systems. This data may include the composition of each material (e.g., the content of elements such as C, Si, and Mn), unit price (e.g., differentiated by batch), batch inventory, and the grid number of the storage location (e.g., scrap steel yard silo number). The material inventory data in the database is updated based on real-time inventory, updating the remaining inventory and storage location of each batch of materials in real time, for example, updating once per minute. Table 3 shows the material inventory data for some materials.

[0038] Table 3. Inventory data for some materials

[0039] Obtain various process feeding constraint data, including process steel output constraints, yield constraints, scrap steel type and weight constraints, alloy type and weight constraints, etc. For example, the input-output yield coefficient, steel output upper and lower limits, whether materials are allowed to be added, and material addition upper and lower limits for each process (such as alloy melting furnace, electric furnace, AOD, etc.).

[0040] Among these, there are many material names, which require management by major categories from the perspective of production site use. For example, cutting slag and stainless steel shavings (foam) both belong to slag steel, thus achieving material classification management. In addition, for the same steel grade, due to slight differences in some elements, a new steel tapping mark will be generated. For ease of management, the steel tapping marks need to be managed by major categories, thus achieving steel grade classification management.

[0041] Key processing steps in the data preprocessing process may include: establishing a material-inventory location mapping, establishing a static inventory location mapping relationship for each material batch, recording the physical location coordinates and price information of the batch, the operating radius of the steel grabber, the division of the storage area, and converting process constraints (such as composition range, process loading rules, etc.) into mathematical constraint expressions.

[0042] Step S102: Call the batching model, input the model input data into the batching model, use the batching model to solve, and output the steel grade batching list for each heat.

[0043] Specifically, the batching model is a mathematical model containing multiple constraints and an objective function. Under these constraints, the batching model aims to minimize the total cost of all materials used in all processes across all furnace batches as the objective function. These multiple constraints are dynamically set, taking into account factors such as changes in production plans, the physical distribution of raw materials, real-time market price fluctuations, equipment operating status, and inventory supply and demand balance.

[0044] Data governance is performed on material and cost data, which are then sent to the model for training to build the final batching model. Specifically, multiple constraints are constructed, and the objective function is to minimize the total cost of all materials used in all processes across all heats, resulting in the batching model. These constraints are managed uniformly to ensure the feasibility of the optimization results. Core constraint types include: composition constraints, process constraints, resource constraints, and cost constraints. Each material is labeled with its corresponding inventory location; for example, scrap steel batch #101 is in grid G3, and batch #102 is in grid G5.

[0045] Among them, composition constraints are aimed at the key elements (such as C, Si, Mn, P, S, Cr, Ni, Mo, V, Ti, etc.) required for different steel smelting processes. The focus is on establishing element content balance equations for high-value alloying elements Cr and Ni, requiring that the composition of the steel batches for each heat strictly fall within the process target composition range of the steel grade, so as to ensure composition consistency and quality stability.

[0046] Process constraints include mandatory process requirements such as the total amount of raw materials charged in different processes not exceeding the upper limit of the furnace charging capacity of the process, and the amount of high-carbon ferrochrome used in a single furnace not less than the lower limit (e.g., 30 tons), as well as the time sequence logic constraints of material flow between processes.

[0047] Resource constraints require that the actual usage of each raw material must not exceed the upper limit of its real-time available inventory, while also mandating the use of lower-priced scrap steel for each heat to reduce raw material costs, thus achieving the dual goals of optimal resource allocation and cost control.

[0048] The cost constraint is that the cost of raw materials for a single furnace batch does not exceed the preset target cost per ton of steel, while also minimizing the total cost per unit time.

[0049] For upstream processes (e.g., alloy melting furnaces or electric furnaces): Steel output = Input amount of this process × Process yield coefficient (considering material loss).

[0050] For the AOD process: Steel output = (Total steel output from preceding processes + AOD input for this process) × AOD yield coefficient (considering material loss). It should be noted that only AOD requires the sum of preceding outputs; other processes are calculated independently.

[0051] Specifically, the constraints include: production planning constraints, process material balance constraints, process capacity constraints, material input constraints, material inventory constraints, steel grade material restriction constraints, material batch selection constraints, charging efficiency constraints, and composition balance constraints.

[0052] (1) Production planning constraints can be expressed as: ,

[0053] Where f represents the furnace number; F represents the set of furnace numbers. N represents the total number of furnace trips per day, 40≤N≤60; The indicator parameter represents the process route r corresponding to furnace f. This indicates that heat f uses process route r, otherwise it is 0; G represents the steel grade set. , R represents the total number of steel grades; R represents the set of process routes. , Indicates the total number of process routes; This is an indicator function; it takes a value of 1 when the steel grade of heat f is g, and 0 otherwise. This represents the planned number of heats for steel grade g using process route r. The meaning of this production planning constraint is to ensure that the actual number of heats produced for each steel grade g using process route r equals the planned number. .

[0054] (2) The material balance constraint of the process can be expressed as: ,

[0055] Where f represents the furnace number; F represents the set of furnace numbers; s represents the process; and S represents the set of processes. , This represents the total number of processes, including alloy melting furnace, electric furnace, AOD, converter, and three-stage desulfurization; m represents the material; M represents the set of material types. , 'b' represents the total number of material types; 'b' represents the material batch; 'B' represents the set of material batches. , This represents the total number of batches of a certain material; z represents the inventory grid; Z represents the set of scrap yard inventory grids. , Indicates the total number of grid cells in the inventory area; X fsmbz This indicates the input weight (in tons) of material batch b from storage grid z in process s for furnace f. fs Input represents the total input (tons) of furnace f in process s; fs The expression means that the total input of furnace f in process s is equal to the sum of the input of all materials in that process.

[0056] ,

[0057] Among them, C s1 This represents the input-output ratio (yield) of process s1 (alloy melting furnace). fs1 Output represents the total input (tons) of furnace f in process s1; fs1 Output indicates the amount of steel produced (in tons) in process s1 of heat f; fs1 The expression means that the amount of steel produced in furnace f in process s1 is equal to the input quantity multiplied by the input-output coefficient.

[0058] ,

[0059] Among them, C s2 This represents the input-output ratio (yield) of process s2 (electric furnace). fs2 Output represents the total input (tons) of furnace f in process s2; fs2 Output indicates the amount of steel produced (in tons) in process s2 of heat f; fs2 The expression means that the amount of steel produced in furnace f in process s2 is equal to the input quantity multiplied by the input-output coefficient.

[0060] ,

[0061] Among them, C s3 Indicates the input-output ratio (yield) of process s3 (AOD); Input fs3 Output represents the total input (tons) of furnace f in process s3; fs3 Output indicates the steel output (tons) of heat f in process s3; fs3 The expression means that the output of AOD furnace is equal to (the output of the previous process plus the input of this process) multiplied by the input-output coefficient of AOD furnace.

[0062] The preceding processes (alloy melting furnace, electric furnace, three-stage desulfurization, etc.) only consider the amount added in this process, and only the final process (AOD) considers the output of the preceding processes.

[0063] (3) The process capacity constraint can be expressed as: ,

[0064] in, Indicates the minimum steel output limit (tons) for process s; This indicates the maximum steel output limit (tons) for process s. This represents the steel output (in tons) of heat f in process s. The above expression means that the steel output of each process must be within the process's capacity; it cannot be lower than the minimum output limit, nor can it exceed the maximum output limit. This is a fundamental requirement for safe operation and process quality.

[0065] (4) The constraint on the amount of material added can be expressed as: ,

[0066] in, This indicates the minimum amount (in tons) of material m to be added in process s. This represents the maximum allowable amount (in tons) of material m in process s. The above expression means that the amount of each material added in each process must meet the upper and lower limits of the process requirements. For example, adding too little of some materials will affect the smelting effect, while adding too much may lead to excessive composition; some materials are not allowed to be added in certain processes, etc.

[0067] (5) Material inventory constraints can be expressed as: ,

[0068] in, This represents the current inventory (in tons) of material batch b in batch m. The above expression means that the total amount of material batch b used in all processes across all furnace batches cannot exceed the current inventory of that batch, ensuring the feasibility of material usage.

[0069] (6) The material restrictions on steel grades can be expressed as: ,

[0070] in, This indicates the type of steel corresponding to heat number f; Indicates steel type The set of permitted material types The above expression means that for the steel grade corresponding to heat f... Only the material types permitted by this steel grade can be used. If material m is not in the steel grade... Permitted material set If the input of this material in all processes of that furnace batch is zero, then the input amount of this material must be zero.

[0071] (7) The material batch selection constraint can be expressed as: ,

[0072] in, ,Right now A variable that can be 0 or 1 indicates whether to select or not to use the material batch; =1 indicates that furnace f uses material batch b (m) in process s; otherwise, it is 0. Variable X can only take a non-zero value when variable Y is activated (equal to 1); if Y is 0, then X must be forced to 0; if furnace f does not use material batch b (m) in process s (i.e., ... If the input of the material is 0, then the input amount of the material in this furnace and this process must be 0.

[0073] In this embodiment, materials are combined in batches as the smallest unit, taking into full account the price differences between different batches of materials, thus achieving refined cost control.

[0074] (8) The loading efficiency constraint can be expressed as: ,

[0075] in, A variable that is either 0 or 1. =1 indicates that heat f uses material m from grid z in process s; otherwise, it is 0. For each heat of steel in each process, the number of grids from which material m originates in the inventory cannot exceed 2.

[0076] ,

[0077] in, This indicates the grid position of material batch b in batch m. The above expression means that if furnace f uses material m from grid z in process s, then It must be 1.

[0078] ,

[0079] The above expression means that if material batch m (b) is used, then the grid containing that batch must be used. Materials.

[0080] The scrap yard is artificially divided into storage grids to store different batches of scrap steel and alloys. If the same material source is used in more than two storage grids when charging each furnace, it will lead to excessively long charging time. This constraint ensures the feasibility of the charging operation.

[0081] (9) The compositional equilibrium constraint can be expressed as: ,

[0082] in, This indicates the content (%) of steel composition k in the steel produced by process s at heat f; K represents the set of constituent elements. , It indicates the total number of constituent elements, including C, Si, Mn, P, S, Cr, Ni, etc. This represents the percentage content of component k in batch b of material m. The above expression means that the steel output composition of alloy melting furnaces and electric furnaces is calculated by a weighted average of the components of the input materials.

[0083] ,

[0084] in, This indicates the content (%) of steel composition k in the steel produced by process s3 of heat f. This indicates the content (%) of steel composition k in the steel produced by process s1 in heat f. This indicates the content (%) of steel composition k in the steel produced by process s2 in heat f; X fs3mbz This indicates the weight (in tons) of material batch b from inventory grid z used in process s3 for furnace f. The above expression means that the steel composition of the AOD furnace is calculated as a weighted average of the steel composition from the preceding process and the composition of the materials input in this process.

[0085] ,

[0086] in, This indicates the minimum content requirement (%) of component k in steel grade g. This indicates the maximum required content (%) of component k in steel grade g. The composition of steel produced by AOD must meet the target composition requirements for steel grade g.

[0087] The constructed objective function is:

[0088] in, This represents the purchase price (yuan / ton) of material m in batch b; X fsmbzThe usage is specifically represented by the weight (in tons) of material batch b from inventory grid z used in furnace f in process s. The objective function is to minimize the total cost of all materials used in all processes across all furnaces.

[0089] This application constructs a batching model by receiving production plan data for 40 to 60 heats of stainless steel per day and summarizing it according to steel type and process route; establishing a process co-optimization model including converter, three-stage desulfurization, alloy melting furnace, electric furnace, and AOD refining furnace; dividing the scrap yard into multiple grids and setting a constraint that the same material source does not exceed two grids when charging each heat of steel; performing combination optimization with material batches as the smallest unit, considering the price differences of different batches of materials; and using mixed integer programming combined with heuristic algorithms to solve the optimization model to minimize the total cost.

[0090] This application integrates with the steel plant's MES, secondary systems, and CE systems to form a complete batching system. The batching model is invoked, and pre-processed model input data is input into the model. The model is then used to solve the problem, outputting the steel grade batching list for each heat, thus generating a detailed batching plan for each heat. The steel grade batching list may include: basic heat information, process-level bill of materials, composition information, and cost details. Specifically, the basic heat information includes: steel grade, heat number, process name, process route, and planned time. The process-level bill of materials includes: the types of materials input into each process, material batches, inventory grid positions, the weight (tons) of each material input into the process, and the expected steel output (tons) for the process. Composition information includes: the composition content of materials and the predicted composition content of steel output from each process. Cost details include: the cost of each material and the total cost.

[0091] Because the batching model involves the overall batching of the daily stainless steel smelting plan, continuous production plan data for more than 45 heats, and needs to consider more than 100 materials, each with multiple batches, and the coupling constraint between material batches and inventory grid locations, the dimensionality of decision variables grows exponentially. These decision variables include: material input variables (such as...) , , ) and process flow variables (such as , , Meanwhile, the grid constraint of inventory locations (≤2 sources of the same material in inventory locations) leads to combinatorial explosion characteristics, making it difficult for traditional single mathematical programming solvers to converge in a short time. Therefore, this application adopts a four-layer collaborative optimization system of constraint variable pre-screening, furnace decomposition, constraint local fine-tuning, and global optimization for solving the problem. The first layer is constraint variable pre-screening, which quickly identifies and eliminates obviously infeasible raw material combinations and constraints based on historical data and expert rules, significantly reducing the problem dimension and solution complexity. The second layer is furnace decomposition, which decomposes the multi-furnace coupled optimization problem into independent optimization subproblems with steel grade and process route as the smallest unit, and solves them synchronously using parallel computing technology, significantly improving computational efficiency. The third layer is constraint local fine-tuning, which adaptively adjusts local constraints based on the material dependencies between processes and dynamic changes in inventory for each furnace optimization result, ensuring that the batching list for all furnaces can be given at once. The fourth layer is global optimization, which integrates the optimization results of each furnace and uses a global search algorithm to further optimize the total cost per unit time under the premise of satisfying all hard constraints, obtaining the globally optimal batching scheme.

[0092] Through a four-layer collaborative optimization system, this application is no longer limited to the constraints of scrap steel bulk density and composition in a single furnace. Instead, it takes into account the collaboration of multiple furnaces and the hard constraints of complex processes. This breaks through the single optimization logic of existing technologies that only aim to maximize the steelmaking efficiency of scrap steel. It effectively handles the resource allocation conflicts and strong process constraints among multiple furnaces. This application abandons the approach of relying on black-box deep learning models and instead constructs a solution framework with clear hard constraint mechanisms and high interpretability.

[0093] Specifically, before solving the problem, the massive amount of raw data, including inventory and available materials, is first cleaned and its dimensionality reduced to eliminate invalid variables, narrow the solution space, and achieve pre-screening of constraint variables. Specifically, for each heat, the steel grade to be produced is determined based on the current heat's production plan data, and the material library is traversed to eliminate steel grades that are not required. The process specifications prohibit certain material types, generating a whitelist of available materials to achieve process availability screening. For example, some precious alloys can only be used with certain steel grades, while some inexpensive materials should be used as much as possible. Then, for each category of material in the whitelist, all inventory batches of that material are sorted according to unit price. The inventory batches are sorted from low to high, and the top-ranked batches (with a preset quantity) are selected as decision variables and input into the batching model. The remaining high-priced batches are directly marked as unavailable, thus achieving price gradient sorting and truncation. Those skilled in the art can set the preset quantity according to actual needs; for example, setting the preset quantity to 5 indicates selecting the top 5 cheapest inventory batches.

[0094] After pre-screening the constraint variables, heat decomposition is performed, and parallel optimization is conducted based on single-heat cost optimization. Specifically, the originally large-scale, nondeterministic, polynomial-difficulty mixed-integer programming problem involving multiple heats is decoupled into N independent single-heat optimizations. The decoupling logic temporarily ignores cost fluctuations caused by the use of materials in different steel grades and heats, and ignores the constraints of inventory usage order across heats. Parallel computation is used to construct a single-heat optimization sub-model for each heat, and the objective function of the sub-model is adjusted to minimize the single-heat cost, while retaining the specific composition constraints, weight constraints, and mesh constraints of that heat.

[0095] During the single-heat solution process, if the sub-model has no solution due to insufficient inventory or overly tight constraints, the constraint relaxation mechanism will be automatically triggered. The constraints will be gradually relaxed in order of priority, realizing local fine-tuning of constraints to ensure that a feasible solution is obtained. The steel grade batching list for that heat will be generated, realizing feasibility repair based on constraint graded relaxation.

[0096] The constraint relaxation mechanism includes relaxation priority rules, which include: Level 1 relaxation, Level 2 relaxation, Level 3 relaxation, and a feedback mechanism. Specifically, Level 1 relaxation is used to relax the target value of steel output. For example, if the steel output of a process must meet the upper limit requirement, and this cannot be met, the upper and lower limits are each allowed to be relaxed by 1 ton. Level 2 relaxation is used to fine-tune the target value of process composition. If there is still no solution after Level 1 relaxation, the target value of the pre-solution composition is allowed to fluctuate slightly within the standard specification range. For example, the target chromium content is shifted from the median specification towards the surplus direction by 0.0001%. Level 3 relaxation is used for high-price substitution. If Level 1 and Level 2 relaxation are ineffective, the price truncation restriction in the price gradient sorting and truncation is temporarily lifted, allowing the model to call substitute materials with slightly higher prices but within the available material whitelist. The feedback mechanism records the solution and its corresponding relaxation cost immediately upon obtaining a feasible solution and stops further relaxation.

[0097] Because of the sequential solution and rolling inventory update mechanism, a globally feasible solution is automatically obtained when the last heat is solved. The system automatically accumulates the costs of each heat to obtain the overall total cost. In the sequential solution process, the order of heats is fixed, and this approach may not yield the globally optimal solution. To obtain the overall optimal solution, a grouped rolling optimization algorithm based on steel species was designed, which realizes global cost verification and optimization of the total cost of each heat.

[0098] This application achieves globally optimal control of batching costs and precise and stable composition accuracy by introducing intelligent optimization algorithms and real-time dynamic adjustment mechanisms. It also provides batching technicians with a comprehensive batching solution. A precise batching mechanism based on material inventory location is established, achieving refined management on a case-by-case basis. It has the ability to instantly reschedule under production disturbances to cope with emergencies and ensure production continuity. Furthermore, it optimizes global costs based on multi-dimensional dynamic factors such as raw material prices, inventory, and production rhythm, thereby fundamentally eliminating passive high-price substitutions caused by material shortages. This improves the scientific nature of the batching process for technicians and significantly enhances batching efficiency, accuracy, and the overall benefits of the steel plant.

[0099] Figure 2 The flowchart of the grouped rolling optimization algorithm based on steel races in this application is shown. Figure 1 ,like Figure 2 As shown, the algorithm includes the following steps: Step S201: Group each heat batch according to the composite primary key formed by the steel grade and process route to form multiple steel families, and add them to the steel family list.

[0100] All heats are grouped according to the composite primary key formed by steel grade and process route, forming multiple steel families. A steel family list is set up, and the newly formed steel families are added to the steel family list for unified management.

[0101] Step S202: Perform rolling time-domain optimization independently on each steel race in the steel race list to obtain the target batching optimization result for each steel race.

[0102] Each steel race performs rolling time-domain optimization, and different steel races are isolated from each other and do not affect each other. Specifically, each steel race performs sequential solution, inventory rolling, and intra-group backtracking.

[0103] Step S203: Summarize the target batching optimization results for each steel class to form a global cost optimization plan, and determine the steel batching list for each heat based on the global cost optimization plan.

[0104] After rolling time-domain optimization is completed for each steel class in the steel class list, the target batching optimization results for each steel class are used to form a global cost optimization scheme. That is, the global cost optimization scheme contains the target batching optimization results for each steel class. Then, the steel batching list for each heat is determined based on the global cost optimization scheme.

[0105] Figure 3 The flowchart of the grouped rolling optimization algorithm based on steel races in this application is shown. Figure 2 ,like Figure 3 As shown, the algorithm includes the following steps: Step S301: Group each heat batch according to the composite primary key formed by the steel grade and process route to form multiple steel families, and add them to the steel family list.

[0106] Step S302: Sort multiple steel races in the steel race list according to the planned time in the production plan data of each furnace to form a baseline sequence.

[0107] After grouping, the steel races were sorted according to the planned time sequence of the furnaces corresponding to each steel race, forming a baseline sequence S0, thus realizing the baseline sorting.

[0108] Step S303: For the baseline sequence, perform intra-group rolling optimization for each steel race in the baseline sequence to obtain the material optimization results for each steel race, calculate the cost of each steel race and sum them up to obtain the baseline total cost corresponding to the baseline sequence.

[0109] Solve the baseline scheme formed by the baseline sequence S0. Perform intra-group rolling optimization for each steel race according to the sorting order in the baseline sequence, calculate the cost of each steel race, and then sum the costs of all steel races to obtain the baseline total cost C0.

[0110] Building upon the intra-group rolling optimization, we further optimize the execution order among steel races by trying different sorting schemes to find the optimal sequence that minimizes the total global cost.

[0111] Step S304: Adjust the execution order among the various steel races to generate multiple candidate sequences.

[0112] Multiple candidate sequences {S1, S2, ..., S...} can be generated using heuristics and other rules. k}

[0113] Step S305: For each candidate sequence, perform intra-group rolling optimization for each steel race in the candidate sequence to obtain the material optimization results for each steel race, calculate the cost of each steel race and sum them up to obtain the total candidate cost corresponding to the candidate sequence.

[0114] After generating multiple candidate sequences, candidate schemes can be evaluated. For each candidate sequence S, according to that candidate sequence S... i The new sorting order is used to perform intra-group rolling optimization for each steel race, for example, using the updated global inventory. The cost of each steel race is calculated, and then the costs of all steel races are summed to obtain the candidate sequence S. i The corresponding total candidate cost C i .

[0115] Step S306: Based on the baseline total cost and multiple candidate total costs, select the target sequence from the baseline sequence and multiple candidate sequences, and take the batching optimization result of each steel race corresponding to the target sequence as the target batching optimization result of each steel race.

[0116] After solving the baseline and candidate solutions, the optimal sequence selection can be performed.

[0117] C ’ =min{C0, C1, ..., C k} S ’ =argmin{S0, S1, ..., S k} The cost-optimal sequence S ’ As the target sequence, output S ’ and its corresponding solutions, the solutions of which include S ’ The corresponding material optimization results for each steel race.

[0118] Step S307: Summarize the target batching optimization results for each steel class to form a global cost optimization plan, and determine the steel class batching list for each heat based on the global cost optimization plan.

[0119] This application constructs a four-layer collaborative optimization system consisting of constraint variable pre-screening, furnace decomposition, constraint local fine-tuning, and global optimization. It also systematically overcomes long-standing technical challenges in stainless steelmaking batching optimization, such as the large number of material varieties, numerous inventory grids, many furnaces, and large computational load, by employing steel race grouping, rolling time-domain optimization, sequence adjustment, and dynamic backtracking.

[0120] Traditional batching optimization methods often employ one-time global optimization or parallel solution strategies, treating all heats as independent problems and processing them simultaneously, neglecting the temporal dependence of material consumption and dynamic changes in inventory. This application introduces a rolling time-domain optimization mechanism, solving for each heat sequentially according to the production timeline and updating material inventory status in real time. Dynamic inventory awareness is established to ensure the optimization scheme is executable in actual production. Through time-constraint modeling, the actual logic of how material consumption in previous heats affects subsequent heats is accurately reflected. Furthermore, the concept of steel class grouping is proposed, using steel type and process route as composite primary keys for group optimization, achieving collaborative optimization of highly correlated heats within a group. It also achieves an organic combination of local fine-grained optimization and global sequential optimization, achieving the best balance between computational efficiency and optimization quality, effectively overcoming the temporal blind spots of static optimization.

[0121] Step S103: After the steel grade batching list for each heat is approved, the corresponding material inventory is locked, and charging and smelting are carried out according to the steel grade batching list for each heat.

[0122] Figure 4A schematic diagram illustrating the principle of an intelligent batching method for stainless steel smelting based on a multi-heat rolling plan according to an embodiment of this application is shown, as follows: Figure 4 As shown, the daily production plan data for each heat is imported, sorted according to the import order, and grouped and summarized by steel type and process route to determine the number of smelting heats. Then, multiple steel types are formed by combining steel types and process routes, and the model is called for calculation. Based on material inventory data and various process feeding constraints, the allowable materials for each steel type, process definitions, and process output steel composition requirements are determined. The batching model is used to solve the problem, achieving the optimal total cost for multiple heats, multiple steel types, and multiple process routes, and obtaining the steel batching list for each heat.

[0123] If inventory meets requirements, the model outputs complete batching lists for all steel grades. If inventory is insufficient, or there are conflicts with process target composition constraints or steel output constraints, the constraints are appropriately relaxed according to process requirements, and the model is automatically called back to ensure that appropriate batching lists can be output. If inventory shortages are determined based on the steel grade batching lists and material inventory data for each heat, the batching model outputs shortage prediction results and feeds them back to the purchasing system so that the purchasing system can procure the relevant materials. Optionally, a purchasing plan can also be generated based on the shortage prediction results and fed back to the purchasing system.

[0124] The system outputs steel grade batching lists according to the specified format and determines whether they need to be reviewed. If no review is required, no inventory reduction is made. If review is required, all steel grade batching lists are checked to ensure they meet the requirements. If all requirements are met, the review is passed, the corresponding material inventory is locked, inventory occupancy is recorded, and actual charging and smelting are carried out based on the steel grade batching lists for each heat.

[0125] This application establishes pre-solution and AOD (Alternating Dry Oven) loading weight constraint equations based on the constraints of alloy and scrap steel inventory in steel enterprises, the constraints of different steel grades on scrap ratio, alloy and scrap type in production plans, the constraints of scrap steel loading quantity, and the constraints of scrap steel and alloy addition. An objective equation is established with the goal of minimizing the total cost of multiple heats. A linear programming mathematical model is constructed based on the objective equation and used as a batching model. The batching model calculates the addition quantities of various alloys and scrap steel, i.e., the steel grade batching list for each heat. Through solving the model, batching lists for multiple steel grades, multiple process routes, and multiple heats can be output. The output is based on steel grade and process route as the smallest unit, minimizing the number of steel grade batching lists to facilitate physical batching and loading on the production site.

[0126] In addition, this application also designs a closed-loop self-learning feedback mechanism. After the batching scheme is executed, the actual smelting endpoint composition data, cost data and process parameters can be collected in real time. Through machine learning algorithms, the element yield parameters, constraint boundaries and inventory sources in the model are corrected online to achieve continuous self-optimization of the batching model and continuously improve the rationality and economy of multi-heat batching of stainless steel.

[0127] According to the intelligent batching method for stainless steel smelting based on multi-heat rolling planning provided in the embodiments of this application, a batching model is constructed. The batching model takes minimizing the total cost of all materials used in all processes of all heats as the objective function and simultaneously embeds multiple constraints. The batching model is used to solve the problem, outputting the steel grade batching list for each heat. This achieves the automatic generation of steel grade batching lists with high availability, optimal cost, and strong feasibility, realizing refined batching of steel grades for a single heat and transforming from manual calculation to automated real-time optimization. In the solution process, a four-layer collaborative optimization system of constraint variable pre-screening—heat decomposition—constraint local fine-tuning—global optimization is adopted, which can efficiently solve complex batching optimizations involving multiple materials, multiple steel grades, multiple heats, and multiple process routes. A precise material inventory location-based system is also established. The batching mechanism achieves refined management with a single-order, single-policy approach, possessing the ability to instantly reschedule under production disturbances to cope with emergencies and ensure production continuity. It also optimizes global costs based on multi-dimensional dynamic factors such as raw material prices, inventory, and production rhythm, fundamentally preventing passive high-price substitutions due to material shortages. This enhances the scientific nature of the batching process for technical personnel, significantly improving batching efficiency, accuracy, and the overall benefits of the steel plant. The solution realizes intelligent batching for multiple heats of stainless steel based on multi-constraint coupled modeling and an efficient hybrid solution strategy. It effectively solves problems in existing technologies such as high reliance on manual labor, static and coarse batching schemes, inability to dynamically respond to production disturbances, and lack of global cost coordination. It achieves intelligent, precise, and optimized batching processes, significantly reducing the overall cost per ton of steel and ensuring continuous and stable production operation.

[0128] Figure 5 A structural block diagram of an intelligent batching system for stainless steel smelting based on a multi-heat rolling schedule according to an embodiment of this application is shown, as follows: Figure 5 As shown, the system includes: a data acquisition module 510, a data preprocessing module 520, a model processing module 530, and an execution module 540.

[0129] The data acquisition module 510 is suitable for: acquiring production plan data for each furnace within a unit of time, acquiring material inventory data, and various process feeding constraint data.

[0130] The data preprocessing module 520 is suitable for preprocessing production planning data, material inventory data, and various process feeding constraint data before using them as model input data.

[0131] The model processing module 530 is suitable for: calling the batching model, inputting model input data into the batching model, solving the problem using the batching model, and outputting the steel grade batching list for each heat. The batching model, under multiple constraints, solves the problem with the objective function of minimizing the total cost of all materials used in all processes across all heats.

[0132] The execution module 540 is suitable for: locking the corresponding material inventory after the steel grade batching list for each heat is approved, and carrying out charging and smelting according to the steel grade batching list for each heat.

[0133] Optionally, the system also includes a model building module 550, which is adapted to build multiple constraints and construct a batching model with the objective function of minimizing the total cost of all materials used in all processes across all furnaces.

[0134] Optionally, multiple constraints include: production planning constraints, process material balance constraints, process capacity constraints, material input constraints, material inventory constraints, steel grade material restriction constraints, material batch selection constraints, charging efficiency constraints, and composition balance constraints.

[0135] Optionally, the model processing module 530 is further adapted to: for each heat, determine the steel grade to be produced based on the production plan data of the current heat, traverse the material library, remove the material types prohibited in the process specifications of the steel grade, and generate a whitelist of available materials; for each type of material in the whitelist of available materials, sort all inventory batches of that type of material according to the unit price from low to high, and select the inventory batches with the highest preset quantity as decision variables to input into the batching model.

[0136] Optionally, the model processing module 530 is further adapted to: group each heat according to the composite primary key formed by the steel grade and process route to form multiple steel classes, and add them to the steel class list; independently perform rolling time-domain optimization on each steel class in the steel class list to obtain the target batching optimization result for each steel class; summarize the target batching optimization results of each steel class to form a global cost optimization scheme, and determine the steel grade batching list for each heat based on the global cost optimization scheme.

[0137] Optionally, the model processing module 530 is further adapted to: sort multiple steel races in the steel race list according to the planned time in the production plan data of each heat, forming a baseline sequence; for the baseline sequence, perform intra-group rolling optimization on each steel race in the baseline sequence to obtain the batching optimization result of each steel race, calculate the cost of each steel race and accumulate them to obtain the baseline total cost corresponding to the baseline sequence; adjust the execution order between each steel race to generate multiple candidate sequences; for each candidate sequence, perform intra-group rolling optimization on each steel race in the candidate sequence to obtain the batching optimization result of each steel race, calculate the cost of each steel race and accumulate them to obtain the candidate total cost corresponding to the candidate sequence; based on the baseline total cost and the multiple candidate total costs, select a target sequence from the baseline sequence and the multiple candidate sequences, and take the batching optimization result of each steel race corresponding to the target sequence as the target batching optimization result of each steel race.

[0138] The descriptions of the above modules refer to the corresponding descriptions in the method embodiments, and will not be repeated here.

[0139] According to the intelligent batching system for stainless steel smelting based on multi-heat rolling planning provided in this application embodiment, a batching model is constructed. The batching model takes minimizing the total cost of all materials used in all processes across all heats as the objective function and simultaneously embeds multiple constraints. The batching model is used to solve the problem, outputting steel grade batching lists for each heat. This achieves the automatic generation of highly available, cost-optimal, and feasible steel grade batching lists, realizing refined batching of steel grades for a single heat and transforming the process from manual calculation to automated real-time optimization. During the solution process, a four-layer collaborative optimization system is adopted: constraint variable pre-screening—heat decomposition—constraint local fine-tuning—global optimization. This system can efficiently solve complex batching optimizations involving multiple materials, multiple steel grades, multiple heats, and multiple process routes. A precise system based on material inventory location is also established. The batching mechanism achieves refined management with a single-order, single-policy approach, possessing the ability to instantly reschedule under production disturbances to cope with emergencies and ensure production continuity. It also optimizes global costs based on multi-dimensional dynamic factors such as raw material prices, inventory, and production rhythm, fundamentally preventing passive high-price substitutions due to material shortages. This enhances the scientific nature of the batching process for technical personnel, significantly improving batching efficiency, accuracy, and the overall benefits of the steel plant. The solution realizes intelligent batching for multiple heats of stainless steel based on multi-constraint coupled modeling and an efficient hybrid solution strategy. It effectively solves problems in existing technologies such as high reliance on manual labor, static and coarse batching schemes, inability to dynamically respond to production disturbances, and lack of global cost coordination. It achieves intelligent, precise, and optimized batching processes, significantly reducing the overall cost per ton of steel and ensuring continuous and stable production operation.

[0140] This application provides a non-volatile computer storage medium storing at least one executable instruction or computer program that enables a processor to perform the operation corresponding to the intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan in any of the above method embodiments.

[0141] This application provides a computer program product, which includes at least one executable instruction or computer program that enables a processor to perform the operation corresponding to the intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan in any of the above method embodiments.

[0142] Figure 6 The diagram shows a structural schematic of a computing device according to one embodiment of the present application. The specific embodiments of the present application do not limit the specific implementation of the computing device.

[0143] like Figure 6 As shown, the computing device may include: a processor 602, a communications interface 604, a memory 606, and a communications bus 608.

[0144] The processor 602, communication interface 604, and memory 606 communicate with each other via communication bus 608. Communication interface 604 is used to communicate with other network elements such as clients or other servers. Processor 602 executes program 610, specifically performing the relevant steps in the above-described embodiment of the intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan for computing devices.

[0145] Specifically, program 610 may include program code that includes computer operation instructions.

[0146] The processor 602 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. The computing device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0147] Memory 606 is used to store program 610. Memory 606 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0148] Specifically, program 610 can be used to cause processor 602 to execute the intelligent batching method for stainless steel smelting based on a multi-heat rolling plan in any of the above method embodiments. The specific implementation of each step in program 610 can be found in the corresponding descriptions of the steps and units in the above embodiments of intelligent batching for stainless steel smelting based on a multi-heat rolling plan, and will not be repeated here. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the equipment and modules described above can be referred to the corresponding process descriptions in the foregoing method embodiments, and will not be repeated here.

[0149] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, the embodiments of this application are not directed to any particular programming language. It should be understood that the contents of the embodiments of this application described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best implementation of the embodiments of this application.

[0150] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0151] Similarly, it should be understood that, in order to simplify this disclosure and aid in understanding one or more of the various inventive aspects, in the foregoing description of exemplary embodiments of the present application, various features of the present application embodiments are sometimes grouped together into a single embodiment, figure, or description thereof. However, this approach to disclosure should not be construed as reflecting an intention that the claimed embodiments of the present application require more features than expressly recited in each claim. Rather, as reflected in the following claims, inventive aspects lie in fewer than all features of a single foregoing disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the present application.

[0152] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0153] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are meant to be within the scope of the embodiments of this application and form different embodiments. For example, in the following claims, any one of the claimed embodiments can be used in any combination.

[0154] The various component embodiments of this application can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components according to the embodiments of this application. The embodiments of this application can also be implemented as device or apparatus programs (e.g., computer programs and computer program products) for performing part or all of the methods described herein. Such programs implementing the embodiments of this application can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0155] It should be noted that the above embodiments are illustrative of the embodiments of this application and not limiting of the embodiments of this application, and those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. Embodiments of this application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

Claims

1. A smart batching method for stainless steel smelting based on multi-heat rolling planning, characterized in that, include: The production plan data for each furnace within a unit time period is obtained, as well as the material inventory data and various process feeding constraint data. The production plan data, the material inventory data, and the various process feeding constraint data are preprocessed and then used as model input data. The batching model is invoked, and the input data of the model is input into the batching model. The batching model is then used to solve the problem and output the steel batching list for each heat. The batching model is solved under multiple constraints with the objective function of minimizing the total cost of all materials used in all processes of all heats. Once the steel batching list for each heat is approved, the corresponding material inventory is locked, and charging and smelting are carried out according to the steel batching list for each heat.

2. The method according to claim 1, characterized in that, Before invoking the ingredient model, the method further includes: Multiple constraints are constructed, and the total cost of all materials used in all processes across all furnaces is minimized as the objective function to obtain the batching model.

3. The method according to claim 1, characterized in that, Multiple constraints include: production planning constraints, process material balance constraints, process capacity constraints, material input constraints, material inventory constraints, steel grade material restriction constraints, material batch selection constraints, charging efficiency constraints, and composition balance constraints.

4. The method according to claim 1, characterized in that, The step of inputting the model input data into the ingredient model further includes: For each heat, the steel grade to be produced is determined based on the production plan data of the current heat. The material library is traversed, and material types that are prohibited in the process specifications of the steel grade are removed to generate a whitelist of available materials. For each type of material in the available material whitelist, all inventory batches of that type of material are sorted from low to high according to unit price, and the inventory batches with the highest preset quantity are selected as decision variables and input into the ingredient distribution model.

5. The method according to claim 1, characterized in that, The step of using the batching model to solve the problem and outputting the steel batching list for each heat further includes: Each heat cycle is grouped according to the composite primary key formed by steel grade and process route to form multiple steel families, which are then added to the steel family list. Perform rolling time-domain optimization independently on each steel race in the steel race list to obtain the target batching optimization result for each steel race; The target batching optimization results of each steel class are summarized to form a global cost optimization plan, and the steel class batching list for each heat is determined based on the global cost optimization plan.

6. The method according to claim 5, characterized in that, The step of independently performing rolling time-domain optimization on each steel race in the steel race list to obtain the target batching optimization result for each steel race further includes: According to the planned time in the production plan data of each heat, the multiple steel races in the steel race list are sorted to form a baseline sequence; For the benchmark sequence, intra-group rolling optimization is performed on each steel race in the benchmark sequence to obtain the batching optimization result of each steel race. The cost of each steel race is calculated and accumulated to obtain the benchmark total cost corresponding to the benchmark sequence. Adjust the execution order among the various steel races to generate multiple candidate sequences; For each candidate sequence, perform intra-group rolling optimization for each steel race in the candidate sequence to obtain the batching optimization results for each steel race. Calculate the cost of each steel race and sum them up to obtain the total candidate cost corresponding to the candidate sequence. Based on the baseline total cost and multiple candidate total costs, a target sequence is selected from the baseline sequence and multiple candidate sequences, and the batching optimization result of each steel race corresponding to the target sequence is taken as the target batching optimization result of each steel race.

7. A smart batching system for stainless steel smelting based on multi-furnace rolling planning, characterized in that, include: The data acquisition module is suitable for acquiring production plan data for each furnace within a unit of time, acquiring material inventory data, and various process feeding constraint data; The data preprocessing module is suitable for preprocessing the production plan data, the material inventory data, and various process material feeding constraint data before using them as model input data; The model processing module is adapted to call the batching model, input the model input data into the batching model, solve the problem using the batching model, and output the steel grade batching list for each heat. The batching model is solved under multiple constraints with the objective function of minimizing the total cost of all materials used in all processes of all heats. The execution module is suitable for locking the corresponding material inventory after the steel grade batching list for each heat is approved, and for charging and smelting according to the steel grade batching list for each heat.

8. A computing device, comprising: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the operation corresponding to the intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan as described in any one of claims 1-6.

9. A computer storage medium storing at least one executable instruction that causes a processor to perform an operation corresponding to the intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan as described in any one of claims 1-6.

10. A computer program product comprising at least one executable instruction that causes a processor to perform an operation corresponding to the intelligent batching method for stainless steel smelting based on a multi-furnace rolling plan as described in any one of claims 1-6.

Images