Molten iron distribution method and control system thereof
Through the improved quadratic planning model and hybrid optimization algorithm, the control problem of traditional molten iron distribution methods in complex environments is solved, intelligent and automated management of molten iron distribution is realized, and production efficiency and product quality are improved.
Patent Information
- Application Number
- CN202410921811.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-10
- Publication Date
- 2024-11-08
AI Technical Summary
Traditional molten iron distribution methods are difficult to achieve efficient and precise control in complex production environments, resulting in a decline in the quality and production efficiency of steel products. The existing optimization algorithms have problems such as high algorithm complexity, poor real-time performance and poor adaptability.
The improved quadratic planning model is used to combine a hybrid Archimedes optimization algorithm and a finite memory quasi-Newtonian method to realize intelligent decision-making and automatic adjustment by real-time monitoring of molten iron, providing an accurate and stable molten iron distribution solution.
It improves the accuracy of molten iron distribution and the stability of the system, reduces operating costs, adapts to different production scales and process requirements, and reduces the demand for manual intervention and computing resources.
Smart Images

Figure CN118915636A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of steel manufacturing industry control, and in particular relates to a molten iron distribution method and a control system thereof. Technical Background
[0002] In the steelmaking process, molten iron distribution is a crucial link, which directly affects the quality, production efficiency and cost control of steel products. Traditional molten iron distribution methods often rely on manual experience or simple control algorithms. These methods are often difficult to achieve efficient and accurate control in the face of complex production environments and changing demands, resulting in uneven molten iron distribution, affecting the overall quality and production efficiency of steel products.
[0003] With the development of industrial automation and intelligent manufacturing, the traditional molten iron distribution control method can no longer meet the needs of modern steel production. Therefore, researching and developing an intelligent control system for molten iron distribution based on advanced optimization algorithms has become an urgent need to improve steel production efficiency and product quality. By introducing optimization algorithms and intelligent control technologies, real-time monitoring and intelligent decision-making of the molten iron distribution process can be achieved, thereby improving the accuracy and reliability of molten iron distribution, reducing manual intervention, improving production efficiency, and reducing energy consumption and costs.
[0004] At present, there have been some studies and applications that attempt to use optimization algorithms and other technologies to optimize molten iron distribution, but these methods often have problems such as high algorithm complexity, poor real-time performance, and weak adaptability in practical applications; for example, the document Xuan H, Tang L. Scheduling a hybrid flow shop with batch production at the last stage [J]. Computers & Operations Research, 2007, 34 (9): 2718-2733. discloses that the main process of steel production is regarded as a hybrid flow shop scheduling problem (HFS) containing multiple constraints, and an integer programming model is established using specific actual production constraints, and a Lagrangian relaxation algorithm based on batch decoupling is proposed for this problem. In this algorithm, after relaxing the capacity constraints by Lagrangian multipliers, the relaxed problem is decomposed into sub-problems, and the sub-problems are solved by using dynamic programming; because it establishes an integer programming model, it means that as the scale of the problem increases, the solution time may increase exponentially, and large-scale integer programming problems are very time-consuming to solve, and the requirements for computing resources are also high. Therefore, developing a molten iron distribution system that combines the latest optimization algorithms and intelligent control technologies, aiming to address the limitations of existing methods and improve the intelligence level and control performance of the molten iron distribution system, is one of the key technologies for improving the automation and intelligence level of steel production. Summary of the invention
[0005] The purpose of the present invention is to provide a molten iron distribution method and a control system thereof to solve the problem of inaccurate control of liquid metal distribution in the steel production industry. The molten iron distribution process described in this application refers to the molten iron allocation subject. On the premise that the iron output of the iron plant has been estimated and the future consumption plan of the steel plant has been determined, the amount of molten iron to be distributed is determined by monitoring the amount of circulating molten iron in all areas of the current system. All areas in the current system include the iron plant area, the steel plant area, and the in-transit area; molten iron is distributed according to the steel mill's steelmaking water demand combined with the actual iron output of different iron plants.
[0006] The present invention provides a molten iron distribution method, which improves the existing linear programming model and establishes a more reasonable quadratic programming model for molten iron distribution, thereby realizing intelligent decision-making, real-time monitoring and automatic adjustment of the molten iron distribution process. The system will combine sensor technology to monitor the molten iron volume of the system in real time, calculate the molten iron distribution plan through an intelligent optimization algorithm, thereby providing an accurate and stable molten iron distribution plan, improving the accuracy of molten iron distribution and reducing the operating costs of the enterprise.
[0007] In a first aspect, the present invention provides a method for distributing molten iron, such as Figure 1As shown, the molten iron distribution method comprises the steps of:
[0008] Step S1, according to the molten iron planning scenario of the present application, collect resource parameters, planning parameters, and priority information to establish a secondary planning model, wherein the resource parameters include the maximum, minimum safety, and optimal values of the molten iron circulation volume, the planning parameters refer to the planned iron output of the iron plant, the planned furnace loading volume, the number of furnaces, and the planned scrap preheating amount of the steel plant, and the priority information refers to the production priority of each steel plant;
[0009] The objective function of the quadratic programming model may be a quadratic function, and the constraint of the quadratic programming model may be a linear function; the objective function of the quadratic programming model includes a first objective function, a second objective function and a third objective function; the first objective function is an objective function for minimizing the transportation time of molten iron from the blast furnace to each converter; the second objective function is an objective function for minimizing the difference between the molten iron volume of the steel plant system and the molten iron volume of the optimal system of the steel plant; the third objective function is an objective function for minimizing the difference between the molten iron circulation volume of the blast furnace system and the molten iron circulation volume of the optimal system of the blast furnace; the constraint of the quadratic programming model includes a first constraint and a second constraint, the first constraint is a capacity constraint of the blast furnace, and the second constraint is a maximum / minimum safety stock constraint of the system molten iron circulation volume;
[0010] The first objective function is to minimize the transportation time of molten iron from blast furnace to each converter, that is, to minimize the transportation time of molten iron from each blast furnace to each converter / electric furnace. In other words, within the planning period, the sum of the transportation time of each blast furnace to transport molten iron to each steelmaking plant in each sub-period should be minimized, where the transportation time is represented by the transportation priority, that is,
[0011]
[0012] The second objective function is to minimize the difference between the amount of hot metal in the steel plant system and the optimal amount of hot metal in the steel plant system. That is, within the planning period, the sum of the amount of hot metal received by each steel plant and the amount of hot metal circulating in the steel plant system at the beginning of the plan, minus the amount of hot metal consumed by the steel plant, the smaller the difference between the final value and the optimal amount of hot metal circulating in the steel plant system, the better.
[0013]
[0014] The third objective function is to minimize the difference between the molten iron circulation volume of the blast furnace system and the molten iron circulation volume of the optimal blast furnace system. That is to say, within the planning period, the sum of the production capacity of each blast furnace, the molten iron circulation volume of the blast furnace system at the beginning of the plan, and the planned amount of scrap steel added, minus the amount of molten iron allocated to the steel plant, the smaller the difference between the final value and the molten iron circulation volume of the optimal blast furnace system, the better.
[0015]
[0016] The constraint conditions of the multi-objective quadratic programming model include a first constraint condition and a second constraint condition;
[0017] The first constraint is the capacity constraint of the blast furnace, which means that when performing iron and steel balance, the amount of molten iron produced and delivered by each blast furnace in each sub-period within the planning cycle cannot exceed the sum of the maximum production capacity of the blast furnace in the sub-period, the original molten iron circulation volume of the blast furnace system, and the amount of scrap steel planned to be added on the day.
[0018]
[0019] The second constraint is the maximum / minimum safety inventory constraint of the system's molten iron circulation volume. Since blast furnace production is carried out continuously, for the sake of production safety and stability, and in order to prevent emergencies such as equipment failure, a certain safety inventory (i.e., the system's molten iron circulation volume) should be maintained to ensure safe and stable production in the steelmaking area. The system's molten iron circulation volume has a maximum and minimum safety range. By controlling the weight of the system's molten iron so that it is within the maximum / minimum safety inventory range, it can be avoided that insufficient molten iron supply to the blast furnace causes the subsequent steelmaking to be unable to produce normally, or excessive molten iron supply causes too many heavy molten iron tanks and fewer empty tanks, affecting the safe iron tapping of the blast furnace and causing dangerous situations. The system's molten iron weight cannot exceed the specified maximum / minimum safety inventory, that is,
[0020]
[0021]
[0022] The meanings of the symbols in the objective function and constraint conditions of the multi-objective quadratic programming model are shown in Table 1 below:
[0023] Table 1
[0024] symbol illustrate m Indicates the number of blast furnaces n Indicates the number of steel mills q represents the qth steel plant, q={1,2,…,n} p represents the pth blast furnace, p = {1, 2, ..., m} Z The amount of hot metal circulating in the system of the qth steel plant at the beginning of the plan, in tons B Molten iron circulation volume of the pth blast furnace system BfP The daily production capacity of the pth blast furnace LqCy Planned production furnace number of the qth steel plant R Planned iron loading per furnace for the Qth steel plant LG Planned consumption of hot metal capacity of the qth steel plant HkD Optimal system molten iron circulation value for the qth steel plant Cp The priority of the pth blast furnace to the qth steel plant Mq Maximum system molten iron circulation capacity of the qth steel plant Nq Minimum system molten iron circulation volume of the qth steel plant FWf Planned amount of scrap steel added to the pth blast furnace Wlq The amount of hot metal allocated from the pth blast furnace to the qth steel plant during the planning period
[0025] It should be noted that one ton of scrap steel described in the present invention can only be converted into 0.8 to 0.85 tons of molten iron; and the scrap steel is not added to the blast furnace, but to the molten iron tank. However, after adding the scrap steel, the molten iron tank will still go to the specific blast furnace to receive molten iron, so the model classifies the molten iron produced by scrap steel into the blast furnace.
[0026] Step S2, collecting real-time information on molten iron distribution, the real-time information on molten iron distribution includes the system molten iron circulation volume in each blast furnace area of each iron plant, each steel plant area and in-transit area; the system molten iron circulation volume refers to the molten iron tank from the time it starts to receive iron under the blast furnace until the molten iron tank reaches the steel plant backfilling station or ladle. During this process, the molten iron is carried by the molten iron tank. At a certain time in the plan, the amount of molten iron in the molten iron tank is the system molten iron circulation volume at that time node;
[0027] Step S3, inputting the real-time information of the molten iron distribution obtained in step S2 into the quadratic programming model established in step S1, and using a hybrid AOA algorithm to solve the quadratic programming model and output a molten iron distribution plan.
[0028] Among them, the hybrid AOA is a hybrid AOA optimization algorithm based on the Archimedes Optimization Algorithm (AOA) and the limited-memory quasi-Newton method with bounded constraints (Limited-memory Broyden-Fletcher-Goldfarb-Shanno with Bounds, L-BFGS-B) provided in this application; the algorithm finds the current optimal initial point after a global search of AOA, and the current optimal initial point corresponds to the initial output, that is, the initial molten iron distribution plan, but it is a hypothetical solution. The current optimal initial point is used to input the L-BFGS-B algorithm for solution, and the final molten iron distribution plan is output; it should be noted that the process of using the current optimal initial point to input the L-BFGS-B algorithm for solution is to gradually approximate the true solution with the hypothetical solution determined by AOA, calculate the gradient of the objective function with the hypothetical solution, and use a limited memory method to approximate the Hessian matrix, and find the true optimal solution by constantly adjusting the position. The solution process flow chart of the hybrid AOA is as follows: Figure 2 The specific steps are as follows:
[0029] Step 1: Initialize algorithm parameters, including population size, maximum number of iterations, and C1 to C4. The population is a number of randomly generated initial points, that is, n solutions are assumed. For example, assuming that the population size is 50, 50 assumed solutions are randomly generated; the maximum number of iterations is 100, and C1 to C4 are 2, 6, 2, and 0.5 respectively;
[0030] Step 2: Calculate the fitness of each individual in the population and record the optimal individual position x best , optimal density den best , optimal volume vol best , optimal acceleration acc bestIt should be noted that the individual refers to a specific point in the population, corresponding to a specific hypothetical solution in the population in Step 1;
[0031] Step 3: Update density and volume according to equations (7) and (8);
[0032]
[0033] in, and is the density of the i-th and i+1-th individuals in generation t, and is the volume of the i-th and i+1-th individuals in the t-th generation.
[0034] Step 4: Update the transfer operator and density factor according to formulas (9) and (10);
[0035]
[0036] Among them, t and t max They represent the current number of selections and the maximum number of iterations respectively. If TF≤0.4, the hybrid AOA enters the global exploration phase; otherwise, it enters the local development phase; d is the density factor.
[0037] Step 5: If the transfer factor TF ≤ 0.4, use (11) to update the acceleration and (12) to update the individual position;
[0038]
[0039] in, and is the position of the i-th individual in the t+1 generation and the t generation, x rand is the position of the random individual in the tth generation, a random number between rand∈(0,1).
[0040] Step 6: If the transfer factor TF>0.4, first use the current optimal individual position as the initial point of the L-BFGS-B algorithm to solve. If the solution is successful, output the optimal result; otherwise, use (13) to update the acceleration and (14) to update the individual position;
[0041]
[0042] Among them, acc best is the acceleration of the current optimal individual. Where C2 is a fixed constant; T = TF*C3, and T∈[C3*0.3, 1]; F is a direction factor used to determine the iterative position update direction, which is defined as (15):
[0043]
[0044] Step 7: Determine whether the iteration stop condition is met. If so, exit and output the optimal solution, that is, the allocation plan of this application, such as the allocation plan shown in Table 2. Otherwise, repeat Step 2 to Step 7.
[0045] In a second aspect, the present invention provides a molten iron distribution control system, including a data acquisition module, an optimization algorithm module, a control execution module, a human-computer interaction module, and a data storage module.
[0046] The data acquisition module is used for data acquisition, including the collection of the above-mentioned resource parameters, planning parameters, priority information, and real-time data including the collection of system molten iron circulation volume in each blast furnace area of each iron plant, each steel plant area, and in-transit area;
[0047] The optimization algorithm module is used to solve the best allocation scheme using the above hybrid AOA;
[0048] The control execution module is used to establish a secondary planning model based on the molten iron planning scenario, resource parameters, planning parameters, and priorities of this application, and set the optimal configuration of molten iron resource parameters according to the operating status of the steel plant converter equipment on that day, including the optimal value and maximum and minimum safety value settings of molten iron in the iron plant and steel plant system and the priority of molten iron supply in each steel plant. The allocation plan generated by the system display is determined to be executed by this module and sent to various departments. The plan confirmed to be executed by this module will be stored in the data storage module.
[0049] The human-computer interaction module includes detailed plan viewing and plan parameter modification. Detailed plan viewing means that after allocation, the specific blast furnace to converter allocation plan, as well as the planned molten iron source details and steel plant consumption can be viewed in detail; plan parameter modification means that when abnormal conditions occur, such as converter equipment failure, poor blast furnace conditions, and reduced iron output, the plan parameters can be modified to re-call the algorithm to calculate the allocation plan; through the human-computer interaction module, operators can view and monitor all relevant input data for molten iron allocation in real time. In addition, the system can also display a detailed molten iron allocation plan, including the amount of molten iron allocated from different blast furnaces to different iron plants. During the molten iron allocation process, if any abnormal conditions are encountered or the allocation plan needs to be adjusted according to the actual production situation, the operator can make necessary manual interventions through the interactive module to ensure smooth production.
[0050] The data storage module is used to store the allocation plan, as well as the ironworks iron production plan and steel mill consumption plan used in the allocation plan. The data storage module is one of the core parts of the entire system, responsible for the reliable storage of all collected data and generated allocation plans. This module ensures the integrity and security of the data so that accurate information can be provided for subsequent queries and analysis. The data storage module uses a relational database (MySQL) system. Through effective data storage and management, a solid foundation can be provided for decision support.
[0051] In a third aspect, the present invention provides a computer device / equipment, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above-mentioned molten iron distribution method.
[0052] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon, wherein the computer program / instruction, when executed by a processor, implements the steps of the above-mentioned molten iron distribution method.
[0053] In a fifth aspect, the present invention provides a computer program product, comprising a computer program / instruction, characterized in that the computer program / instruction implements the steps of the above-mentioned molten iron distribution method when executed by a processor.
[0054] Beneficial effects: The present invention proposes a molten iron distribution method and a control system thereof, aiming to solve the molten iron distribution problem of steel enterprises and realize the automation and intelligent management of the molten iron distribution process.
[0055] This method has the following main advantages:
[0056] The quadratic programming model established in the molten iron distribution method provided by the present invention is not required to be an integer solution, which has greater advantages in terms of computational efficiency, modeling simplicity and solution stability; while the integer linear programming established in the document Xuan H, Tang L. Scheduling a hybrid flow shop with batch production at the last stage [J]. Computers & Operations Research, 2007, 34 (9): 2718-2733 faces high computational complexity, and the integer solution required is very sensitive to input data and parameters, and a small change may significantly differ from the optimal solution, and the stability is poor. Therefore, the algorithm proposed in this application can effectively reduce the running time cost and improve the stability of the solution due to its low computational complexity and stable optimal solution.
[0057] The present invention provides a molten iron distribution control system, which collects data in real time, adjusts the distribution plan in real time, adapts to changes in production demand, and can be expanded according to actual needs to adapt to different production scales and process requirements. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] Figure 1 Schematic diagram of the process of molten iron distribution method;
[0059] Figure 2 Hybrid AOA algorithm flow chart;
[0060] Figure 3 A schematic diagram of the structure of the molten iron distribution control system of the present invention;
[0061] Figure 4 It is an iterative process of solving a quadratic programming model using a hybrid AOA algorithm in a specific implementation manner;
[0062] Figure 5 It is an iterative process of solving a quadratic programming model using the Powell algorithm in a specific implementation method;
[0063] Figure 6 This is an iterative process of solving a quadratic programming model using the SLSQP algorithm in a specific implementation manner. DETAILED DESCRIPTION
[0064] Combine the following Figure 3 The present invention is further described in detail by taking specific usage scenarios as examples, but is not intended to limit the scope of the present invention.
[0065] Taking a steel enterprise in Nanjing as an example, an intelligent control system and method for molten iron distribution of the present invention is used to determine the distribution plan from the blast furnace of the iron plant to the converter of the steel plant. The enterprise has two iron plants, one with three blast furnaces, the second with two blast furnaces, three steel plants, and one iron casting station. The maximum production capacity of the five blast furnaces in the iron plant is 5800, 7400, 5800, 5600, and 5600 tons / day respectively; the three steel plants have a total of 7 converters.
[0066] The following is a detailed process for the intelligent control system and method for molten iron distribution:
[0067] Step S1: Data collection, collect the system molten iron volume of the iron plant, steel plant and the current moment through the existing steel smart operation system interface of the enterprise; when the system is operating normally, the data collection module also needs to obtain the executed plan and real-time system molten iron volume from the smart operation system at regular intervals. The real-time molten iron circulation volume of the system is obtained by the sensors on the molten iron tank of the locomotive and the iron tank of the iron plant to obtain the weight of the molten iron in the molten iron tank. For example, the data of the steel plant on a certain day in August 2023 are as follows: the estimated production capacities of the five blast furnaces are 5800, 7400, 5800, 5600 and 5600 tons respectively; the planned production numbers of the seven converters produced by the three steel plants are 36, 36, 36, 43, 43, 43 and 23 respectively, and the furnace input amounts are 152.24, 152.24, 152.24, 103.1, 103.1, 103.1 and 63 respectively; the current molten iron circulation volume of each regional system is 200 tons for No. 1 blast furnace, 300 tons for No. 2 blast furnace, 200 tons for No. 3 blast furnace, 200 tons for No. 4 blast furnace, 100 tons for No. 5 blast furnace; the molten iron of the seven converter systems of the steel plant are 500 tons, 500 tons, 500 tons, 300 tons, 300 tons, 300 tons and 200 tons respectively; 0 tons in the in-transit area;
[0068] Step S2: Present the data collected in step S1 in the intelligent control system for molten iron distribution, and then the operator confirms whether the acquired data is prepared correctly and confirms the data, and then the operator sets the resource parameters. For example, the optimal values of the system molten iron for the three steel mills are set to 1050, 750, and 100 respectively, and the optimal values under the blast furnace are set to 200, 300, 200, 200, and 200 respectively. The maximum safety of the three steel mills is 1800, 1200, and 400 respectively, and the minimum safety is 900, 600, and 100.
[0069] Step S3: Based on the data obtained in step S1 and the resource allocation parameters set in step S2, the priorities of the steel mills are set. For example, even if the priorities of the three steel mills are the same, the algorithm system will output the allocation plan for the day. The iterative process of the model solution is as follows: Figure 4 As shown, the final allocation plan is shown in Table 2, and the amount of molten iron in each blast furnace and steel plant system is shown in Table 3, in tons.
[0070] Table 2 Allocation plan
[0071] Blast Furnace No. 1 No. 2 Blast Furnace No. 3 Blast Furnace No. 4 Blast Furnace No. 5 Blast Furnace A steel plant 4300 7498 4479 0 0 Second Steel Plant 1549 0 0 5651 5652 Three Steel Plants 0 0 1370 0 0
[0072] Table 3 System molten iron quantity after planning
[0073] 1 Blast Furnace 2 Blast Furnace 3 Blast Furnace 4 Blast Furnace 5 Blast Furnace 1Steel Plant 2Steel Plant 3 Steel Plant 151 201 151 149 149 1335 660 120
[0074] Step S4: The distribution data calculated according to step S3 can serve as a foundation for subsequent molten iron transportation. If the operator has questions about the distribution plan or does not understand something, he can check the detailed plan for molten iron distribution. For example, the detailed description of Table 2 above is shown in Table 4 below:
[0075] Table 4 Detailed allocation plan
[0076] 1 Converter 2 Converter 3 Converter 4 Converter 5 Converter 6 Converter 7 Converter 1 Blast Furnace 1123 1446 1731 577 730 242 0 2 Blast Furnace 2899 2479 2120 0 0 0 0 3 Blast Furnace 1404 1500 1575 0 0 0 1370 4 Blast Furnace 0 0 0 2051 965 2635 0 5. Blast Furnace 0 0 0 1656 2589 1407 0
[0077] In order to verify the efficiency of the hybrid AOA algorithm, the Powell algorithm and the SLSQP algorithm are used to solve the molten iron distribution optimization model, namely the quadratic programming model, respectively, and the solution iterative process is obtained as follows: Figure 5 and Figure 6 As shown; Among them, Powell algorithm is a numerical optimization algorithm for unconstrained optimization problems. The algorithm was proposed by Michael JDPowell in 1964. The calculation process can be found in "An efficient method for finding the minimum of a function ofseveral variables without calculating derivatives." The Computer Journal, 7 (2), 155-162. The content recorded in the literature; It is an iterative method that finds the optimal solution by performing a one-dimensional line search at each step; The calculation process of SLSQP algorithm can be found in Kraft, Dieter. "A software package for sequentialquadratic programming." Forschungsbericht-Deutsche Forschungs-undVersuchsanstalt für Luft-und Raumfahrt (1988). The content recorded in the literature; SLSQP is a sequential quadratic programming method used to solve nonlinear optimization problems with equality and inequality constraints. The characteristic of this algorithm is that in each iteration step, it gradually approaches the optimal solution by solving a quadratic programming subproblem. In this experiment, the SLSQP algorithm was used without using Lagrange multipliers. The constraints were incorporated into the objective function, so that the performance of the three algorithms could be tested from multiple angles. The three algorithms were run independently 10 times to obtain the running time, number of iterations, and stability of each algorithm, as shown in Table 5:
[0078] Table 5 Algorithm comparison
[0079] Powell SLSQP Mixed AOA Average running time 1.844 0.29 0.215 Optimal value stability 32696.11349 0.070265307 0.00271693 Average number of iterations 26 26.8 57.2
[0080] When the three algorithms are compared in terms of performance in three aspects, in terms of the average number of iterations, due to the small iteration step size of the hybrid AOA algorithm in the later stage, it takes multiple iterations to finally reach the convergence condition. Therefore, the average number of iterations of the hybrid AOA algorithm is much different from that of the Powell algorithm and the SLSQP algorithm. The running time of the SLSQP algorithm also has a good effect, but because its initial point is randomly generated, the experimental process will not get the optimal solution after the algorithm runs, indicating that the algorithm fails to run. If the initial point is selected appropriately, the algorithm will have a good running effect. The results obtained using the hybrid AOA method are quite superior in terms of average running time and algorithm stability. Especially in terms of stability, if the actual production of a steel plant is taken as the standard, its accuracy only needs to be accurate to the single digit. If the algorithm is judged on the accuracy of its convergence condition, the hybrid AOA algorithm will greatly reduce the number of iterations, and the running time will also be greatly improved. Among them, stability refers to the difference between the maximum and minimum optimal values obtained by the algorithm after 10 tests.
[0081] Step S5: The allocation plan and detailed description calculated through the above steps are stored in the database and can be viewed by the user at any time. In this step, all relevant data of the allocation plan, including the detailed description of each task, are systematically organized and recorded. These data will first be converted into a format suitable for database storage to ensure that they are structured and easy to query. Then, using a specially designed database interface, the data will be uploaded to the database server. After storage is completed, users can access this information through the front-end interface. The system will provide a friendly query function, allowing users to filter and view the details of the allocation plan according to different conditions.
Claims
1. A method for distributing molten iron, characterized in that: The method comprises the steps of: Step S1: Acquire resource parameters, planning parameters, and priority information to establish a quadratic programming model, wherein the resource parameters include the maximum, minimum safe, and optimal values of the molten iron circulation volume, the planning parameters refer to the planned iron output of the iron plant, the planned furnace loading volume, the number of furnaces, and the planned scrap preheating amount of the steel plant, and the priority information refers to the production priority of each steel plant; Step S2, collecting real-time information on molten iron distribution, the real-time information on molten iron distribution includes the system molten iron circulation volume in each blast furnace area of each iron plant, each steel plant area and in-transit area; the system molten iron circulation volume refers to the molten iron tank from the time it starts to receive iron under the blast furnace until the molten iron tank reaches the steel plant backfilling station or ladle. During this process, the molten iron is carried by the molten iron tank. At a certain time in the plan, the amount of molten iron in the molten iron tank is the system molten iron circulation volume at that time node; Step S3, input the real-time information of the molten iron distribution obtained in step S2 into the quadratic programming model established in step S1, and use a hybrid AOA algorithm to solve the quadratic programming model, and output a molten iron distribution plan; wherein the hybrid AOA is a hybrid AOA optimization algorithm based on the Archimedean optimization algorithm and the limited memory quasi-Newton method with bounded constraints; the algorithm finds the current optimal initial point after an AOA global search, and the current optimal initial point corresponds to the initial output, i.e., the initial molten iron distribution plan, but it is a hypothetical solution, and the current optimal initial point is used to input the L-BFGS-B algorithm for solution, and the final molten iron distribution plan is output.
2. A molten iron distribution method as claimed in claim 1, characterized in that: The steps of solving the mixed AOA include: Step 1: Initialize algorithm parameters, including population size, maximum number of iterations, and C1 to C4. The population is a number of randomly generated initial points, that is, n solutions are assumed. For example, assuming that the population size is 50, 50 assumed solutions are randomly generated; the maximum number of iterations is 100, and C1 to C4 are 2, 6, 2, and 0.5 respectively; Step 2: Calculate the fitness of each individual in the population and record the optimal individual position x best , optimal density den best , optimal volume vol best , optimal acceleration acc best It should be noted that the individual refers to a specific point in the population, corresponding to a specific hypothetical solution in the population in Step 1; Step 3: Update density and volume according to equations (7) and (8); in, and is the density of the i-th and i+1-th individuals in generation t, and is the volume of the i-th and i+1-th individuals in the t-th generation; Step 4: Update the transfer operator and density factor according to formulas (9) and (10); Among them, t and t max They represent the current number of generations and the maximum number of iterations respectively; if TF≤0.4, the hybrid AOA enters the global exploration stage; otherwise, it enters the local development stage; d is the density factor; Step 5: If the transfer factor TF ≤ 0.4, use (11) to update the acceleration and (12) to update the individual position; in, and is the position of the i-th individual in the t+1 generation and the t generation, x rand is the position of the random individual in the tth generation, a random number between rand∈(0,1); Step 6: If the transfer factor TF>0.4, first use the current optimal individual position as the initial point of the L-BFGS-B algorithm to solve. If the solution is successful, output the optimal result; otherwise, use (13) to update the acceleration and (14) to update the individual position; Among them, acc best is the acceleration of the current optimal individual; C2 is a fixed constant; T = TF*C3, and T∈[C3*0.3, 1]; F is a direction factor used to determine the iterative position update direction, which is defined as (15): Step 7: Determine whether the iteration stop condition is met. If so, exit and output the optimal solution, i.e., the molten iron distribution plan, such as the distribution plan shown in Table 2. Otherwise, repeat Step 2 to Step 7.
3. A molten iron distribution method as claimed in claim 1, characterized in that: The objective functions of the quadratic programming model include a first objective function, a second objective function and a third objective function; the first objective function is an objective function for minimizing the transportation time of molten iron from the blast furnace to each converter; the second objective function is an objective function for minimizing the difference between the molten iron volume of the steel plant system and the optimal molten iron volume of the steel plant system; the third objective function is an objective function for minimizing the difference between the molten iron circulation volume of the blast furnace system and the optimal molten iron circulation volume of the blast furnace system; the constraints of the quadratic programming model include a first constraint and a second constraint, the first constraint is the production capacity constraint of the blast furnace, and the second constraint is the maximum / minimum safety stock constraint of the system molten iron circulation volume.
4. A molten iron distribution method as claimed in claim 3, characterized in that: The first objective function is shown as follows: Among them, m represents the number of blast furnaces; n represents the number of steel mills; q represents the qth steel mill, q = {1, 2, ..., n}; p represents the pth blast furnace, p = {1, 2, ..., m}; Cpq represents the priority of the pth blast furnace in allocating the qth steel mill; Wpq represents the amount of molten iron allocated from the pth blast furnace to the qth steel mill during the planning period.
5. A molten iron distribution method as claimed in claim 4, characterized in that: The second objective function is as follows: Among them, Zq represents the system molten iron circulation volume of the qth steel plant at the beginning of the plan, in tons; Hq represents the optimal system molten iron circulation value of the qth steel plant; LGq represents the planned molten iron consumption capacity of the qth steel plant, and its value is the product of the number of furnaces LSq and the furnace input amount RRLq, LSq represents the planned number of production furnaces of the qth converted steel plant; RRLq represents the planned amount of molten iron input into one furnace of the qth converted steel plant.
6. A molten iron distribution method as claimed in claim 5, characterized in that: The third objective function is shown in the following formula: Among them, Bp represents the molten iron circulation volume of the p-th blast furnace system; BFp represents the daily production capacity of the p-th blast furnace; FGp represents the amount of scrap steel planned to be added to the p-th blast furnace.
7. A molten iron distribution method as claimed in claim 6, characterized in that: The first constraint condition is as follows: The second constraint is as follows: Among them, Mq represents the maximum system molten iron circulation volume of the qth steel plant; Nq represents the minimum system molten iron circulation volume of the qth steel plant.
8. A molten iron distribution control system, used to execute the molten iron distribution method according to any one of claims 1 to 7, the molten iron distribution control system comprising a data acquisition module, an optimization algorithm module, a control execution module, a human-computer interaction module, and a data storage module; in, The data acquisition module is used to obtain data, including resource parameters, planning parameters, priority information, and system molten iron circulation volume in the blast furnace area of each iron plant, each steel plant area, and the in-transit area; The optimization algorithm module is used to solve the best allocation scheme using the hybrid AOA algorithm; The control execution module sets the optimal configuration of molten iron resource parameters according to the operating status of the steel plant's converter equipment on that day, including the optimal value and safety value setting of molten iron in the iron plant and steel plant systems and the priority of molten iron supply in each steel plant; the allocation plan generated by the system is determined and executed by this module and sent to various departments, and the confirmed plan will be stored in the data storage module; The human-computer interaction module is used to view detailed plans and modify plan parameters; after allocation, the detailed plan view means that the allocation plan from blast furnace to converter, the source of molten iron and the consumption of steel mill can be viewed; The data storage module is used to store allocation plans, ironworks tapping plans and steelworks consumption plans; it ensures the integrity and security of data for subsequent query and analysis; and it uses a relational database (MySQL) system to provide a solid foundation for decision support.
9. A computer device / apparatus comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the molten iron distribution method according to any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the molten iron distribution method described in any one of claims 1 to 7 are implemented.
