High-proportion lump ore blast furnace burden distribution system prediction method based on CFD-DEM

Abstract

The invention discloses a high-proportion lump ore blast furnace burden distribution system prediction method based on CFD-DEM, and the method comprises the following steps: (1) carrying out an experiment on a furnace burden natural stacking angle, and obtaining furnace burden physical property characteristics conforming to an actual project; (2) constructing a three-dimensional mathematical model and performing CFD-DEM numerical simulation calibration according to the real three-dimensional size and the material distribution structure of the blast furnace; (3) constructing a high-proportion lump ore blast furnace burden distribution system prediction model based on a KAN algorithm; according to the method, efficient optimization and advanced prediction of the high-proportion lump ore blast furnace burden distribution system are achieved.

Description

Technical Field

[0001] The present invention relates to the field of energy technologies, and particularly to a prediction method for the burden distribution system of a blast furnace with a high proportion of lump ore based on CFD-DEM. Background Art

[0002] Currently, in China's iron ore pelletizing process, sintering has always been the main method, forming a burden structure of "high basicity sinter + acid pellet + a small amount of natural lump ore". However, the production of sinter and pellet belongs to a process with high energy consumption and high carbon emissions, and natural lump ore directly comes from mining, which is a zero-carbon-emission burden for the blast furnace. On the premise of maintaining the smooth operation and utilization coefficient of the blast furnace unchanged, using a high proportion of lump ore directly charged into the blast furnace for smelting to partially replace the sintering or pelletizing process can significantly reduce the energy consumption and carbon emissions in the sintering or pelletizing process. This is an important path for energy conservation and carbon reduction in the steel industry and also a top priority for low-carbon and safe smelting of the blast furnace. However, the static and dynamic impacts of the change in the burden structure under the increased proportion of lump ore on the blast furnace smelting process are still difficult to predict due to the large number of changing parameters. Under the changes in the types of burden, chute inclination angle, number of charging rings, etc., the change in the layer structure of lump ore and coke cannot be accurately obtained in advance. Summary of the Invention

[0003] Objective of the Invention: The objective of the present invention is to provide a prediction method for the burden distribution system of a blast furnace with a high proportion of lump ore based on CFD-DEM to solve problems such as increased energy consumption during the operation of the blast furnace and reduced service life caused by the unclear distribution of the burden layer structure in the furnace due to the increase in the proportion of lump ore in the blast furnace burden in the existing technical methods.

[0004] Technical Solution: The prediction method for the burden distribution system of a blast furnace with a high proportion of lump ore based on CFD-DEM according to the present invention includes the following steps:

[0005] (1) Conduct experiments on the natural stacking angle of the burden to obtain the burden physical property characteristics consistent with the actual project;

[0006] (2) According to the three-dimensional size and burden structure of an actual blast furnace, construct a three-dimensional mathematical model and perform CFD-DEM numerical simulation calibration;

[0007] (3) Construct a prediction model for the burden distribution system of a blast furnace with a high proportion of lump ore based on the KAN algorithm.

[0008] Furthermore, in step (1), the selected burden includes lump ore such as Robe River, PB, Atlas, Newman, Sierra Leone, Whyalla, sinter, and pellet; the selected coke is the reference coke charged into the furnace; among them, the stacking angles of the lump ore are (degrees) in the above arrangement order: 39 - 40, 30 - 31, 28 - 29, 40 - 41, 40 - 41, 44 - 45, 46 - 47, 38 - 39, and the coke is 35 - 36.

[0009] Further, step (2) is specifically as follows: For the three-dimensional mathematical model of blast furnace burden distribution that has met the engineering error, perform numerical simulation of the burden distribution system under multiple working conditions to obtain data on the layer structure, burden line depth, and radial ore-to-coke ratio under different burden materials, different chute inclination angles, and number of distribution rings, and classify and process the data and establish a database.

[0010] Further, in step (2), the three-dimensional structural dimensions of the blast furnace are as follows: the rotational angular velocity of the chute is 0.1 - 0.2 r / s, the diameter of the feed pipe is 0.7 - 0.8 m, the tilting distance is 0.9 - 1 m, the length of the chute is 3 - 3.5 m, the height of the chute suspension point (distance from the blast furnace throat) is 4.5 - 4.8 m, the diameter of the throat is 7.3 - 7.5 m, the diameter of the furnace waist is 11 - 12 m, the height of the throat is 1.7 - 1.9 m, the height of the hearth is 3.2 - 3.4 m, and the furnace body angle is 82 - 83°. These are used as the benchmark for constructing the three-dimensional mathematical model of the blast furnace.

[0011] Further, in step (2), through comparison with historical burden distribution parameters, perform the calibration step of the three-dimensional mathematical model of blast furnace burden distribution. Select four groups of working conditions with changes in the central coke charging angle of 13°, 15°, 17°, and 19°. The main accuracy measurement indicators are: comparison of the burden layer structure in the vertical cross-section of the blast furnace with the burden layer structure in the furnace obtained by the actual burden hammer, and the layered structure characteristics of burden materials such as the radial ore-to-coke ratio in the furnace. Compare with the calibration working conditions of the established three-dimensional mathematical model to ensure that the overall error is within 5 - 10%.

[0012] Further, in step (2), based on the CFD-DEM method, perform multi-sphere combination simulation on the burden particles to obtain the material flow trajectory, layer structure, radial ore-to-coke ratio, and burden limit angle; simulate the collisions and accumulations generated during the process of the particles entering the furnace from the chute and accumulating.

[0013] Further, step (3) is specifically as follows: Use the layer structure of the blast furnace burden as the prediction outlet, take the stratification characteristics of each burden layer in the horizontal cross-section of the blast furnace as parameter nodes, construct a KAN network, use the parameters such as the type of burden material, chute angle, and number of rings as the inlet, select a single-variable function weight to replace the traditional linear weight between the input-output nodes, divide the cross-section parameters of the burden layer in the blast furnace furnace with a vertical spacing of 1 m as the unit, select nodes at intervals of 0.5 m from the center to the edge in the radial direction for the unit cross-section, and longitudinally obtain the thickness matrix of the burden layer where the nodes are located above and below to construct a high-efficiency optimization prediction model for the blast furnace burden distribution system with a high proportion of lump ore based on the KAN algorithm.

[0014] A prediction system for the blast furnace burden distribution system with a high proportion of lump ore based on CFD-DEM according to the present invention includes:

[0015] An experimental module: used to conduct experiments on the natural stacking angle of the burden to obtain the physical properties of the burden that conform to the actual project;

[0016] Calibration module: used to construct a three-dimensional mathematical model and perform CFD-DEM numerical simulation calibration according to the real three-dimensional dimensions of the blast furnace and the burden distribution structure;

[0017] Prediction module: used to construct a prediction model for the burden distribution system of a high-proportion lump ore blast furnace based on the KAN algorithm.

[0018] An electronic device according to the present invention includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the computer program is loaded into the processor, it implements a method for predicting the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to any one of the above.

[0019] A storage medium according to the present invention stores a computer program, characterized in that when the computer program is executed by a processor, it implements a method for predicting the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to any one of the above.

[0020] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: By using CFD-DEM numerical simulation, parameters such as the accumulation and changes in the layer structure of lump ore and coke particles during the burden distribution process are obtained, providing an efficient and accurate data source for the prediction model constructed based on the KAN algorithm. The prediction model jointly constructed by CFD-DEM numerical simulation and the KAN algorithm has a rapid response and accurate results, realizing the efficient optimization and early prediction of the burden distribution system of a high-proportion lump ore blast furnace. Description of the Drawings

[0021] Figure 1 is a flowchart of the present invention;

[0022] Figure 2 is a schematic diagram of the CFD-DEM numerical simulation process in the present invention;

[0023] Figure 3 is a logic diagram of the KAN efficient optimization prediction model for the burden distribution system of a high-proportion lump ore blast furnace of the present invention. Detailed Embodiments

[0024] The technical solutions of the present invention will be further described below with reference to the drawings.

[0025] As Figure 1 shown, an embodiment of the present invention provides a method for predicting the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM, including the following steps:

[0026] S1 Conduct natural stacking experiments on different furnace charges. The selected furnace charges are lump ores: Robe River, PB, Atlas, Newman, Sierra Leone, Whyalla, sinter, and pellet; the selected coke is the benchmark coke charged into the furnace. Among them, the stacking angles of the lump ores are in the above-listed order: 39.51, 31.52, 28.59, 40.05, 40.88, 44.98, 46.64, 38.01, and that of the coke is 35. The main particle sizes are: Robe River 10 - 16 mm, PB 16 - 25 mm, Atlas 25 - 40 mm, Newman 16 - 40 mm, Sierra Leone 16 - 40 mm, Whyalla 25 - 40 mm.

[0027] S2 As Figure 2 shown, construct a three-dimensional model according to the specific structure and dimensions of the blast furnace in actual engineering. The three-dimensional structural dimensions of the blast furnace are: chute rotation angular velocity 0.125 r / s, feed pipe diameter 0.75 m, tilting distance 0.9 m, chute length 3.4 m, height of the chute suspension point (distance from the blast furnace throat) 4.6 m, throat diameter 7.45 m, furnace waist diameter 11.4 m, throat height 1.8 m, belly height 3.3 m, furnace body angle 82.64°. Use these as the benchmark for constructing the three-dimensional mathematical model of the blast furnace.

[0028] The basic parameters in the numerical simulation are: mass of a single batch of ore 50 - 70 t, mass of a single batch of coke 13 - 16 t, bulk density of a single batch of coke 550 kg / m 3 , bulk density of a single batch of ore 1950 kg / m 3; Use Rosin-Rammler to perform mass cumulative fitting on the particle sizes of different lump ores. Taking the typical burden distribution systems of Blast Furnaces No. 4 and No. 5 as examples, the coke burden distribution system of BF No. 4 is: "Corner positions: 40.5, 38, 35.5, 33, 30.5, 13 - corresponding number of rings: 3, 3, 4.5, 1.5, 1.5, 3.5", and the ore burden distribution system of BF No. 4 is: "Corner positions: 39, 36.5, 34, 31.5, 29 - corresponding number of rings: 3, 2.5, 2.5, 2.5, 2.5". The coke burden distribution system of BF No. 5 is: "Corner positions: 40, 37.5, 35, 32, 29, 13 - corresponding number of rings: 3, 3, 4.5, 1.5, 1.5, 3", and the ore burden distribution system of BF No. 5 is: "Corner positions: 38.5, 36, 33.5, 30.5, 27.5 - corresponding number of rings: 2, 2, 3, 2, 2". By comparing with historical burden distribution parameters, perform the calibration steps of the 3D mathematical model of blast furnace burden distribution. Select four groups of working conditions with changes in the central coke charging angle of 13°, 15°, 17° and 19°. The main accuracy indicators are: comparing the burden layer structure in the vertical section of the blast furnace with the burden layer structure in the furnace obtained by the burden hammer in the actual project, and the layered structure characteristics of burden materials in the furnace such as the radial ore-to-coke ratio, etc., and comparing with the calibration working conditions of the established 3D mathematical model to ensure that the overall error is less than 10%. Based on the CFD-DEM method, perform multi-sphere combination simulation on the burden particles to approximate the characteristics of real non-spherical burden particles, so as to obtain more accurate burden flow trajectories, burden layer structures, radial ore-to-coke ratios and burden limit angles and other burden structure parameters. In the more accurate non-spherical particle fitting process, accurately simulate the collisions and accumulations generated during the process of particles entering the furnace from the chute and accumulating.

[0029] Among them, the super-ellipsoid standard equation is used to describe the burden particles:

[0030]

[0031] Among them, x, y, and z represent the three axes respectively, a, b, and c represent the axis distances of the ellipsoid, and S2 and S1 are constants representing the ellipsoidal variables.

[0032] The particle motion equation is:

[0033]

[0034] Among them, m is the particle mass, v is the velocity vector, t is the time, g is the acceleration due to gravity, and F c is the superposition of the remaining forces.

[0035] S3 is as Figure 3As shown, a high - proportion lump - ore burden distribution system efficient optimization prediction model is constructed based on CFD - DEM numerical simulation data and the KAN algorithm. In the numerical simulation data, the burden layer structure of the blast furnace burden is used as the prediction outlet, and the characteristics of each burden layer in the horizontal cross - section of the blast furnace are used as parameter nodes to construct the KAN network. The description of the KAN algorithm is as follows:

[0036]

[0037] Among them, φ q,p (x p ) belongs to the internal function (the 0th layer in the architecture), and Φ q (x q ) belongs to the external function (the 1st layer in the architecture). The basic composition of the KAN network consists of nodes, edges, and KAN layers. Among them, nodes are formed according to the relationship of edges and perform simple addition operations without including non - linear activation functions; edges have the characteristics of non - linear activation function weights; the KAN layer is a one - dimensional function matrix composed of n - dimensional input and n - dimensional output (n is a set value). In this network, parameters such as burden type, chute angle, and ring number changes are used as the inlet, and the burden layer characteristics of different cross - section nodes are used as the outlet for model construction.

[0038] In the KAN network, B - spline (Basic Spline) is used for construction, which is divided into the following steps:

[0039] (1) Residual function activation

[0040] Use a basic function b(x) so that the activation function Φ x is the sum of the basic function b(x) and the spline function.

[0041] φ(x)=w(b(x)+spline(x))

[0042] b(x)=silu(x)=x / (1 + e -x )

[0043] spline(x)=∑ i c i B i (x)

[0044] (2) Initialization

[0045] Each activation function is initialized to spline(x)≈0 2 , and w is initialized according to the initialization method of the linear layer.

[0046] (3) Update of the spline grid

[0047] Each network is updated in real time according to its input activation value to solve the problem that the spline function is defined in a bounded region while the activation value may exceed the fixed region during the training process.

[0048] Number of parameters: Assume a network has a depth L, equal width N for each layer, and the order of each spline is k (usually 3). On G intervals (G + 1 network points), the total number of KAN parameters can be estimated as O(N 2 L(G + k) ~ O(N 2 LG).

[0049] Taking parameters such as burden type, chute angle, and number of rings as the entrance, using a single-variable function weight to replace the traditional linear weight between input-output nodes, dividing the cross-sectional parameters of the burden layer structure in the blast furnace with a vertical spacing of 1m as the unit, taking the radial direction as the line for the unit cross-section, selecting nodes at intervals of 0.5m from the center to the edge, obtaining the thickness matrix of the burden layer above and below the nodes longitudinally, and constructing an efficient optimization prediction model for the burden distribution system of a high-proportion lump ore blast furnace based on the KAN algorithm.

[0050] An embodiment of the present invention provides a prediction system for the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM, including:

[0051] An experimental module: used to conduct experiments on the natural angle of repose of the burden to obtain the physical properties of the burden that conform to the actual project;

[0052] A calibration module: used to construct a three-dimensional mathematical model and perform CFD-DEM numerical simulation calibration according to the three-dimensional size and burden distribution structure of the actual blast furnace;

[0053] A prediction module: used to construct a prediction model for the burden distribution system of a high-proportion lump ore blast furnace based on the KAN algorithm.

[0054] An embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the computer program is loaded into the processor, it implements a prediction method for the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to any one of the above.

[0055] An embodiment of the present invention provides a storage medium. The storage medium stores a computer program, and is characterized in that when the computer program is executed by a processor, it implements a prediction method for the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to any one of the above.

Claims

1. A prediction method for the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM, characterized in that, It includes the following steps: (1) Conduct experiments on the natural stacking angle of the burden materials to obtain the physical properties of the burden materials that conform to the actual project; (2) According to the three-dimensional size and burden distribution structure of the real blast furnace, construct a three-dimensional mathematical model and conduct CFD-DEM numerical simulation calibration; (3) Build a prediction model for the burden distribution system of a blast furnace with a high proportion of lump ore based on the KAN algorithm.

2. The prediction method for the burden distribution system of a blast furnace with a high proportion of lump ore based on CFD-DEM according to claim 1, wherein In step (1), the burden materials selected are lump ores including Robe River, PB, Atlas, Newman, Sierra Leone, Whyalla, sinter, and pellet; the coke selected is the reference coke charged into the furnace. Among them, the stacking angles of the lump ores are (in degrees) in the above arrangement order: 39 - 40, 30 - 31, 28 - 29, 40 - 41, 40 - 41, 44 - 45, 46 - 47, 38 - 39, and the coke is 35 - 36.

3. A prediction method for the burden distribution system of a blast furnace with a high proportion of lump ore based on CFD-DEM according to claim 1, wherein Step (2) is specifically as follows: Conduct numerical simulation work on the burden distribution system under multiple conditions for the three-dimensional mathematical model of blast furnace burden distribution that has met the engineering error, obtain data on the layer structure, burden line depth, and radial ore-to-coke ratio under different burden materials, different chute inclination angles, and burden distribution rings, and classify and process the data and establish a database.

4. A prediction method for the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to claim 3, characterized in that In step (2), the range of the three-dimensional structure dimensions of the blast furnace is: the chute rotation angular velocity is 0.1 - 0.2 r / s, the diameter of the feed pipe is 0.7 - 0.8 m, the tilting distance is 0.9 - 1 m, the length of the chute is 3 - 3.5 m, the height of the chute suspension point (distance from the blast furnace throat) is 4.5 - 4.8 m, the diameter of the throat is 7.3 - 7.5 m, the diameter of the furnace waist is 11 - 12 m, the height of the throat is 1.7 - 1.9 m, the height of the furnace belly is 3.2 - 3.4 m, and the furnace body angle is 82 - 83°. These are used as the benchmarks for constructing the three-dimensional mathematical model of the blast furnace.

5. A prediction method for the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to claim 4, characterized in that In step (2), through comparison with historical burden distribution parameters, conduct the calibration step of the three-dimensional mathematical model of blast furnace burden distribution. Select four sets of working conditions with changes in the center coke charging angle of 13°, 15°, 17°, and 19°. The main accuracy metrics are: comparison of the burden layer structure in the vertical cross-section of the blast furnace with the burden layer structure in the furnace obtained by the actual burden hammer, and the layered structure characteristics of burden materials such as the radial ore-to-coke ratio in the furnace, and compare with the calibration working conditions of the established three-dimensional mathematical model to ensure that the overall error is within 5 - 10%.

6. A prediction method for the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to claim 5, characterized in that In step (2), based on the CFD-DEM method, conduct multi-sphere combination simulation on the burden material particles to obtain the material flow trajectory, layer structure, radial ore-to-coke ratio, and burden distribution limit angle; Simulate the collisions and accumulations that occur during the process of the particles entering the furnace from the chute and accumulating.

7. A prediction method for the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to claim 1, characterized in that Step (3) is specifically as follows: Use the layer structure of the blast furnace burden as the prediction outlet, take the characteristics of each burden material layer in the horizontal cross-section of the blast furnace as parameter nodes, construct a KAN network, use the parameters such as the type of burden material, chute angle, and ring number change as the inlet, select a single-variable function weight to replace the traditional linear weight between the input-output nodes, divide the cross-section parameters of the burden layer structure in the blast furnace furnace vertically at a vertical interval of 1 m, select nodes at intervals of 0.5 m from the center to the edge in the radial direction for the unit cross-section, and longitudinally obtain the thickness matrix of the burden layer where the nodes are located above and below, and construct an efficient optimization prediction model for the burden distribution system of a blast furnace with a high proportion of lump ore based on the KAN algorithm.

8. A prediction system for the burden distribution system of a blast furnace with a high proportion of lump ore based on CFD-DEM, characterized in that, It includes: Experimental module: used to conduct experiments on the natural stacking angle of burden materials to obtain burden material physical properties consistent with actual engineering; Calibration module: used to construct a three-dimensional mathematical model and perform CFD-DEM numerical simulation calibration based on the three-dimensional size and burden distribution structure of an actual blast furnace; Prediction module: used to construct a prediction model for the burden distribution system of a high-proportion lump ore blast furnace based on the KAN algorithm.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the computer program is loaded into the processor, it implements a method for predicting the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to any one of claims 1-7.

10. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements a method for predicting the burden distribution system of a high-proportion lump ore blast furnace based on CFD-DEM according to any one of claims 1-7.

Images