A fully automatic demoulding and sampling system for molten steel in blast furnaces

Abstract

The present invention belongs to the technical field of metallurgical automation, and provides a fully automatic demoulding and sampling system for molten steel in a blast furnace, comprising a flow pivot modeling, an oxygen field cloud map, an offset decision and a sample optimization module; the flow pivot modeling module acquires data such as molten steel flow rate and temperature, and combines CFD to construct a flow pivot mapping model to obtain an initial optimal sampling point; the oxygen field cloud map module introduces an oxygen mass transfer equation, and outputs a cloud map of oxygen content distribution in the furnace and the oxygen content at the initial optimal sampling point; the offset decision module determines whether to offset based on the oxygen content, obtains an oxygen-containing offsettable area and optimizes the target offsettable area; the sample optimization module selects the optimal sample through multiple samplings in combination with the offset amount and oxygen content; the present invention improves sampling accuracy, reduces the risk of defects caused by excessive oxygen, and is suitable for intelligent blast furnace molten steel sampling scenarios.

Description

Technical Field

[0001] The invention belongs to the technical field of metallurgical automation, in particular to a full-automatic demoulding and sampling system for molten steel in a blast furnace. Background Art

[0002] With the acceleration of the intelligent transformation of the steel industry, precise control of blast furnace molten steel quality based on multi-physics field coupling analysis has become a research hotspot in the field of smelting process optimization. However, there are still significant technical bottlenecks in the blast furnace molten steel sampling link. Therefore, the development of a fully automatic demoulding and sampling system for blast furnace molten steel is of great significance.

[0003] In the prior art, although the existing blast furnace molten steel sampling device can sample and process the molten steel, it does not take into account the influence of the flow characteristics of the blast furnace molten steel and the oxygen content on the blast furnace molten steel sample. Therefore, on the one hand, the sampling device is limited to a fixed-point mechanical sampling mode, which makes it difficult to reflect the uniformity of the overall composition of the molten steel, resulting in a non-representative sample. On the other hand, not considering the oxygen content distribution of the molten steel may lead to the risk of "empty sampling" of the sample or the presence of bubbles on the surface and inside of the sample, making the sampling results inaccurate and causing large human errors, which is not conducive to adjusting and optimizing the production plan of the blast furnace molten steel.

[0004] To this end, the present invention provides a fully automatic demoulding and sampling system for molten steel in a blast furnace. Summary of the Invention

[0005] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0006] The technical solution adopted by the present invention to solve the technical problem is: a blast furnace molten steel fully automatic demoulding sampling system, comprising:

[0007] Flow pivot modeling module: This module obtains blast furnace molten steel flow data and combines it with CFD (fluid dynamics) to build a flow pivot mapping model, which is used to obtain the initial optimal sampling point.

[0008] Oxygen field cloud map module: Add the oxygen mass transfer equation to the flow pivot mapping model, output the oxygen content distribution cloud map in the furnace, and obtain the oxygen content at the initial optimal sampling point;

[0009] Offset decision module: determines whether the initial optimal sampling point needs to be offset based on the oxygen content of the initial optimal sampling point. If necessary, obtains the oxygen-containing offset area, selects the oxygen-containing offset area, and determines the target offset area;

[0010] Sample optimization module: multiple blast furnace molten steel samples are taken in the target offset area, and the target molten steel sampling samples are selected based on the oxygen content of the target sampling point to be determined and the offset processing results.

[0011] Furthermore, the method for obtaining blast furnace molten steel flow data is as follows:

[0012] The process of building a flow pivot mapping model in combination with CFD is as follows:

[0013] Using blast furnace molten steel flow data, CAD software is used to construct a molten steel flow model;

[0014] Mesh the molten steel flow model, obtain the mesh node coordinates of the molten steel flow model, define the initial blast furnace molten steel flow data, calculate the density, viscosity and boundary type of the molten steel based on the initial molten steel properties, and output the initial flow pivot characteristics in combination with the CFD solver;

[0015] Flow pivot features include: molten steel characteristic points, characteristic lines and characteristic area features;

[0016] The blast furnace molten steel flow data is input into the flow pivot mapping model multiple times, and the actual flow pivot characteristics of the model are obtained through the CFD solver. The mean square error is processed with the initial flow pivot characteristics, and the model parameters are adjusted to construct the flow pivot mapping model.

[0017] Furthermore, the process of building a flow pivot mapping model in combination with CFD is as follows:

[0018] The process of obtaining the initial optimal sampling point is:

[0019] By analyzing the flow pivot characteristics output by the model, the flow pivot area is divided into eddy flow pivot area, high flow velocity pivot area and low flow velocity pivot area. Random sampling is performed on different flow pivot areas to obtain the molten steel composition data of the sampling points in different flow pivot areas. The molten steel composition data are divided and summarized according to the regional categories to obtain the molten steel composition data set.

[0020] Among them, the molten steel composition data at sampling points in different flow pivot areas are obtained by installing a direct reading spectrometer in the blast furnace;

[0021] The steel composition data sets of different flow pivot areas are processed with standard deviation to obtain the uniform value of steel element content;

[0022] Select the flow pivot area corresponding to the lowest uniform value of molten steel element content as the candidate sampling area, and randomly generate several candidate sampling points in the candidate sampling area;

[0023] The candidate sampling points are selected by combining the particle swarm optimization algorithm to obtain the initial optimal sampling point.

[0024] Furthermore, the oxygen mass transfer equation is added to the flow pivot mapping model, and the process of outputting the oxygen content distribution cloud map in the furnace is as follows:

[0025] Construct an oxygen mass transfer equation to obtain oxygen concentration data in the molten steel fluid model;

[0026] The oxygen mass transfer equation is combined with the flow pivot mapping model to construct an LSTM model. The flow pivot characteristics and oxygen concentration data output by the CFD solver are used as input and fed into the LSTM model multiple times to train and learn the mapping relationship between the flow pivot mapping model characteristics and oxygen content.

[0027] The flow pivot mapping model features are used as the input layer, the oxygen concentration data change rate is used as the hidden layer, and the oxygen concentration is used as the output layer;

[0028] The blast furnace molten steel flow data corresponding to the grid node coordinates of the molten steel fluid model is passed through the LSTM model to output the oxygen concentration of each grid node coordinate;

[0029] The oxygen concentration of each grid node coordinate is marked one by one with the grid node coordinates in the molten steel fluid model, and the oxygen content distribution cloud map in the furnace is output through the CFD solver.

[0030] Furthermore, the process of obtaining the oxygen content at the initial optimal sampling point is as follows:

[0031] The oxygen content at the initial optimal sampling point can be obtained by obtaining the grid node coordinates corresponding to the initial optimal sampling point in the molten steel fluid model and obtaining the oxygen content corresponding to the coordinates in the oxygen content distribution cloud map in the furnace.

[0032] Furthermore, the process of determining whether to perform an offset process on the initial optimal sampling point according to the oxygen content of the initial optimal sampling point is as follows:

[0033] Compare and analyze the oxygen content at the initial optimal sampling point with the standard oxygen content threshold;

[0034] If the oxygen content at the initial optimal sampling point is less than or equal to the standard oxygen content threshold, it is determined that the initial optimal sampling point does not need to be offset;

[0035] Otherwise, it is determined that the initial optimal sampling point needs to be offset.

[0036] Furthermore, based on the oxygen content distribution cloud map in the furnace, the process of obtaining the oxygen-containing deflectable area is as follows:

[0037] If it is determined that the initial optimal sampling point needs to be offset, the oxygen content distribution cloud map in the furnace is divided into several unit areas of equal area based on the oxygen content distribution cloud map in the furnace. If the oxygen content distribution cloud map in the furnace shows that the oxygen content in the unit area is less than or equal to the standard oxygen content threshold, then the unit area is an oxygen-containing offset area.

[0038] Furthermore, the process of determining the target offset area is as follows:

[0039] Based on any oxygen-containing deflectable region;

[0040] Obtain the shortest distance between the initial optimal sampling point and the oxygen-containing deflectable area to obtain the deflection distance;

[0041] The offset distance is ratioed to the total offset distance to obtain the offset distance ratio;

[0042] Obtain the average oxygen content in the oxygen-containing deflectable area and calculate the ratio with the total average oxygen content to obtain the average oxygen content ratio;

[0043] The offset distance ratio and the average oxygen content ratio are weighted and the output is the comprehensive offset assessment value.

[0044] The oxygen-containing deflectable area corresponding to the minimum deflection comprehensive evaluation value is selected as the target deflection area.

[0045] Furthermore, the process of selecting the target molten steel sampling sample is as follows:

[0046] Gridding the target offset area, obtaining grid node coordinates, and obtaining multiple target sampling points to be determined;

[0047] Sampling molten steel from a blast furnace at each target sampling point to be determined, to obtain molten steel sampling samples from a plurality of target sampling points to be determined;

[0048] Obtain the actual sampling offset ratio of molten steel and the oxygen content ratio of molten steel sampling samples;

[0049] The actual sampling offset ratio of the molten steel and the oxygen content ratio of the molten steel sampling sample are summed to obtain the sample selection value of the molten steel sampling sample;

[0050] The molten steel sampling sample at the target sampling point to be determined corresponding to the minimum sample selection value is selected as the molten steel target sampling sample.

[0051] Furthermore, the process of obtaining the actual sampling offset ratio of molten steel and the oxygen content ratio of the molten steel sampling sample is as follows:

[0052] Obtaining the initial optimal sampling point of the blast furnace molten steel when sampling each molten steel sampling sample, and calculating the position deviation with the target sampling point to be determined, to obtain the actual sampling offset of the molten steel sampling sample;

[0053] Calculate the ratio of the actual sampling offset of the molten steel sample to the actual sampling offset total to obtain the actual sampling offset ratio of the molten steel;

[0054] The oxygen content of the initial optimal sampling point of the blast furnace molten steel is obtained when the molten steel sampling sample is sampled, and the oxygen content ratio of the molten steel sampling sample is calculated by proportional calculation with the total oxygen content.

[0055] The beneficial effects of the present invention are as follows: flow pivot modeling module: obtains blast furnace molten steel flow data, combines CFD to construct a flow pivot mapping model, and obtains the initial optimal sampling point through the flow pivot mapping model; oxygen field cloud map module: adds the oxygen mass transfer equation to the flow pivot mapping model, outputs the oxygen content distribution cloud map in the furnace, and obtains the oxygen content of the initial optimal sampling point; offset decision module: judges whether the initial optimal sampling point needs to be offset according to the oxygen content of the initial optimal sampling point. If necessary, the oxygen-containing offset area is determined according to the oxygen content distribution cloud map in the furnace, and the initial optimal sampling point is offset according to the oxygen-containing offset area; sample selection module: samples the blast furnace molten steel multiple times, and selects the target sampling sample of the molten steel based on the oxygen content of the target sampling point and the offset processing result. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The present invention will be further described below with reference to the accompanying drawings.

[0057] Figure 1 This is a module block diagram of a fully automatic demoulding and sampling system for molten steel in a blast furnace according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0059] Example 1: Please refer to Figure 1 As shown, a fully automatic demoulding and sampling system for molten steel in a blast furnace according to an embodiment of the present invention includes the following modules:

[0060] Flow pivot modeling module: This module obtains blast furnace molten steel flow data and combines it with CFD (fluid dynamics) to build a flow pivot mapping model, which is used to obtain the initial optimal sampling point.

[0061] The process of obtaining blast furnace molten steel flow data is as follows:

[0062] Blast furnace molten steel flow data includes but is not limited to: molten steel flow rate, temperature, pressure and liquid level height;

[0063] Ultrasonic Doppler flowmeters are installed on the side walls and bottom of the blast furnace to obtain molten steel flow rate data through high-frequency sampling;

[0064] Monitor the temperature distribution of molten steel by arranging infrared thermal imagers;

[0065] Using a pressure sensor array to measure the dynamic pressure field inside the blast furnace;

[0066] By using a laser level meter to monitor the molten steel level in real time;

[0067] In the flow pivot modeling module, the process of building a flow pivot mapping model in combination with CFD is as follows:

[0068] Among them, CFD is based on the Navier-Stokes equations and uses discretization methods to convert the continuous flow of molten steel into a set of algebraic equations;

[0069] The specific steps are: using CAD software to construct a molten steel fluid model and perform mesh division on the molten steel fluid model;

[0070] Define the initial blast furnace molten steel flow data, calculate the density, viscosity, and boundary type (such as high-speed flow boundary and low-speed flow boundary) of the molten steel based on the initial molten steel properties, and combine with the CFD solver to output the initial flow pivot characteristics;

[0071] Flow pivot features include: molten steel characteristic points, characteristic lines and characteristic regions. The molten steel characteristic points are the vortex centers, the molten steel characteristic lines are the velocity dividing lines, and the molten steel characteristic regions are the high velocity region and the low velocity region.

[0072] The blast furnace molten steel flow data is input into the flow pivot mapping model multiple times. The actual flow pivot characteristics of the model are obtained through the CFD solver. The mean square error is processed with the initial flow pivot characteristics, and the model parameters are adjusted to construct the flow pivot mapping model.

[0073] In the flow pivot modeling module, the process of obtaining the initial optimal sampling point through the flow pivot mapping model is as follows:

[0074] Install a direct reading spectrometer in the blast furnace to obtain molten steel composition data, which includes but is not limited to the content of elements such as carbon, silicon, and manganese in the molten steel;

[0075] By analyzing the flow pivot characteristics output by the flow pivot mapping model, the flow pivot region is divided into eddy flow pivot region, high flow velocity flow pivot region and low flow velocity flow pivot region. Random sampling is performed on different flow pivot regions to obtain the molten steel composition data of the sampling points in different flow pivot regions. The molten steel composition data are divided and summarized according to the regional categories to obtain the molten steel composition data set.

[0076] The steel composition data sets of different flow pivot areas are processed with standard deviation to obtain the uniform value of steel element content;

[0077] Select the flow pivot area corresponding to the lowest uniform value of molten steel element content as the candidate sampling area, and randomly generate several candidate sampling points in the candidate sampling area;

[0078] Exemplarily, the process of determining the candidate sampling area is as follows:

[0079] In the 150-ton LF refining furnace scenario, the flow pivot mapping model outputs the following flow pivot characteristics: a stable vortex forms at node (5,5,5). Random sampling around the node yields a molten steel composition data set of {0.003%, 0.004%, 0.005%} for iron content. Standard deviation processing is performed, and the standard deviation of the iron content in this region is 0.001%. Similarly, the standard deviation of the iron content at node (8,5,3) in the high-speed flow region is 0.005%, and the iron content at node (1,1,1) in the near-wall low-speed region is 0.016%. Therefore, the vortex center region (e.g., near node 5,5,5) should be selected as the candidate sampling area.

[0080] The candidate sampling points are selected by combining the particle swarm optimization algorithm to obtain the initial optimal sampling point;

[0081] For example, the method of selecting candidate sampling points in combination with the particle swarm optimization algorithm is as follows:

[0082] In the 150-ton LF refining furnace, the candidate sampling areas have been identified by the flow pivot mapping model as (5, 5, 5), (8, 5, 3), (1, 1, 1), (10, 2, 7), randomly select 20 candidate sampling points in the candidate sampling area, combine the particle swarm optimization algorithm, define the algorithm parameters, calculate the initial fitness, and perform iterative updates: adjust the particle position through the speed update formula to make the particles gradually move to the , , , Regional aggregation, after 50 iterative updates, the final fitness of the candidate sampling points in the candidate sampling area is output as (0.92), (0.82), (0.85), (0.90), we get The final fitness of the candidate sampling point is the highest, and it is selected according to the output result The (5, 5, 5) node is the initial optimal sampling point;

[0083] Oxygen field cloud map module: Add the oxygen mass transfer equation to the flow pivot mapping model, output the oxygen content distribution cloud map in the furnace, and obtain the oxygen content at the initial optimal sampling point;

[0084] In the oxygen field cloud map module, the process of constructing the oxygen mass transfer equation is as follows:

[0085] Based on Fick's second law, the oxygen mass transfer equation is established:

[0086]

[0087] in, is the oxygen concentration, is the molten steel velocity vector, is the gradient of oxygen concentration, is the oxygen diffusion coefficient (temperature-dependent), is the divergence of the oxygen concentration gradient, As oxygen source;

[0088] The calculation formula is:

[0089]

[0090] in, is the mass transfer coefficient between furnace gas and molten steel, a is the specific surface area of ​​gas-liquid interface, is the oxygen saturation concentration;

[0091] In the oxygen field cloud map module, the oxygen mass transfer equation is combined with the flow pivot mapping model to output the oxygen content distribution cloud map in the furnace as follows:

[0092] The flow pivot characteristics and oxygen concentration data output by the CFD solver are used as input and fed into the LSTM model multiple times to train and learn the mapping relationship between the flow pivot mapping model characteristics and oxygen content.

[0093] The pivot mapping model features are used as the input layer, the oxygen concentration data change rate is used as the hidden layer, and the oxygen concentration is used as the output layer;

[0094] The blast furnace molten steel flow data corresponding to the CFD-divided grid nodes is passed through the LSTM model to output the oxygen concentration of each grid node;

[0095] The oxygen concentration of each grid node is marked one by one with the grid nodes in the molten steel flow model, and a cloud map of the oxygen content distribution in the furnace is output through the CFD solver;

[0096] For example, taking a 150-ton LF refining furnace as the object, CFD is used to divide a 10×10×10 three-dimensional grid (a total of 1000 nodes) to simulate the dynamic distribution of oxygen content in molten steel under bottom blowing argon stirring. The steady-state flow field is calculated by the CFD solver, and the fluid mechanics parameters of each node are extracted: molten steel flow rate V: node (5,5,5): 0.32m / s, direction along the -z axis), temperature T (node ​​(5,5,5): 1873K), turbulence intensity k (node ​​(5,5,5): 0.015m² / s²), and the initial oxygen content is set. The center of the molten pool (node ​​(5,5,5)): C0 = 0.002%, the edge of the molten pool (node ​​(1,1,1)): C0 = 0.005%, set the node characteristics [v,T,k] of the flow pivot mapping model (node ​​(5,5,5) input: [0.32,1873,0.015]) The oxygen concentration change rate is 0.000008% / s, and the output predicted oxygen concentration is 0.0018%. The predicted oxygen concentration of 1000 nodes is mapped to the CFD grid. Through the CFD solver, the oxygen content distribution cloud map in the furnace is output;

[0097] In the oxygen field cloud map module, the oxygen content at the initial optimal sampling point is obtained as follows:

[0098] Obtain the position corresponding to the initial optimal sampling point in the molten steel fluid model and obtain the oxygen content corresponding to the position, so as to obtain the oxygen content at the initial optimal sampling point;

[0099] Offset decision module: Determines whether the initial optimal sampling point needs to be offset based on the oxygen content at the initial optimal sampling point. If necessary, obtains the oxygen-containing offset area based on the oxygen content distribution cloud map in the furnace, selects the oxygen-containing offset area, and determines the target offset area.

[0100] In the offset decision module, the process of determining whether to perform offset processing of the initial optimal sampling point according to the oxygen content of the initial optimal sampling point is as follows:

[0101] Compare and analyze the oxygen content at the initial optimal sampling point with the standard oxygen content threshold, where the standard oxygen content threshold is set by the sampling personnel;

[0102] If the oxygen content at the initial optimal sampling point is less than or equal to the standard oxygen content threshold, it is determined that the initial optimal sampling point does not need to be offset;

[0103] If the oxygen content at the initial optimal sampling point is greater than the standard oxygen content threshold, it is determined that the initial optimal sampling point needs to be offset;

[0104] For example, according to the oxygen content distribution cloud map in the furnace, the oxygen content at the initial optimal sampling point is , the standard oxygen threshold is , by comparing the oxygen content of the initial optimal sampling point with the standard oxygen content threshold, it is determined that the initial optimal sampling point needs to be offset;

[0105] It can be understood that if the oxygen content at the initial optimal sampling point is greater than the standard oxygen content threshold, the purpose of determining that the initial optimal sampling point needs to be offset is that if the collected sample has an excessively high oxygen content, it may cause a "blank sampling" phenomenon in the sample. Therefore, by determining that the initial optimal sampling point needs to be offset, the "blank sampling" phenomenon can be prevented, thereby ensuring the accuracy of the sample.

[0106] In the offset decision module, the process of obtaining the oxygen-containing offset area based on the oxygen content distribution cloud map in the furnace is as follows:

[0107] If it is determined that the initial optimal sampling point needs to be offset, the oxygen content distribution cloud map in the furnace is divided into several unit areas of equal area based on the oxygen content distribution cloud map in the furnace. If the oxygen content distribution cloud map in the furnace shows that the oxygen content in the unit area meets the requirements (the oxygen content is less than or equal to the standard oxygen content threshold), the corresponding unit area is defined as an oxygen-containing offset area;

[0108] It should be noted that if there are adjacent unit regions that are both oxygen-containing and deflectable regions, they are merged to form a final oxygen-containing and deflectable region;

[0109] In the migration decision module, the process of selecting oxygen-containing migration areas and determining the target migration areas is as follows:

[0110] Based on any oxygen-containing deflectable region;

[0111] Obtain the shortest distance between the initial optimal sampling point and the oxygen-containing deflectable area to obtain the deflection distance;

[0112] Calculate the ratio of the offset distance to the total offset distance to obtain the offset distance ratio;

[0113] The total displacement distance is the sum of the shortest distances between the initial optimal sampling point and each oxygen-containing displacement area;

[0114] Obtain the average oxygen content in the oxygen-containing deflectable area and calculate the ratio with the total average oxygen content to obtain the average oxygen content ratio;

[0115] The total mean oxygen content is the sum of the average oxygen contents of all oxygen-containing and deflectable regions;

[0116] The offset distance ratio and the average oxygen content ratio are weighted and the output is the offset comprehensive assessment value; the weight coefficients of the offset distance ratio and the average oxygen content ratio are 2:8 respectively;

[0117] The oxygen-containing deflectable area corresponding to the minimum deflection comprehensive evaluation value is selected as the target deflection area;

[0118] It is understandable that the significance of obtaining the comprehensive evaluation value of the offset is:

[0119] The comprehensive offset assessment value is calculated by combining the offset distance ratio and the average oxygen content ratio. The purpose is to find an area with a short offset distance and low oxygen content within multiple offsettable areas. This ensures that the sampling meets the uniform flow of molten steel, and the oxygen content at the final offset sampling point is low, reducing the risk of empty sampling. Therefore, the offset distance and oxygen content are used to comprehensively determine the offset selection of the initial optimal sampling point, thereby improving sampling accuracy.

[0120] Sample optimization module: multiple blast furnace molten steel samples are taken in the target offset area, and the target molten steel sampling samples are selected based on the oxygen content of the target sampling point to be determined and the offset processing results;

[0121] In the sample optimization module, the process of selecting the sample based on the oxygen content of the target sampling point to be determined and the offset processing result is as follows:

[0122] The target offset area is grid-divided to obtain a plurality of target sampling points to be determined;

[0123] Sampling molten steel from a blast furnace at each target sampling point to be determined, to obtain molten steel sampling samples from a plurality of target sampling points to be determined;

[0124] According to the flow pivot mapping model, the initial optimal sampling point of the blast furnace molten steel is obtained when each molten steel sampling sample is sampled, and the position deviation is calculated with the target sampling point to be determined to obtain the actual sampling offset of the molten steel sampling sample;

[0125] Calculate the ratio of the actual sampling offset of the molten steel sample to the actual sampling offset total to obtain the actual sampling offset ratio of the molten steel;

[0126] The total actual sampling offset is the sum of the actual sampling offsets of all molten steel sampling samples;

[0127] The oxygen content of the initial optimal sampling point of the blast furnace molten steel is obtained by the oxygen content distribution cloud map in the furnace, and the oxygen content ratio of the molten steel sampling sample is obtained by proportional calculation with the total oxygen content value;

[0128] The total oxygen content is the sum of the oxygen contents of the initial optimal sampling points of the blast furnace molten steel when all molten steel samples are sampled;

[0129] The actual sampling offset ratio of the molten steel is summed with the oxygen content ratio of the molten steel sampling sample to obtain the sample selection value of the molten steel sampling sample;

[0130] Select the molten steel sampling sample corresponding to the minimum sample selection value as the molten steel target sampling sample;

[0131] It should be noted that the physical meaning of the sample selection value is:

[0132] The sample selection value is obtained by summing the actual sampling offset ratio of molten steel and the proportion of oxygen content in the molten steel sampling sample. The reason is that the actual sampling offset ratio of molten steel reflects the proportion of the actual sampling offset to the total actual sampling offset, and the oxygen content proportion of molten steel sampling sample reflects the proportion of the oxygen content at the initial optimal sampling point to the total oxygen content. The smaller the ratio of the actual sampling offset ratio of molten steel and the oxygen content proportion of molten steel sampling sample, the higher the sampling accuracy of the molten steel sampling sample. The actual location point of blast furnace molten steel sampling meets the sampling uniformity and the low oxygen control requirements at the same time. Therefore, the optimal collection sample is obtained through the sample selection value.

[0133] The technical solution of an embodiment of the present invention is to obtain blast furnace molten steel flow data, construct a flow pivot mapping model in combination with computational fluid dynamics (CFD), and obtain an initial optimal sampling point through the flow pivot mapping model, add an oxygen mass transfer equation to the flow pivot mapping model, output an oxygen content distribution cloud map in the furnace, and obtain the oxygen content of the initial optimal sampling point, determine whether to perform offset processing of the initial optimal sampling point based on the oxygen content of the initial optimal sampling point, and if necessary, obtain an oxygen-containing offsettable area based on the oxygen content distribution cloud map in the furnace, select the oxygen-containing offsettable area, determine the target offset area, sample the blast furnace molten steel multiple times in the target offset area, and select the target sampling sample of the molten steel based on the oxygen content of the target sampling point to be determined and the offset processing result.

[0134] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A fully automatic demoulding and sampling system for molten steel in a blast furnace, characterized by: include: Flow pivot modeling module: This module obtains blast furnace molten steel flow data and builds a flow pivot mapping model in combination with CFD to obtain the initial optimal sampling point. The process of building a flow pivot mapping model in combination with CFD is as follows: Using blast furnace molten steel flow data, CAD software is used to construct a molten steel flow model; Meshing the molten steel fluid model to obtain the mesh node coordinates of the molten steel fluid model; Define the initial blast furnace molten steel flow data, calculate the density, viscosity and boundary type of the molten steel based on the initial molten steel properties, and combine with the CFD solver to output the initial flow pivot characteristics; Flow pivot features include: molten steel characteristic points, characteristic lines and characteristic area features; The blast furnace molten steel flow data is input into the flow pivot mapping model multiple times. The actual flow pivot characteristics of the model are obtained through the CFD solver. The mean square error is processed with the initial flow pivot characteristics, and the model parameters are adjusted to construct the flow pivot mapping model. The process of obtaining the initial optimal sampling point is: By analyzing the flow pivot characteristics output by the flow pivot mapping model, the flow pivot region is divided into eddy flow pivot region, high flow velocity flow pivot region and low flow velocity flow pivot region. Random sampling is performed on different flow pivot regions to obtain the molten steel composition data of the sampling points in different flow pivot regions. The molten steel composition data are divided and summarized according to the regional categories to obtain the molten steel composition data set. Among them, the molten steel composition data at sampling points in different flow pivot areas are obtained by installing a direct reading spectrometer in the blast furnace; The steel composition data sets of different flow pivot areas are processed with standard deviation to obtain the uniform value of steel element content; Select the flow pivot area corresponding to the lowest uniform value of molten steel element content as the candidate sampling area, and randomly generate several candidate sampling points in the candidate sampling area; The candidate sampling points are selected by combining the particle swarm optimization algorithm to obtain the initial optimal sampling point; Oxygen field cloud map module: Add the oxygen mass transfer equation to the flow pivot mapping model, output the oxygen content distribution cloud map in the furnace, and obtain the oxygen content at the initial optimal sampling point; Offset decision module: Determines whether the initial optimal sampling point needs to be offset based on the oxygen content at the initial optimal sampling point. If necessary, the oxygen-containing offset area is obtained based on the oxygen content distribution cloud map in the furnace, and the oxygen-containing offset area is selected to determine the target offset area. Sample optimization module: multiple blast furnace molten steel samples are taken in the target offset area, and the target molten steel sampling samples are selected based on the oxygen content of the target sampling point to be determined and the offset processing results.

2. A blast furnace molten steel fully automatic demoulding sampling system according to claim 1, characterized in that: The process of adding the oxygen mass transfer equation to the flow pivot mapping model and outputting the oxygen content distribution cloud map in the furnace is as follows: Construct an oxygen mass transfer equation to obtain oxygen concentration data in the molten steel fluid model; The oxygen mass transfer equation is combined with the flow pivot mapping model to construct an LSTM model. The flow pivot characteristics and oxygen concentration data output by the CFD solver are used as input and fed into the LSTM model multiple times to train and learn the mapping relationship between the flow pivot mapping model characteristics and oxygen content. The flow pivot mapping model features are used as the input layer, the oxygen concentration data change rate is used as the hidden layer, and the oxygen concentration is used as the output layer; The blast furnace molten steel flow data corresponding to the grid node coordinates of the molten steel fluid model is passed through the LSTM model to output the oxygen concentration of each grid node coordinate; The oxygen concentration of each grid node coordinate is marked one by one with the grid node coordinates in the molten steel fluid model, and the oxygen content distribution cloud map in the furnace is output through the CFD solver.

3. The fully automatic demoulding and sampling system for molten steel in a blast furnace according to claim 1, characterized in that: The process of obtaining the oxygen content at the initial optimal sampling point is: The oxygen content at the initial optimal sampling point can be obtained by obtaining the grid node coordinates corresponding to the initial optimal sampling point in the molten steel fluid model and obtaining the oxygen content corresponding to the coordinates in the oxygen content distribution cloud map in the furnace.

4. The fully automatic demoulding and sampling system for molten steel in a blast furnace according to claim 1, characterized in that: The process of determining whether to perform an initial optimal sampling point offset process based on the oxygen content at the initial optimal sampling point is as follows: Compare and analyze the oxygen content at the initial optimal sampling point with the standard oxygen content threshold; If the oxygen content at the initial optimal sampling point is less than or equal to the standard oxygen content threshold, it is determined that the initial optimal sampling point does not need to be offset; Otherwise, it is determined that the initial optimal sampling point needs to be offset.

5. The fully automatic demoulding and sampling system for molten steel in a blast furnace according to claim 1, characterized in that: Based on the oxygen content distribution cloud map in the furnace, the process of obtaining the oxygen-containing deflectable area is as follows: If it is determined that the initial optimal sampling point needs to be offset, the oxygen content distribution cloud map in the furnace is divided into several unit areas of equal area based on the oxygen content distribution cloud map in the furnace. If the oxygen content distribution cloud map in the furnace shows that the oxygen content in the unit area is less than or equal to the standard oxygen content threshold, then the unit area is an oxygen-containing offset area.

6. The fully automatic demoulding and sampling system for molten steel in a blast furnace according to claim 1, characterized in that: The process of determining the target offset area is: Based on any oxygen-containing deflectable region; Obtain the shortest distance between the initial optimal sampling point and the oxygen-containing deflectable area to obtain the deflection distance; The offset distance is ratioed to the total offset distance to obtain the offset distance ratio; Obtain the average oxygen content in the oxygen-containing deflectable area and calculate the ratio with the total average oxygen content to obtain the average oxygen content ratio; The offset distance ratio and the average oxygen content ratio are weighted and the output is the comprehensive offset assessment value. The oxygen-containing deflectable area corresponding to the minimum deflection comprehensive evaluation value is selected as the target deflection area.

7. The fully automatic demoulding and sampling system for molten steel in a blast furnace according to claim 1, characterized in that: The process of selecting target molten steel sampling samples is as follows: Gridding the target offset area, obtaining grid node coordinates, and obtaining multiple target sampling points to be determined; Sampling molten steel from a blast furnace at each target sampling point to be determined, to obtain molten steel sampling samples from a plurality of target sampling points to be determined; Obtain the actual sampling offset ratio of molten steel and the oxygen content ratio of molten steel sampling samples; The actual sampling offset ratio of the molten steel and the oxygen content ratio of the molten steel sampling sample are summed to obtain the sample selection value of the molten steel sampling sample; The molten steel sampling sample at the target sampling point to be determined corresponding to the minimum sample selection value is selected as the molten steel target sampling sample.

8. The fully automatic demoulding and sampling system for molten steel in a blast furnace according to claim 7, characterized in that: The process of obtaining the actual sampling offset ratio of molten steel and the oxygen content ratio of the molten steel sampling sample is as follows: Obtaining the initial optimal sampling point of the blast furnace molten steel when sampling each molten steel sampling sample, and calculating the position deviation with the target sampling point to be determined, to obtain the actual sampling offset of the molten steel sampling sample; Calculate the ratio of the actual sampling offset of the molten steel sample to the total actual sampling offset to obtain the actual sampling offset ratio of the molten steel; The oxygen content of the initial optimal sampling point of the blast furnace molten steel is obtained when the molten steel sampling sample is sampled, and the oxygen content ratio of the molten steel sampling sample is calculated by proportional calculation with the total oxygen content.

Images