Method and device for determining furnace type parameters of a gas-based shaft furnace

Abstract

This invention discloses a method and apparatus for determining the furnace type parameters of a gas-based vertical shaft furnace. The method includes: establishing a numerical simulation model of the gas-based vertical shaft furnace based on benchmark furnace type parameters; simulating multiple sets of preset furnace type parameters using the numerical simulation model to obtain performance indicators corresponding to each furnace type parameter; training a pre-established random forest model based on the furnace type parameters and their corresponding performance indicators to obtain a trained random forest model; and optimizing the furnace type parameters based on the trained random forest model and a pre-established objective function to obtain preferred furnace type parameters and their corresponding preferred performance indicators. This invention enables rapid prediction and collaborative optimization of the performance indicators of a gas-based vertical shaft furnace, reduces reliance on repetitive numerical simulations, thereby shortening the furnace type parameter optimization cycle and reducing computational costs.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a method and apparatus for determining the furnace type parameters of a gas-based vertical shaft furnace. Background Technology

[0002] The furnace structure of a gas-based vertical shaft furnace has a significant impact on the gas-solid two-phase flow, mass and heat transfer processes, and reduction reaction behavior within the furnace. Furnace parameters (including the height-to-diameter ratio of the reduction zone, the height of the cooling zone, and the inclination angle of the cooling zone) directly affect the gas flow distribution, the solid downward flow process, and the gas-solid contact state, thereby influencing the metallization rate, carbon content, and reducing gas utilization rate. Therefore, how to efficiently and quantitatively determine the key furnace parameters of a gas-based vertical shaft furnace under multiple operating conditions and parameters has become an urgent technical problem to be solved.

[0003] In existing technologies, numerical simulation methods based on computational fluid dynamics (CFD) are commonly used to optimize furnace parameters. For example, by constructing blast furnace geometric models with different throat-to-waist diameter ratios, and under conditions such as a fixed effective volume, multiphysics coupling calculations are used to iteratively solve the gas flow field, temperature field, and reaction process. The optimal furnace parameters are then selected based on indicators such as energy consumption, yield, and furnace internal conditions. This type of method can achieve quantitative analysis of multiphase flow and thermochemical behavior within the furnace to a certain extent, and has higher accuracy compared to traditional empirical methods.

[0004] However, the aforementioned methods typically optimize only single furnace parameters and often employ fixed effective volume settings during simulations, making it difficult to reflect the coupled effects of effective volume and operating conditions caused by variations in furnace parameters. Furthermore, these methods rely on repeated CFD simulations for each operating condition, resulting in high computational costs and long cycles. This makes it difficult to construct multi-condition databases and conduct data-driven rapid predictions and multi-parameter collaborative optimization, thus failing to meet the practical needs for efficient design and optimization of gas-based vertical shaft furnace parameters.

[0005] This section is intended to provide background or context for the embodiments of the invention set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section. Summary of the Invention

[0006] This invention provides a method for determining the furnace type parameters of a gas-based vertical shaft furnace, which solves the problems in the prior art that rely on a single numerical simulation, resulting in a long optimization cycle, difficulty in achieving collaborative optimization of multiple furnace type parameters, and inability to quantitatively analyze the impact of furnace type parameters on key performance indicators, thereby improving the efficiency and accuracy of furnace type parameter optimization.

[0007] The method for determining the furnace parameters of this gas-based vertical shaft furnace includes: A numerical simulation model of a gas-based vertical shaft furnace was established based on the parameters of the benchmark furnace type. The numerical simulation model is used to simulate multiple preset furnace parameters to obtain the performance indicators corresponding to each furnace parameter. The pre-established random forest model is trained based on the furnace type parameters and their corresponding performance indicators to obtain the trained random forest model. The furnace parameters are optimized based on the trained random forest model and the pre-established objective function to obtain the preferred furnace parameters and their corresponding preferred performance indicators.

[0008] In some embodiments, establishing a numerical simulation model of a gas-based vertical shaft furnace based on reference furnace type parameters includes: A geometric model of the gas-based vertical shaft furnace is established based on the reference furnace type parameters, and the computational domain is determined based on the geometric model. The computational domain is divided into networks to obtain multiple computational units. The operating parameters of the gas-based vertical shaft furnace are applied at the boundary positions of the computing unit and the computing domain, and gas-solid phase resistance equation, gas-solid phase heat transfer equation and gas-solid phase chemical reaction equation are established based on the operating parameters.

[0009] In some embodiments, the operating parameters include: the composition of the reducing gas and the composition of the cooling gas; the establishment of the gas-solid phase resistance equation, the gas-solid phase heat transfer equation, and the gas-solid phase chemical reaction equation based on the operating parameters includes: The gas-solid phase resistance equation is established based on the solid phase volume fraction, gas phase volume fraction, gas density, relative velocity between the gas and solid phases, solid particle diameter, gas viscosity, and gas-solid phase resistance. The gas-solid phase heat transfer equation is established based on the convective heat transfer coefficient, gas phase thermal conductivity, gas phase volume fraction, solid phase volume fraction, Nusselt number, and solid particle diameter. Based on the composition of the reducing gas and the composition of the cooling gas, respectively construct the iron oxide reduction reaction equation, carburizing reaction equation, reforming reaction equation and water-gas replacement reaction equation.

[0010] In some embodiments, training a pre-established random forest model based on the furnace type parameters and their corresponding performance indicators to obtain a trained random forest model includes: Set the hyperparameters of the random forest model; the hyperparameters include: number of decision trees, maximum number of features participating in the candidate when splitting each leaf node, maximum depth of the decision tree, minimum number of samples per leaf node, minimum number of samples required for leaf node re-split, and maximum number of leaf nodes, i.e., random number seed; The random forest model is trained based on the hyperparameters, the furnace type parameters, and their corresponding performance indicators to obtain the trained random forest model.

[0011] In some embodiments, optimizing the furnace parameters based on the trained random forest model and a pre-established objective function to obtain preferred furnace parameters and their corresponding preferred performance indicators includes: Set the value range and population size of the furnace type parameters, and initialize the population based on the value range; The trained random forest model is invoked to predict the individuals in the population, and the predicted values ​​of metallization rate, furnace charge temperature and carbon content corresponding to the individuals are obtained. The fitness value is determined based on the predicted values ​​of the metallization rate, the furnace charge temperature, and the carbon content, and the objective function. The population is updated based on the fitness value; The above steps are executed iteratively until the number of iterations reaches the preset number of iterations or the fitness value is less than the preset threshold, at which point the preferred furnace type parameters and their corresponding preferred performance indicators are output.

[0012] In some embodiments, the performance indicators include: metallization rate, furnace charge temperature, and carbon content; establishing the objective function includes: An objective function is constructed based on the metallization rate, the furnace charge temperature, the carbon content, and their corresponding weighting coefficients.

[0013] In some embodiments, updating the population based on the fitness value includes: Multiple target individuals are selected from the population based on their fitness values; The furnace type parameters of each target individual are recombined to generate multiple preferred individuals; The furnace type parameters of each of the preferred individuals are randomly perturbed according to a preset probability to form an updated population.

[0014] This invention also provides a device for determining the furnace type parameters of a gas-based vertical shaft furnace, which solves the problems in the prior art that rely on a single numerical simulation, resulting in a long optimization cycle, difficulty in achieving collaborative optimization of multiple furnace type parameters, and inability to quantitatively analyze the impact of furnace type parameters on key performance indicators, thereby improving the efficiency and accuracy of furnace type parameter optimization.

[0015] The device for determining the furnace parameters of the gas-based vertical shaft furnace includes: The numerical simulation model building module is used to establish a numerical simulation model of the gas-based vertical furnace based on the parameters of the benchmark furnace type. The simulation calculation module is used to simulate multiple preset furnace parameters through the numerical simulation model to obtain the performance indicators corresponding to each furnace parameter. The model training module is used to train the pre-established random forest model based on the furnace type parameters and their corresponding performance indicators to obtain the trained random forest model. The parameter optimization module is used to optimize the furnace parameters based on the trained random forest model and the pre-established objective function to obtain the preferred furnace parameters and their corresponding preferred performance indicators.

[0016] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described method for determining the furnace type parameters of a gas-based vertical shaft furnace.

[0017] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for determining the furnace type parameters of a gas-based vertical shaft furnace.

[0018] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described method for determining the furnace type parameters of a gas-based vertical shaft furnace.

[0019] The method and apparatus for determining furnace parameters of a gas-based vertical shaft furnace provided in this invention combine a numerical simulation model based on computational fluid dynamics with a random forest model. A simulation database is constructed based on multiple sets of furnace parameters, and the random forest model is then used to predict the furnace parameters, obtaining performance indicators such as metallization rate, carbon content, and charge temperature. This allows model prediction to replace repetitive numerical simulation calculations in the subsequent optimization of furnace parameters, significantly reducing computational complexity and time costs. By collaboratively optimizing multiple furnace parameters under objective function constraints, global search and comprehensive adjustment of key parameters are achieved, improving the efficiency and accuracy of furnace parameter determination. Furthermore, this method can quantitatively analyze the impact of each furnace parameter on performance indicators, providing a basis for furnace design and operating condition adjustment, thereby enhancing the applicability and stability of the method under complex operating conditions. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating a method for determining the furnace type parameters of a gas-based vertical shaft furnace in one embodiment of the present invention. Figure 2 This is a flowchart illustrating the method for determining the furnace type parameters of a gas-based vertical shaft furnace in another embodiment of the present invention; Figure 3This is a flowchart illustrating the method for determining the furnace type parameters of a gas-based vertical shaft furnace in another embodiment of the present invention; Figure 4 This is a flowchart illustrating the method for determining the furnace type parameters of a gas-based vertical shaft furnace in another embodiment of the present invention; Figure 5 This is a flowchart illustrating the method for determining the furnace type parameters of a gas-based vertical shaft furnace in another embodiment of the present invention; Figure 6 This is a flowchart illustrating the method for determining the furnace type parameters of a gas-based vertical shaft furnace in another embodiment of the present invention; Figure 7 This is a schematic diagram of the furnace type parameter determination device for a gas-based vertical shaft furnace in an embodiment of the present invention; Figure 8 This is a schematic diagram of the physical structure of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments and their descriptions are used to explain the present invention, but are not intended to limit the present invention. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other. The acquisition, storage, use, and processing of data in the technical solutions of this application all comply with relevant laws and regulations. The user information in the embodiments of this application is obtained through legal and compliant means, and the acquisition, storage, use, and processing of user information have been authorized and agreed upon by the customer.

[0022] To facilitate understanding of the technical solution provided in this application, the relevant content of the technical solution in this application will be explained below.

[0023] To address the following problems in the optimization of gas-based vertical shaft furnace parameters in existing technologies: reliance on single computational fluid dynamics (CFD) simulations leads to long computation cycles, high design costs, difficulty in achieving multi-parameter collaborative optimization, and an inability to quantitatively analyze the impact of various furnace parameters on key performance indicators such as metallization rate, carbon content, and charge temperature, this invention provides a method for determining gas-based vertical shaft furnace parameters based on a combination of computational fluid dynamics and machine learning. By establishing a numerical simulation model of the gas-based vertical shaft furnace to simulate multiple sets of furnace parameters, a simulation database of furnace parameters and performance indicators is constructed. A random forest model is then trained based on this simulation database to establish the mapping relationship between furnace parameters and performance indicators. By combining an objective function and intelligent optimization algorithms to perform global search and optimization of furnace parameters, the efficient determination of key furnace parameters is achieved, improving furnace design efficiency and optimization accuracy, and enabling quantitative evaluation of the impact of various furnace parameters on the production performance of the vertical shaft furnace.

[0024] This invention provides a method for determining the furnace type parameters of a gas-based vertical shaft furnace. For example... Figure 1 As shown, the method for determining the furnace type parameters includes steps 101 to 104.

[0025] Step 101: Establish a numerical simulation model of the gas-based vertical shaft furnace based on the reference furnace type parameters.

[0026] Step 102: Simulate multiple sets of preset furnace parameters using a numerical simulation model to obtain the performance indicators corresponding to each furnace parameter.

[0027] Step 103: Train the pre-established random forest model based on the furnace type parameters and their corresponding performance indicators to obtain the trained random forest model.

[0028] Step 104: Optimize the furnace parameters based on the trained random forest model and the pre-established objective function to obtain the preferred furnace parameters and their corresponding preferred performance indicators.

[0029] According to the above embodiments, by combining a computational fluid dynamics-based numerical simulation model with a random forest model, and constructing a simulation database based on multiple sets of furnace parameters, the random forest model is then used to predict furnace parameters, obtaining performance indicators such as metallization rate, carbon content, and charge temperature. This allows model prediction to replace repetitive numerical simulation calculations in subsequent furnace parameter optimization, significantly reducing computational complexity and time costs. By collaboratively optimizing multiple furnace parameters under objective function constraints, global search and comprehensive adjustment of key parameters are achieved, improving the efficiency and accuracy of furnace parameter determination. Furthermore, this method can quantitatively analyze the impact of each furnace parameter on performance indicators, providing a basis for furnace design and operating condition adjustment, thereby enhancing the applicability and stability of the method under complex operating conditions.

[0030] In some embodiments, such as Figure 2 As shown, step 101 includes steps 201 to 203.

[0031] Step 201: Establish a geometric model of the gas-based vertical shaft furnace based on the reference furnace type parameters, and determine the computational domain based on the geometric model.

[0032] Step 202: Divide the computational domain into a network to obtain multiple computational units.

[0033] Step 203: Apply the operating parameters of the gas-based vertical shaft furnace to the boundary positions of the calculation unit and the calculation domain, and establish the gas-solid phase resistance equation, gas-solid phase heat transfer equation and gas-solid phase chemical reaction equation based on the operating parameters.

[0034] In this embodiment of the invention, a numerical simulation model is established to describe the energy and mass transfer process of the gas-solid two-phase system in a gas-based vertical shaft furnace, so as to characterize the gas-solid phase flow behavior, heat transfer process and reaction process between multi-component gas and solid in the furnace.

[0035] Specifically, based on the reference furnace parameters, a three-dimensional geometric model of the gas-based vertical shaft furnace was established using the 3D modeling software SolidWorks. This 3D geometric model includes a reduction zone, a transition zone, and a cooling zone. The internal interconnected spaces of the reduction zone, transition zone, and cooling zone were extracted, and defined as a single computational domain or multiple interconnected sub-computational domains according to operating requirements. Based on this, the boundary types of this computational domain were identified, determining the reducing gas inlet boundary, cooling gas inlet boundary, furnace top outlet boundary, furnace wall boundary, and symmetry boundary for subsequent boundary condition settings. During the establishment of the 3D geometric model, the structural parameters of each region were initialized. Specifically, the height-to-diameter ratio of the reduction zone was set to 1.8, the height of the cooling zone was set to 6 m, and the inclination angle of the cooling zone was set to 15°, serving as input parameters for the reference furnace type; this invention is not limited to these parameters.

[0036] After the 3D geometric model is constructed, it is exported as a STEP format file and imported into the mesh generation software ANSYS Meshing for spatial discretization. Preferably, an unstructured tetrahedral mesh is used to divide the computational domain, and local mesh refinement is performed at the reducing gas inlet boundary, cooling gas inlet boundary, furnace top outlet boundary, furnace wall boundary, symmetric boundaries, and regions with significant gas-solid interaction to improve computational accuracy and numerical stability. Simultaneously, the mesh quality is checked to ensure that mesh orthogonality and element distortion rate meet the requirements of numerical computation.

[0037] Subsequently, the operating parameters of the gas-based vertical shaft furnace were set according to the production conditions, and these operating parameters were input into the numerical simulation model as boundary conditions and initial conditions. The operating parameters include: reducing gas temperature, reducing gas flow rate, reducing gas composition and its proportion, cooling gas temperature, cooling gas flow rate, and cooling gas composition and its proportion, etc.

[0038] For example, the reducing gas inlet is set as a velocity inlet or mass flow inlet boundary condition, the furnace top or outlet area is set as a pressure outlet boundary condition, and the furnace wall is set as an adiabatic wall or a thermal boundary condition with a given heat transfer coefficient.

[0039] The reducing gas temperature is set to 975℃, and the flow rate is set to 1100 Nm³. 3 / ton of pellets, with the following composition and volume fractions: H2 52.5%, CO 33%, CO2 3%, H2O 6%, CH4 4%, and N2 1.5%. The cooling gas temperature is set to 30℃, and the flow rate is set to 110 Nm³. 3 The pellets per ton, with the following composition and volume fraction: CH4 97%, N2 3%, are not limited thereto in this invention. The above parameters are used to describe the gas flow, heat exchange, and reaction processes within the furnace, thus providing a data foundation for subsequent multi-condition simulation calculations.

[0040] According to the above embodiments, by establishing a geometric model and determining the computational domain based on the reference furnace parameters, the gas-solid two-phase flow and reaction space inside the gas-based vertical shaft furnace can be accurately characterized. The computational domain is meshed and discretized into multiple computational cells, providing a computational basis for the numerical solution of the multiphysics equations. Operating parameters are applied to the computational cells and boundary locations, and gas-solid interphase resistance, heat transfer, and chemical reaction models are introduced, enabling the numerical simulation model to comprehensively reflect the flow, heat transfer, and reaction processes, thereby improving the accuracy and stability of the numerical simulation results.

[0041] In some embodiments, the operating parameters include: the composition of the reducing gas and the composition of the cooling gas. For example... Figure 3 As shown, step 203 includes steps 301 to 303.

[0042] Step 301: Establish the gas-solid phase resistance equation based on the solid phase volume fraction, gas phase volume fraction, gas density, relative velocity between the gas and solid phases, solid particle diameter, gas viscosity, and gas-solid phase resistance.

[0043] Step 302: Establish the gas-solid phase heat transfer equation based on the convective heat transfer coefficient, gas phase thermal conductivity, gas phase volume fraction, solid phase volume fraction, Nusselt number, and solid particle diameter.

[0044] Step 303: Construct the iron oxide reduction reaction equation, carburizing reaction equation, reforming reaction equation, and water-gas replacement reaction equation according to the composition of the reducing gas and the cooling gas.

[0045] In this embodiment of the invention, a gas-solid interphase drag equation is set in the CFD simulation software to characterize the momentum exchange process between the gas and solid phases. Preferably, the Ergun equation is used to calculate the momentum exchange term between the gas and solid phases, and its gas-solid interphase drag equation expression is:

[0046] in, Indicates the volume fraction of the solid phase. Indicates the gas phase volume fraction. Indicates the dynamic viscosity of the gas. Indicates the diameter of solid particles. Indicates gas density, Represents the solid phase velocity vector. Represents the gas phase velocity vector. This represents the gas-solid phase drag. This gas-solid phase drag equation is introduced as a source term into the momentum conservation equation to simulate the coupled flow behavior of the gas-solid two phases.

[0047] Furthermore, a gas-solid phase heat transfer model is set up in the CFD simulation software to describe the convective heat transfer process between the gas and solid phases. Preferably, the gas-solid phase heat transfer coefficient is calculated using the following expression:

[0048] in, Indicates thermal conductivity. Indicates the gas phase volume fraction. Indicates the volume fraction of the solid phase. Represents the Nusel number, Let represent the diameter of the solid particles, and h represent the gas-solid interphase convective heat transfer coefficient. The gas-solid interphase heat transfer coefficient is used to establish the energy exchange relationship between the gas and solid phases and is introduced as a source term into the energy conservation equation to simulate the gas-solid interphase heat transfer process.

[0049] A gas-solid phase chemical reaction equation was set in the CFD simulation software to describe the chemical reaction process between gaseous reducing gas and solid charge in a gas-based vertical shaft furnace and its impact on mass transport and energy transfer.

[0050] Specifically, the gas-solid phase chemical reactions include reduction reaction, carburizing reaction, reforming reaction and water-gas replacement reaction, and their corresponding chemical reaction equations and reaction rates are shown in Table 1.

[0051] Table 1

[0052] Table 1 above shows the stoichiometric relationships of each chemical reaction and the corresponding chemical reaction rates. Among them, the reaction rates... The unit is , is used to express the rate of each chemical reaction per unit volume.

[0053] Based on the aforementioned chemical reaction equations, a reaction heat source term is further constructed to achieve coupled calculation of the chemical reaction process and the temperature field. This term is constructed by multiplying the reaction enthalpy by the reaction rate. The reaction enthalpy is calculated using the enthalpy calculation formula and the temperature correction equation.

[0054] Specifically, the enthalpy of each substance participating in the chemical reaction at temperature T is calculated based on the temperature correction equation, and its expression is as follows:

[0055] in, This represents the enthalpy of the i-th substance at temperature T. This indicates the enthalpy of the substance at standard conditions (298 K). This represents the isobaric specific heat capacity of the substance at temperature T, where T represents the absolute temperature.

[0056] Based on the enthalpy values ​​of each substance, the enthalpy of each chemical reaction at temperature T is determined according to the enthalpy calculation formula, which is as follows:

[0057] in, This represents the enthalpy of the nth chemical reaction at temperature T. This represents the enthalpy of the product at temperature T. This indicates the enthalpy of the reactant at temperature T.

[0058] Furthermore, the enthalpy of reaction was calculated for the carburizing reaction and the carbon monoxide dismutation reaction (corresponding to reactions 9 and 10 in Table 1). and .

[0059] Based on the reaction rates and corresponding enthalpies of each chemical reaction, the reaction heat source term for the gas-based vertical shaft furnace is established, and its expression is as follows:

[0060] in, The term representing the heat source of reaction per unit volume is used as a source term in the energy conservation equation. The weighting coefficients represent the weighting factors for oxygen-related reactions. This represents the reaction rate of the nth chemical reaction. This represents the enthalpy of the corresponding chemical reaction at temperature T. This represents the weighting coefficient for carbon-related reactions. These represent the reaction rates of the 9th and 10th chemical reactions, respectively. These represent the enthalpy of the 9th and 10th chemical reactions, respectively.

[0061] The aforementioned reaction heat source term is introduced as a source term into the energy conservation equation, thereby realizing the coupled calculation between the chemical reaction process and the temperature field to simulate the thermodynamic behavior of various reaction systems in a gas-based vertical shaft furnace.

[0062] According to the above embodiments, by introducing the components of reducing gas and cooling gas into the operating parameters, and establishing gas-solid phase resistance equations, heat transfer equations, and chemical reaction equations respectively, a synergistic description of gas-solid two-phase flow, heat transfer, and reaction processes can be achieved in the same numerical model. This improves the numerical simulation model's ability to characterize the multi-physics coupling behavior inside a gas-based vertical shaft furnace. By incorporating the reaction mechanism and physical property parameters into the calculation process, it helps to accurately reflect the influence of furnace type parameter changes on flow field distribution, temperature field, and composition evolution, thus improving the reliability and consistency of the simulation results.

[0063] In some embodiments, step 102 specifically includes: based on the numerical simulation model of the gas-based vertical shaft furnace established in step 101, performing numerical modeling and simulation calculations for multiple preset sets of furnace parameters. That is, by changing the furnace structure parameters and repeatedly executing step 101, numerical simulations are performed on the operating conditions corresponding to different furnace structures to obtain the performance indicators corresponding to each set of furnace parameters. The furnace parameters include: reduction zone height, reduction zone diameter ratio, transition zone height, cooling zone height, and cooling zone inclination angle.

[0064] After completing the above multi-condition numerical simulations, the simulation results for each group were stored. Specifically, key performance indicators at the outlet boundary of the gas-based vertical shaft furnace were extracted from the calculation results. The outlet boundary is the exit position where Direct Reduced Iron (DRI) exits from the furnace. Key performance indicators include metallization rate, carbon content, and furnace charge temperature.

[0065] Preferably, key performance indicators in the outlet section or outlet area are spatially averaged to obtain representative average metallization rate, average carbon content, and average charge temperature, thereby reducing the impact of local fluctuations on the results and improving the stability and comparability of the data.

[0066] Furthermore, a simulation database for gas-based vertical shaft furnaces was constructed using the furnace type parameters as input variables and the corresponding key performance indicators as output variables. This simulation database is used to characterize the mapping relationship between furnace type structural parameters and performance indicators, and to provide a data foundation for the subsequent training of machine learning models.

[0067] For example, each set of furnace parameters and its corresponding key performance indicators constitutes a sample data record. For instance, when the furnace parameters are [10, 2, 2.5, 9, 15], the corresponding key performance indicators are metallization rate of 92.8%, carbon content of 1.35%, and DRI temperature of 380℃. By summarizing the simulation results corresponding to multiple sets of different furnace parameters, a dataset containing multiple samples is formed, thereby establishing a complete simulation database for gas-based vertical shaft furnaces.

[0068] In some embodiments, such as Figure 4 As shown, step 103 includes steps 401 to 402.

[0069] Step 401: Set the hyperparameters of the random forest model. Hyperparameters include: number of decision trees, maximum number of features participating in the candidate selection when splitting at each leaf node, maximum depth of the decision tree, minimum number of samples per leaf node, minimum number of samples required for further splitting of a leaf node, and maximum number of leaf nodes, i.e., the random number seed.

[0070] Step 402: Train the random forest model based on the hyperparameters, furnace type parameters and their corresponding performance indicators to obtain the trained random forest model.

[0071] In this embodiment of the invention, the machine learning model used can be a Random Forest (RF) model, which is used to establish a nonlinear mapping relationship between furnace parameters and key performance indicators.

[0072] Specifically, the gas-based vertical shaft furnace simulation database constructed in step 102 is divided into a training set and a test set according to a preset ratio. Preferably, the ratio of the training set to the test set is 8:2.

[0073] The random forest model is trained using the training set, and its performance is evaluated using the test set. Evaluation metrics include: goodness of fit R0. 2Mean Absolute Error (MAE) and Mean Squared Error (MSE) are used to measure the consistency and error level between model predictions and simulation results.

[0074] Furthermore, the hyperparameters of the random forest model are set and adjusted to improve its prediction accuracy and stability. These hyperparameters include: the number of decision trees (n_estimators), the maximum number of features participating in each node split (max_features), the maximum depth of the decision trees (max_depth), the minimum number of samples per leaf node (min_samples_leaf), the minimum number of samples required for further splitting of internal nodes (min_samples_split), the maximum number of leaf nodes (max_leaf_nodes), and the random number seed (random_state). By appropriately configuring these hyperparameters, the random forest model can achieve good generalization ability while avoiding overfitting.

[0075] During model training, the Random Forest Regression interface in the Python machine learning library is called to execute the training program, fitting the input furnace parameters with the corresponding key performance indicators to obtain the trained Random Forest model. After model training is complete, the model prediction results are post-processed to generate a comparison file between the original simulation data and the model prediction results. This comparison file is stored in CSV format and includes the furnace parameters, corresponding simulation performance indicators, and model prediction values ​​to facilitate analysis of the model fitting effect.

[0076] In addition, the trained random forest model is saved as a serialized file, preferably in pkl format, so that the model can be directly called for performance prediction without repeated training during subsequent furnace parameter optimization, thereby improving the overall computational efficiency.

[0077] According to the above embodiments, by reasonably setting the hyperparameters of the random forest model and training the random forest model based on the furnace type parameters and their corresponding performance indicators, the fitting ability of the random forest model to complex nonlinear relationships can be effectively improved, enabling it to accurately represent the mapping relationship between furnace type parameters and key performance indicators such as metallization rate, carbon content, and furnace charge temperature. By adjusting the hyperparameters, a balance can be achieved between model accuracy and generalization ability, reducing the risk of overfitting and improving the stability and reliability of model predictions.

[0078] In some embodiments, such as Figure 5 As shown, step 104 includes steps 501 to 505.

[0079] Step 501: Set the range of values ​​for furnace type parameters and population size, and initialize the population based on the range of values.

[0080] Step 502: Call the trained random forest model to predict the individuals in the population and obtain the predicted values ​​of metallization rate, furnace charge temperature and carbon content for each individual.

[0081] Step 503: Determine the fitness value based on the predicted values ​​of metallization rate, furnace charge temperature and carbon content, and the objective function.

[0082] Step 504: Update the population based on fitness values.

[0083] Step 505: Iteratively execute the above steps until the number of iterations reaches the preset number of iterations or the fitness value is less than the preset threshold, then output the optimal furnace type parameters and their corresponding optimal performance indicators.

[0084] In this embodiment of the invention, after obtaining the trained Random Forest (RF) model, a model interpretation method based on Shapley values ​​is further introduced to calculate the contribution of each forest type parameter to the model prediction results. Preferably, the TreeSHAP interpretation algorithm is used to adapt to the tree structure of the Random Forest model, thereby achieving efficient calculation of feature contribution values.

[0085] Specifically, all or part of the data in the simulation database is selected as background data. The expected value of the random forest model's prediction results is obtained by statistically analyzing the model prediction results of the background data, and this expected value is used as a reference benchmark. Based on the reference benchmark, the trained random forest model is used to calculate the corresponding SHAP value for the furnace type parameters of each sample in the background data, so as to characterize the marginal contribution of each furnace type parameter in a single sample.

[0086] Statistical analysis was performed on the SHAP values ​​of all samples. Preferably, the average absolute value of the SHAP values ​​of each furnace type parameter was used as the global influence weight index to measure the overall influence of each furnace type parameter on the target performance indicators. The target performance indicators include the metallization rate, carbon content, and charge temperature of Direct Reduced Iron (DRI). Based on the global influence weight, the parameters of each furnace type were ranked and visualized using bar charts and other methods, thus completing the furnace type parameter importance analysis of the random forest model.

[0087] Based on the above analysis results, furnace parameters that have a significant impact on key performance indicators can be identified and determined as priority furnace parameters for optimization, so as to provide a basis for subsequent furnace parameter optimization.

[0088] According to the above embodiments, by introducing a swarm intelligence optimization strategy based on a set range of furnace parameters, and combining this with a trained random forest model to quickly predict and evaluate the fitness of individuals in the population, efficient search and optimization of furnace parameters can be achieved without repeating costly numerical simulations. By iteratively updating the population and gradually approaching the optimal solution, the combination of furnace parameters under multi-objective constraints is synergistically optimized, thereby improving the efficiency and stability of parameter optimization, enhancing the globality and reliability of the optimization results, and providing a fast and feasible parameter decision-making basis for the structural design of gas-based vertical shaft furnaces.

[0089] In some embodiments, the machine learning model can be rebuilt when the simulation database or machine learning model is updated.

[0090] Specifically, when it is necessary to update the simulation database of the gas-based vertical shaft furnace, steps 101 to 103 can be repeated to retrain the random forest model based on the updated simulation database.

[0091] When the random forest model needs to be updated, step 103 can be re-executed by adjusting the model hyperparameters. These hyperparameters include: number of decision trees, maximum number of features, maximum tree depth, minimum number of samples per leaf node, minimum number of samples required for internal node repartitioning, maximum number of leaf nodes, and random number seed.

[0092] When it is necessary to change the machine learning model, other types of regression models can be selected as alternatives. For example, backpropagation neural network (BP neural network) models or CatBoost models can be used. After resetting the corresponding hyperparameters, the model training can be completed to establish a new machine learning model for the gas-based vertical shaft furnace.

[0093] In some embodiments, performance indicators include: metallization rate, charge temperature, and carbon content. Establishing the objective function specifically includes: constructing an objective function based on the metallization rate, charge temperature, carbon content, and their corresponding weighting coefficients.

[0094] In this embodiment of the invention, one or more optimization objectives are set, and an objective function is constructed based on these objectives. An intelligent optimization algorithm is then used to optimize and search for the furnace parameters of the gas-based vertical shaft furnace, thereby obtaining a preferred combination of key furnace parameters. The optimization objectives include: increasing the metallization rate of direct reduced iron (DRI), increasing the carbon content, and decreasing the furnace charge temperature.

[0095] Specifically, based on the above optimization objective, the objective function is constructed, and its expression is as follows:

[0096] in, Describe the objective function. The furnace type parameters include structural parameters such as the height of the reduction zone, the diameter ratio of the reduction zone, the height of the transition zone, the height of the cooling zone, and the inclination angle of the cooling zone. This represents the metallization rate predicted based on the trained random forest model. This represents the carbon content predicted based on the random forest model. This represents the furnace charge temperature predicted based on the random forest model. , and These are the weighting coefficients corresponding to each performance index, used to adjust the importance of different optimization objectives in the overall optimization process.

[0097] According to the above embodiments, by optimizing the objective function, the furnace parameters can be synergistically optimized under multi-objective constraints, thereby obtaining the furnace parameter combination with the best overall performance.

[0098] In some embodiments, such as Figure 6 As shown, step 504 includes steps 601 to 603.

[0099] Step 601: Select multiple target individuals from the population based on their fitness values.

[0100] Step 602: Reorganize the furnace type parameters of each target individual to generate multiple preferred individuals.

[0101] Step 603: Randomly perturb the furnace type parameters of each preferred individual according to a preset probability to form an updated population.

[0102] In this embodiment of the invention, a range of values ​​is set for the furnace type parameters to construct a search space for optimization variables. Specifically, the height of the reduction zone... The value range is set to The height-to-diameter ratio of the reduction zone The value range is set to Cooling zone height The value range is set to Cooling zone tilt angle The value range is set to The value ranges of the above parameters together constitute the search space for furnace type parameter optimization.

[0103] Within the search space, a population-based intelligent optimization algorithm is used to generate an initial population. Specifically, multiple individuals are randomly sampled within the parameter value range to generate each individual. This represents a combination of furnace parameters, expressed in the following form:

[0104] Where i represents the individual ID. Preferably, the population size is set to N, for example, 50 individuals.

[0105] For each individual in the population The trained random forest model is called to predict performance and obtain the corresponding metallization. Carbon content and furnace charge temperature Then, the predicted results are substituted into the objective function to calculate the fitness value of the corresponding individual. To evaluate the overall performance of this group of furnace parameters.

[0106] Based on fitness values, a selection operation is performed on individuals in the population to identify target individuals with high fitness. A crossover operation is then performed on these target individuals, recombining their parameters to generate new individuals. Simultaneously, a mutation operation is performed on the parameters of the newly generated individuals with a preset probability (e.g., 0.08), randomly perturbing some parameters to enhance population diversity and prevent the algorithm from getting trapped in local optima. After selection, crossover, and mutation operations, a new generation of the population is formed.

[0107] Repeat the above steps until a preset termination condition is met. Termination conditions include: reaching the maximum number of iterations (e.g., 20 generations), or the fitness change being less than a preset threshold for several consecutive generations. When the termination condition is met, output the individual with the best fitness as the optimal furnace parameter combination and provide the corresponding predicted performance index.

[0108] According to the above embodiments, this invention combines computational fluid dynamics numerical simulation with machine learning (ML) data-driven methods. While ensuring the accuracy of the physical mechanisms, it utilizes machine learning models to replace repetitive numerical simulation calculations, thereby shortening the optimization cycle of key furnace parameters and achieving rapid prediction and optimization under multiple furnace parameter conditions, thus reducing the design and analysis costs of gas-based vertical shaft furnaces. This method has good scalability, enabling coordinated adjustment of multiple furnace parameters within a unified framework. Furthermore, it can rapidly predict and optimize corresponding operating conditions based on different design requirements such as the effective volume of the gas-based vertical shaft furnace and the target metallization rate to obtain a matching combination of furnace parameters. By establishing a mapping relationship between furnace parameters and key performance indicators, this invention achieves quantitative analysis of the impact of furnace parameter changes on the production performance of gas-based vertical shaft furnaces, overcoming the shortcomings of existing technologies that rely on engineering experience and are difficult to quantify and evaluate. This makes the furnace design process more quantifiable and verifiable.

[0109] This application provides a device for determining the furnace type parameters of a gas-based vertical shaft furnace, applied to the aforementioned method for determining the furnace type parameters of a gas-based vertical shaft furnace. This device and the method described in one embodiment of this application are based on the same inventive concept and solve similar problems. Therefore, the implementation of the device is the same as the method described in one embodiment of this application, and repetitions will not be repeated. The terms "unit" or "module" used below refer to a combination of software and / or hardware that performs a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0110] like Figure 7 As shown, the furnace type parameter determination device 700 for the gas-based vertical shaft furnace includes: The numerical simulation model construction module 701 is used to establish a numerical simulation model of a gas-based vertical furnace based on the parameters of the reference furnace type.

[0111] The simulation calculation module 702 is used to simulate multiple preset furnace parameters through a numerical simulation model to obtain the performance indicators corresponding to each furnace parameter.

[0112] The model training module 703 is used to train a pre-established random forest model based on furnace type parameters and their corresponding performance indicators to obtain the trained random forest model.

[0113] The parameter optimization module 704 is used to optimize the furnace parameters based on the trained random forest model and the pre-established objective function to obtain the preferred furnace parameters and their corresponding preferred performance indicators.

[0114] In some embodiments, the numerical simulation model building module includes: The geometric modeling submodule is used to establish a geometric model of the gas-based vertical shaft furnace based on the reference furnace type parameters, and to determine the computational domain based on the geometric model.

[0115] The meshing submodule is used to partition the computational domain into a network, resulting in multiple computational units.

[0116] The physical model construction submodule is used to apply the operating parameters of the gas-based vertical shaft furnace at the boundary positions of the computing unit and computing domain, and to establish the gas-solid phase resistance equation, gas-solid phase heat transfer equation and gas-solid phase chemical reaction equation based on the operating parameters.

[0117] In some embodiments, the operating parameters include: the composition of the reducing gas and the composition of the cooling gas. The physical model construction submodule includes: The drag equation construction unit is used to establish the gas-solid phase drag equation based on the solid phase volume fraction, gas phase volume fraction, gas density, relative velocity between the gas and solid phases, solid particle diameter, gas viscosity, and gas-solid phase drag.

[0118] The heat transfer equation construction unit is used to establish the gas-solid phase heat transfer equation based on the convective heat transfer coefficient, gas phase thermal conductivity, gas phase volume fraction, solid phase volume fraction, Nusselt number, and solid particle diameter.

[0119] The chemical reaction model building unit is used to construct iron oxide reduction reaction equations, carburizing reaction equations, reforming reaction equations, and water-gas replacement reaction equations based on the composition of the reducing gas and the cooling gas.

[0120] In some embodiments, the model training module includes: The hyperparameter setting submodule is used to set the hyperparameters of the random forest model. Hyperparameters include: number of decision trees, maximum number of features participating in the candidate selection when splitting at each leaf node, maximum depth of the decision tree, minimum number of samples per leaf node, minimum number of samples required for further splitting of a leaf node, and maximum number of leaf nodes, i.e., the random number seed.

[0121] The model training submodule is used to train the random forest model based on hyperparameters, furnace type parameters and their corresponding performance indicators, and obtain the trained random forest model.

[0122] In some embodiments, the parameter optimization module includes: The population initialization submodule is used to set the range of values ​​for furnace type parameters and the population size, and to initialize the population based on the range of values.

[0123] The model prediction submodule is used to call the trained random forest model to predict the individuals in the population and obtain the predicted values ​​of metallization rate, furnace charge temperature and carbon content for each individual.

[0124] The fitness calculation submodule is used to determine the fitness value based on the predicted values ​​of metallization rate, furnace charge temperature, and carbon content, and the objective function.

[0125] The population update submodule is used to update the population based on fitness values.

[0126] The iterative calculation submodule is used to iteratively execute the above steps until the number of iterations reaches the preset number of iterations or the fitness value is less than the preset threshold, and then outputs the optimal furnace type parameters and their corresponding optimal performance indicators.

[0127] In some embodiments, the apparatus further includes an objective function construction module. The objective function construction module is used to construct an objective function based on the metallization rate, furnace charge temperature, carbon content, and their corresponding weighting coefficients.

[0128] In some embodiments, the population update submodule includes: Individual selection unit, used to select multiple target individuals from the population based on fitness values.

[0129] The parameter recombination unit is used to recombine the furnace parameters of each target individual to generate multiple preferred individuals.

[0130] The mutation update unit is used to randomly perturb the furnace type parameters of each preferred individual according to a preset probability to form an updated population.

[0131] Figure 8 This is a schematic diagram of the physical structure of a computer device provided in an embodiment of the present invention, such as... Figure 8 As shown, the computer device includes a processor 801, a memory 802, and a bus 803.

[0132] The processor 801 and the memory 802 communicate with each other via the bus 803.

[0133] The processor 801 is used to call program instructions in the memory 802 to execute the methods provided in the above-described method embodiments.

[0134] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for determining the furnace type parameters of a gas-based vertical shaft furnace.

[0135] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described method for determining the furnace type parameters of a gas-based vertical shaft furnace.

[0136] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0137] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0138] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0139] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0140] In the description of this specification, the references to terms such as "an embodiment," "a specific embodiment," "some embodiments," "for example," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0141] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for determining the furnace type parameters of a gas-based vertical shaft furnace, characterized in that, include: A numerical simulation model of a gas-based vertical shaft furnace was established based on the parameters of the benchmark furnace type. The numerical simulation model is used to simulate multiple sets of preset furnace parameters to obtain the performance indicators corresponding to each furnace parameter. The pre-established random forest model is trained based on the furnace type parameters and their corresponding performance indicators to obtain the trained random forest model. The furnace parameters are optimized based on the trained random forest model and the pre-established objective function to obtain the preferred furnace parameters and their corresponding preferred performance indicators.

2. The method according to claim 1, characterized in that, The numerical simulation model of the gas-based vertical shaft furnace established based on the reference furnace type parameters includes: A geometric model of the gas-based vertical shaft furnace is established based on the reference furnace type parameters, and the computational domain is determined based on the geometric model. The computational domain is divided into multiple computational units. The operating parameters of the gas-based vertical shaft furnace are applied at the boundary positions of the computing unit and the computing domain, and gas-solid phase resistance equation, gas-solid phase heat transfer equation and gas-solid phase chemical reaction equation are established based on the operating parameters.

3. The method according to claim 2, characterized in that, The operating parameters include: the composition of the reducing gas and the composition of the cooling gas; the establishment of the gas-solid phase resistance equation, the gas-solid phase heat transfer equation, and the gas-solid phase chemical reaction equation based on the operating parameters includes: The gas-solid phase resistance equation is established based on the solid phase volume fraction, gas phase volume fraction, gas density, relative velocity between the gas and solid phases, solid particle diameter, gas viscosity, and gas-solid phase resistance. The gas-solid phase heat transfer equation is established based on the convective heat transfer coefficient, gas phase thermal conductivity, gas phase volume fraction, solid phase volume fraction, Nusselt number, and solid particle diameter. Based on the composition of the reducing gas and the composition of the cooling gas, respectively construct the iron oxide reduction reaction equation, carburizing reaction equation, reforming reaction equation and water-gas replacement reaction equation.

4. The method according to claim 1, characterized in that, The process of training a pre-established random forest model based on the furnace type parameters and their corresponding performance indicators to obtain a trained random forest model includes: Set the hyperparameters of the random forest model; the hyperparameters include: number of decision trees, maximum number of features participating in the candidate when splitting each leaf node, maximum depth of the decision tree, minimum number of samples per leaf node, minimum number of samples required for leaf node re-split, and maximum number of leaf nodes, i.e., random number seed; The random forest model is trained based on the hyperparameters, the furnace type parameters, and their corresponding performance indicators to obtain the trained random forest model.

5. The method according to claim 1, characterized in that, The optimization of the furnace parameters based on the trained random forest model and the pre-established objective function to obtain the preferred furnace parameters and their corresponding preferred performance indicators includes: Set the value range and population size of the furnace type parameters, and initialize the population based on the value range; The trained random forest model is invoked to predict the individuals in the population, and the predicted values ​​of metallization rate, furnace charge temperature and carbon content corresponding to the individuals are obtained. The fitness value is determined based on the predicted values ​​of the metallization rate, the furnace charge temperature, and the carbon content, and the objective function. The population is updated based on the fitness value; The above steps are executed iteratively until the number of iterations reaches the preset number of iterations or the fitness value is less than the preset threshold, at which point the preferred furnace type parameters and their corresponding preferred performance indicators are output.

6. The method according to claim 5, characterized in that, The performance indicators include: metallization rate, furnace charge temperature, and carbon content; the objective function includes: An objective function is constructed based on the metallization rate, the furnace charge temperature, the carbon content, and their corresponding weighting coefficients.

7. The method according to claim 5, characterized in that, The update of the population based on the fitness value includes: Multiple target individuals are selected from the population based on their fitness values; The furnace type parameters of each target individual are recombined to generate multiple preferred individuals; The furnace type parameters of each of the preferred individuals are randomly perturbed according to a preset probability to form an updated population.

8. A device for determining the furnace type parameters of a gas-based vertical shaft furnace, characterized in that, include: The numerical simulation model building module is used to establish a numerical simulation model of the gas-based vertical furnace based on the parameters of the benchmark furnace type. The simulation calculation module is used to simulate multiple preset furnace parameters through the numerical simulation model to obtain the performance indicators corresponding to each furnace parameter. The model training module is used to train the pre-established random forest model based on the furnace type parameters and their corresponding performance indicators to obtain the trained random forest model. The parameter optimization module is used to optimize the furnace parameters based on the trained random forest model and the pre-established objective function to obtain the preferred furnace parameters and their corresponding preferred performance indicators.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

11. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

Images