Smelting simulation methods, apparatus, computer equipment, storage media and software products

Abstract

This application relates to a smelting simulation method, apparatus, computer equipment, storage medium, and computer program product. The method includes: acquiring raw material input parameters; inputting the raw material input parameters into a preset smelting simulation model, simulating a preset smelting process for the raw materials, and generating simulation results for the output of associated elements; the preset smelting process includes a first preset smelting process where the raw material input parameters are greater than a preset threshold and a second preset smelting process where the raw material input parameters are less than or equal to a preset threshold; and outputting the simulation results for the output of associated elements. By using this method to input the acquired raw material input parameters into a preset smelting simulation model and simulate the preset smelting process for the raw materials, the entire process of metal smelting corresponding to the raw materials can be simulated, thereby generating simulation results for the output of associated elements more accurately. Therefore, the output and flow of associated elements can be simulated more accurately.

Description

Technical Field

[0001] This application relates to the field of metal smelting technology, and in particular to a smelting simulation method, apparatus, computer equipment, storage medium and computer program product. Background Technology

[0002] During metal smelting, some other elements (i.e., byproducts) enter the main element products and waste along with the metal products. These byproducts can lead to a decline in the quality of the main element products and environmental pollution. Therefore, it is necessary to simulate the output and flow of these byproducts during metal smelting to enhance production management. Generally, the output and flow of these byproducts can be simulated by simulating the metal smelting process.

[0003] Traditional techniques employ backpropagation (BP) neural network models to simulate the metal smelting process, thereby simulating the production and flow of byproducts. However, using traditional techniques to simulate the production and flow of byproducts suffers from inaccuracies. Summary of the Invention

[0004] Therefore, it is necessary to provide a smelting simulation method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the accuracy of smelting simulation in response to the above-mentioned technical problems.

[0005] Firstly, this application provides a smelting simulation method. The method includes:

[0006] Obtain the input parameters of raw materials;

[0007] The input parameters of the raw materials are input into a preset smelting simulation model to simulate the preset smelting process of the raw materials and generate simulation results of associated element production. The preset smelting process includes a first preset smelting process in which the input parameters of the raw materials are greater than a preset threshold and a second preset smelting process in which the input parameters of the raw materials are less than or equal to the preset threshold.

[0008] Output the simulation results of the associated element production.

[0009] In one embodiment, the step of inputting the input parameters of the raw materials into a preset smelting simulation model, simulating a preset smelting process of the raw materials, and generating simulation results of associated element production includes:

[0010] The input parameters of the raw materials are input into the preset smelting simulation model to simulate the first preset smelting process of the raw materials and generate intermediate simulation results of the raw materials.

[0011] The intermediate simulation results of the raw materials are input into the preset smelting simulation model to simulate the second preset smelting process of the raw materials and generate the simulation results of the by-product output.

[0012] In one embodiment, the first preset smelting process includes a smelting process and a blowing process; the preset smelting simulation model includes a preset long short-term memory model;

[0013] The step of inputting the raw material input parameters into the preset smelting simulation model, simulating the first preset smelting process of the raw material, and generating intermediate simulation results of the raw material includes:

[0014] The input parameters of the raw materials are input into the preset long short-term memory model to simulate the smelting process of the raw materials and generate smelting simulation results; the smelting simulation results include the output parameters of the smelting process.

[0015] The output parameters of the smelting process are input into the preset long short-term memory model to simulate the blowing process of the raw material and generate intermediate simulation results of the raw material; the intermediate simulation results include the output parameters of the blowing process.

[0016] In one embodiment, the method further includes:

[0017] Obtain a first sample dataset; the first sample dataset includes a first training sample set and a first verification sample set, the first training sample set includes first historical input parameters of multiple raw materials, and the first verification sample set includes first historical output parameters of the multiple raw materials.

[0018] For multiple raw materials, the first historical input parameters of the raw materials are input into an initial long short-term memory model for training to obtain the first predicted output parameters of each raw material.

[0019] Based on the first predicted output parameters of the raw materials and the first historical output parameters of the raw materials, the value of the first loss function is calculated, and the model parameters of the initial long short-term memory model are adjusted according to the value of the first loss function to obtain the preset long short-term memory model.

[0020] In one embodiment, the second preset smelting process includes a refining process, an electrolysis process, and / or a slag beneficiation process; the preset smelting simulation model includes a preset Bayesian network model; the preset Bayesian network model includes a first preset Bayesian network model corresponding to the refining process, a second preset Bayesian network model corresponding to the electrolysis process, and / or a third preset Bayesian network model corresponding to the slag beneficiation process.

[0021] The step of inputting the intermediate simulation results into the preset smelting simulation model to simulate the second preset smelting process of the raw materials and generating the simulation results of the by-product output includes:

[0022] The intermediate simulation results are input into the first preset Bayesian network model to simulate the refining process of the raw materials and generate refining simulation results.

[0023] The refining simulation results are input into the second preset Bayesian network model to perform electrolysis simulation on the electrolysis process of the raw materials, generating electrolysis simulation results, and determining the simulation results of the by-product output based on the electrolysis simulation results; and / or

[0024] The intermediate simulation results are input into the third preset Bayesian network model to simulate the slag beneficiation process of the raw materials, generate slag beneficiation simulation results, and determine the simulation results of the associated element output based on the slag beneficiation simulation results.

[0025] In one embodiment, the method further includes:

[0026] Obtain a second sample dataset; the second sample dataset includes a second training sample set and a second validation sample set, the second training sample set includes second historical input parameters of multiple raw materials, and the second validation sample set includes second historical output parameters of the multiple raw materials.

[0027] For multiple raw materials, the second historical input parameters of the raw materials and the intermediate simulation results of the raw materials are input into the initial Bayesian network model for training to obtain the second predicted output parameters of each raw material.

[0028] Based on the second predicted output parameters of the raw materials and the second historical output parameters of the raw materials, the value of the second loss function is calculated, and the model parameters of the initial Bayesian network model are adjusted according to the value of the second loss function to obtain the preset Bayesian network model.

[0029] Secondly, this application also provides a smelting simulation apparatus. The apparatus includes:

[0030] The acquisition module is used to acquire the input parameters of raw materials;

[0031] The smelting simulation module is used to input the input parameters of the raw materials into a preset smelting simulation model, simulate the preset smelting process of the raw materials, and generate simulation results of the output of associated elements; the preset smelting process includes a first preset smelting process in which the input parameters of the raw materials are greater than a preset threshold and a second preset smelting process in which the input parameters of the raw materials are less than or equal to the preset threshold.

[0032] The output module is used to output the simulation results of the associated element production.

[0033] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method in any of the embodiments of the first aspect described above.

[0034] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the method in any of the embodiments of the first aspect described above.

[0035] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the method in any of the embodiments of the first aspect described above.

[0036] The aforementioned smelting simulation method, apparatus, computer equipment, storage medium, and computer program product acquire raw material input parameters; input these parameters into a preset smelting simulation model; simulate a preset smelting process for the raw materials; and generate simulation results for associated element production. The preset smelting process includes a first preset smelting process where raw material input parameters are greater than a preset threshold and a second preset smelting process where raw material input parameters are less than or equal to a preset threshold; and outputs simulation results for associated element production. Since the preset smelting process includes both the first and second preset smelting processes, representing the entire metal smelting process, this application, by inputting the acquired raw material input parameters into a preset smelting simulation model and simulating the preset smelting process for the raw materials, can simulate the entire metal smelting process corresponding to the raw materials, thereby generating simulation results for associated element production more accurately. Therefore, it can more accurately simulate the production and flow of associated elements. Attached Figure Description

[0037] Figure 1 This is a diagram illustrating the application environment of a smelting simulation method in one embodiment;

[0038] Figure 2 This is a schematic flowchart of a smelting simulation method in one embodiment;

[0039] Figure 3 This is a flowchart illustrating the smelting simulation steps in one embodiment;

[0040] Figure 4This is a flowchart illustrating the intermediate simulation result generation steps in one embodiment;

[0041] Figure 5 This is a schematic diagram of the first preset smelting process in one embodiment;

[0042] Figure 6 This is a flowchart illustrating the first training step in another embodiment;

[0043] Figure 7 This is a flowchart illustrating the steps for generating simulation results of associated element production in one embodiment;

[0044] Figure 8 This is a schematic diagram of a preset smelting process in one embodiment;

[0045] Figure 9 This is a flowchart illustrating the second training step in another embodiment;

[0046] Figure 10 This is a schematic diagram of the structure of a Bayesian network in one embodiment;

[0047] Figure 11 This is a schematic flowchart of a smelting simulation method in one optional embodiment;

[0048] Figure 12 This is a structural block diagram of a smelting simulation apparatus in one embodiment;

[0049] Figure 13 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0051] In the metal smelting process, some other elements (i.e., byproducts) enter the main element products and waste along with the main element products. These byproducts not only lead to a decline in the quality of the main element products and environmental pollution, but also require resource recovery or safe disposal. Therefore, in the metal smelting process, it is necessary to simulate the output and flow of these byproducts to enhance production management. Generally, the output and flow of these byproducts can be simulated by simulating the metal smelting process, thereby determining the distribution of byproducts under different operating conditions or variables, and thus controlling the output and flow of these byproducts. However, because metal smelting (such as copper smelting) processes are affected by multiple operating variables, and the influence of each operating variable on element flow is nonlinear and has strong time lag characteristics, it is impossible to precisely control byproducts in practical applications.

[0052] In traditional techniques, a backpropagation (BP) neural network model is used to simulate the metal smelting process, thereby simulating the production and flow of associated elements.

[0053] However, there is a severe uneven distribution of data across different stages of metal smelting. For example, a large amount of production data exists in the critical stages of crude metal smelting (i.e., a large number of historical input and output parameters), while very little production data exists in other stages of refining metal smelting (i.e., a small number of historical input and output parameters). Generally, BP neural network models can only simulate the metal smelting process for stages with large amounts of data. Therefore, when using BP neural network models to simulate metal smelting, only the critical stages of crude metal smelting can be simulated, not other stages of refining. Consequently, BP neural network models cannot simulate the entire smelting production process. Furthermore, BP neural network models cannot accurately simulate the output and flow of associated elements from a holistic perspective; that is, simulating the output and flow of associated elements using traditional techniques results in inaccurate simulations.

[0054] The smelting simulation method provided in this application can be applied to, for example... Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on the cloud or other network servers. Server 104 obtains the input parameters of raw materials from terminal 102; server 104 inputs the input parameters of raw materials into a preset smelting simulation model, simulates a preset smelting process for the raw materials, and generates simulation results of associated element output; the preset smelting process includes a first preset smelting process where the input parameters of raw materials are greater than a preset threshold and a second preset smelting process where the input parameters of raw materials are less than or equal to the preset threshold; and outputs the simulation results of associated element output. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart vehicle devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0055] In one embodiment, such as Figure 2 As shown, a smelting simulation method is provided, which can be applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0056] Step 220: Obtain the input parameters of raw materials.

[0057] Smelting simulation refers to using machine learning models to simulate the production and flow of associated metal elements during metal smelting under different operating conditions. Operating conditions include, but are not limited to, raw material input parameters, temperature conditions, and environmental conditions. In this embodiment, metal smelting includes copper smelting. Accordingly, in this embodiment, raw materials refer to the original materials used for copper smelting, such as copper sulfide concentrate and copper oxide ore. Raw material input parameters include, but are not limited to, the amount of each raw material input during copper smelting, operating parameters (such as blast volume, oxygen supply, current density, and dosage), and the composition of each element (such as the amount of copper, arsenic, and lead). Optionally, the user can input the raw material input parameters on terminal 102, so that server 104 can obtain the raw material input parameters from terminal 102.

[0058] Step 240: Input the raw material input parameters into the preset smelting simulation model, simulate the preset smelting process of the raw materials, and generate simulation results of associated element output; the preset smelting process includes a first preset smelting process where the raw material input parameters are greater than a preset threshold and a second preset smelting process where the raw material input parameters are less than or equal to the preset threshold.

[0059] Optionally, server 104 can input the raw material input parameters into a preset smelting simulation model to simulate a preset smelting process of the raw materials and generate simulation results of associated element output. The preset smelting simulation model is a machine learning model for simulating metal smelting. Optionally, the preset smelting simulation model can be a probabilistic graphical model, or a model combining a probabilistic graphical model and a neural network model. Of course, this embodiment does not limit the preset smelting simulation model. The preset smelting process refers to the entire metal smelting process, including a first preset smelting process where the raw material input parameters are greater than a preset threshold and a second preset smelting process where the raw material input parameters are less than or equal to the preset threshold. The amount of production data can be determined based on the preset threshold. Of course, this embodiment does not limit the preset threshold. The first preset smelting process is used to roughly refine the target element from the raw materials (i.e., roughly extract the target element), and a large amount of production data exists during the rough refining process; the second preset smelting process is used to refine the target element from the raw materials (i.e., more finely extract the target element), and only a small amount of production data exists during the refining process. The production data includes historical input parameters and historical output parameters for raw materials. For example, in a copper smelting simulation, the first preset smelting process refers to the smelting process for crude copper elements, and the second preset smelting process refers to the smelting process for refined copper elements.

[0060] Step 260: Output the simulation results of the associated element production.

[0061] Optionally, server 104 can output the simulation results of associated element production to terminal 102, allowing users to view these results on terminal 102. The simulation results refer to the production and flow of associated metal elements determined based on the metal smelting simulation process. Specifically, the simulation results show the production (e.g., the content of associated elements in each product) and flow of associated elements in each product at every stage of the metal smelting process. In practical applications, the production and flow of associated elements throughout the entire metal smelting process under different operating conditions can be controlled based on the simulation results.

[0062] In the aforementioned smelting simulation method, the input parameters of raw materials are obtained; these parameters are input into a preset smelting simulation model to simulate a preset smelting process for the raw materials, generating simulation results for the output of associated elements. The preset smelting process includes a first preset smelting process where the input parameters of the raw materials are greater than a preset threshold, and a second preset smelting process where the input parameters are less than or equal to the preset threshold. The simulation results for the output of associated elements are then output. Since the preset smelting process includes both the first preset smelting process where the input parameters of the raw materials are greater than the preset threshold and the second preset smelting process where the input parameters are less than or equal to the preset threshold, the preset smelting process represents the entire metal smelting process. Therefore, by inputting the obtained input parameters of the raw materials into the preset smelting simulation model and simulating the preset smelting process of the raw materials, this application can simulate the entire metal smelting process corresponding to the raw materials, thereby generating simulation results for the output of associated elements more accurately. Thus, the output and flow of associated elements can be simulated more accurately.

[0063] In one embodiment, such as Figure 3 As shown, the input parameters of the raw materials are input into the preset smelting simulation model to simulate the preset smelting process of the raw materials, and to generate simulation results of the output of associated elements, including:

[0064] Step 320: Input the raw material input parameters into the preset smelting simulation model, simulate the first preset smelting process of the raw materials, and generate intermediate simulation results of the raw materials.

[0065] Optionally, the preset smelting simulation model can be a probabilistic graphical model, or it can be a model combining a probabilistic graphical model and a neural network model. The probabilistic graphical model uses a graph to represent the joint probability distribution of variables related to the model. Probabilistic graphical models are suitable for small datasets, can simulate causal relationships between variables, and can ignore missing data. Neural network models are a mathematical method for simulating actual neural networks and are suitable for situations with sufficient data. When the preset smelting simulation model only includes a probabilistic graphical model, the server 104 can input the raw material input parameters into the preset probabilistic graphical model to simulate the first preset smelting process of the raw materials and generate intermediate simulation results for the raw materials. When the preset smelting simulation model is a model combining a probabilistic graphical model and a neural network model, the server 104 can input the raw material input parameters into the combined preset model to simulate the first preset smelting process of the raw materials and generate intermediate simulation results for the raw materials. The intermediate simulation results for the raw materials refer to the simulation results for the first preset smelting process of the raw materials.

[0066] Step 340: Input the intermediate simulation results of the raw materials into the preset smelting simulation model, simulate the second preset smelting process of the raw materials, and generate the simulation results of the output of associated elements.

[0067] Optionally, when the preset smelting simulation model only includes a probabilistic graphical model, the server 104 can input the intermediate simulation results of the raw materials into the preset probabilistic graphical model to simulate the second preset smelting process of the raw materials and generate simulation results of associated element production. When the preset smelting simulation model is a preset model combining a probabilistic graphical model and a neural network model, the server 104 can input the intermediate simulation results of the raw materials into the aforementioned combined preset model to simulate the second preset smelting process of the raw materials and generate simulation results of associated element production. The simulation results of associated element production refer to the simulation results of associated elements in each product substance at each stage of the entire metal smelting process of the raw materials.

[0068] In this embodiment, the input parameters of the raw materials are input into a preset smelting simulation model to simulate the first preset smelting process of the raw materials, generating intermediate simulation results for the raw materials. The intermediate simulation results are then input into the preset smelting simulation model to simulate the second preset smelting process of the raw materials, generating simulation results for the production of associated elements. By simulating the first preset smelting process where the input parameters of the raw materials are greater than a preset threshold and the second preset smelting process where the input parameters of the raw materials are less than or equal to a preset threshold, the entire process of metal smelting corresponding to the raw materials can be simulated more accurately, thereby generating simulation results for the production of associated elements more accurately.

[0069] In one embodiment, such as Figure 4 As shown, the first preset smelting process includes a smelting process and a blowing process; the preset smelting simulation model includes a preset long short-term memory model.

[0070] The raw material input parameters are input into a preset smelting simulation model to simulate the first preset smelting process of the raw materials, generating intermediate simulation results for the raw materials, including:

[0071] Step 420: Input the raw material input parameters into the preset long short-term memory model, perform smelting simulation on the raw material smelting process, and generate smelting simulation results; the smelting simulation results include the output parameters of the smelting process.

[0072] Among them, such as Figure 5 As shown, Figure 5This is a schematic diagram of a first preset smelting process in one embodiment. The first preset smelting process includes a smelting process and a blowing process. The initial input parameters of the raw materials and the parameters input in the smelting process jointly affect the output results of the smelting process, and the initial input parameters of the raw materials, the parameters input in the smelting process, and the parameters input in the blowing process jointly affect the output results of the blowing process. The smelting process can transfer intermediate products to the blowing process. For example, the server 104 can input the input parameters of the raw materials into a preset long short-term memory model to simulate the smelting process of the raw materials and generate smelting simulation results. The input parameters of the raw materials include the initial input parameters of the raw materials and the parameters input in the smelting process. The parameters input in the smelting process include the blast volume of the smelting process, the oxygen supply of the smelting process, the amount of quartz added in the smelting process, the amount of coal added in the smelting process, the amount of return material added in the smelting process, and the amount of mixed ore added in the smelting process, etc. Melting simulation is used to simulate the impact of initial input parameters of raw materials and parameters input into the smelting process on the output parameters of the smelting process. The results of the smelting simulation include the output parameters of the smelting process.

[0073] In this embodiment, the preset smelting simulation model includes a preset long short-term memory (LSTM) model. LSTM is a neural network model. LSTM can achieve selective memory through gating networks, solving the gradient vanishing and exploding problems in long-sequence deep learning. In LSTM, when passing data from the previous time point t-1 to the next time point t, both short-term and long-term memories can be passed simultaneously. Based on the current input data x... t Weight parameter W and previous short-term memory parameter S t-1 By performing calculations, the short-term memory parameter S can be obtained. t Long-term memory parameter L t Long-term memory parameter L can be obtained by simultaneously adding some new memories and forgetting some old memories. t The calculation formula is shown in equation (1) below:

[0074] L t =L t,new -L t,old (1)

[0075] New long-term memory parameter L t,new The calculation formula is shown in equation (2) below:

[0076] L t,new =f(W new ,x t ,S t-1 (2)

[0077] Old long-term memory parameter L t,old The calculation formula is shown in equation (3) below:

[0078] L t,old =f(W old ,x t ,S t-1 )·L t-1 (3)

[0079] Among them, S t-1 The short-term memory parameters at time t-1; L t-1 The long-term memory parameter at time t-1; x t The input data is at time t; W new For new memory weights; W old The old memory weights are f; the activation function is L. t,new For new long-term memory parameters; L t,old For old long-term memory parameters; L t These are parameters for long-term memory.

[0080] Step 440: Input the output parameters of the smelting process into the preset long short-term memory model, perform a blowing simulation of the raw material blowing process, and generate intermediate simulation results of the raw material; the intermediate simulation results include the output parameters of the blowing process.

[0081] For example, server 104 can use some output parameters of the smelting process as intermediate products and input these intermediate products into a preset long short-term memory model to simulate the blowing process of raw materials, generating intermediate simulation results for the raw materials. Server 104 can also use another part of the output parameters of the smelting process as by-products or waste, and stop metal smelting on the by-products or waste. The input parameters of the raw materials include parameters input into the blowing process, such as the blast volume, oxygen supply, quartz addition, and cold charge addition. The blowing simulation is used to simulate the impact of the initial input parameters of the raw materials, the parameters input into the smelting process, and the parameters input into the blowing process on the output parameters of the blowing process. The intermediate simulation results include the output parameters of the blowing process.

[0082] In this embodiment, the input parameters of the raw materials are input into a preset long short-term memory model to simulate the smelting process of the raw materials and generate smelting simulation results. The output parameters of the smelting process are input into the preset long short-term memory model to simulate the blowing process of the raw materials and generate intermediate simulation results of the raw materials. By sequentially simulating the smelting process and the blowing process, the first preset smelting process can be simulated more accurately, thereby generating intermediate simulation results of the raw materials more accurately.

[0083] In one embodiment, such as Figure 6 As shown, the smelting simulation method also includes:

[0084] Step 620: Obtain the first sample dataset; the first sample dataset includes a first training sample set and a first verification sample set. The first training sample set includes the first historical input parameters of multiple raw materials, and the first verification sample set includes the first historical output parameters of multiple raw materials.

[0085] Optionally, server 104 can obtain a first sample dataset from the historical production process. The first sample dataset includes a first training sample set and a first validation sample set. The first training sample set includes first historical input parameters for multiple raw materials, and the first validation sample set includes first historical output parameters for multiple raw materials. In this embodiment, the first sample dataset can be divided according to time order. For example, taking consecutive days t as a group, the daily data is merged and sliced ​​with the data from the previous t-1 days in chronological order to generate a time-series dataset suitable for the Long Short-Term Memory (LSTM) model. The first sample dataset can also be divided into a training set, a validation set, and a test set in a 7:1:2 ratio. Of course, this embodiment does not limit the specific division method of the first sample dataset. Furthermore, since fluctuations in different data can lead to differences in the learning ability of the LTM model, this embodiment can perform regularization processing on the first sample dataset. The data in the regularized first sample dataset satisfies a distribution with a mean of 0 and homoscedasticity.

[0086] Step 640: For multiple raw materials, input the first historical input parameters of the raw materials into the initial long short-term memory model for training to obtain the first predicted output parameters of each raw material.

[0087] Step 660: Calculate the value of the first loss function based on the first predicted output parameters and the first historical output parameters of the raw materials, and adjust the model parameters of the initial long short-term memory model based on the value of the first loss function to obtain the preset long short-term memory model.

[0088] Optionally, firstly, server 104 can set the basic structural parameters of the initial long short-term memory model for both the smelting and blowing processes. These basic structural parameters include the number of network layers *l*, the number of intermediate network layers (also called hidden layers) *h*, the activation function *tanh*, the number of training data sets *b*, and the model calculation method. Furthermore, to avoid overfitting, output data of a proportion *d* can be randomly selected from the intermediate network layers, and this proportion of output data can be set to zero.

[0089] In this embodiment, the activation function selected for the initial long short-term memory model is the tanh function, for the following reasons: Compared with other activation functions such as the Sigmoid function, the tanh function is close to a linear function at (0,0). Therefore, the tanh function can greatly reduce the computational load and save running time compared to exponential calculation. When training deep neural networks, the Sigmoid function only has a large change in the interval [-1,1], and there will be a gradient vanishing problem outside this interval; while the tanh function changes more smoothly. Therefore, the tanh function can more conveniently perform data transmission in deep neural networks. The calculation formula of the tanh function is shown in the following formula (4):

[0090]

[0091] Where tanh(x) is the selected activation function.

[0092] In this embodiment, the selected model calculation method is the momentum-driven stochastic gradient descent algorithm. Momentum-driven stochastic gradient descent is suitable for unconstrained extreme value problems in nonlinear programming. When using gradient descent, it is necessary to select an initial point, determine the search direction as the negative gradient direction, and set the search step size to λ, continuously updating the sample data until an approximate minimum point is found. After incorporating momentum, the search direction is determined as the weighted average of the current gradient descent direction and the previous gradient descent direction, and a momentum weight β is set.

[0093] Secondly, for the smelting and blowing processes, server 104 can determine the optimal basic structural parameters of the initial long short-term memory model based on different basic structural parameters of the initial long short-term memory model, so as to achieve better model prediction results. The optimal basic structural parameters of the initial long short-term memory model are shown in the table below.

[0094] Table 1

[0095]

[0096] Next, for multiple raw materials, server 104 can input the first historical input parameters of the raw materials into the initial long short-term memory model, and use the stochastic gradient descent algorithm with momentum to train the model and obtain the first predicted output parameters for each raw material. Then, based on the first predicted output parameters (predicted values) and the first historical output parameters (actual values) of the raw materials, server 104 can calculate the value of the first loss function and adjust the model parameters of the initial long short-term memory model according to the value of the first loss function to generate the target model parameters of the initial long short-term memory model. Finally, server 104 can adjust the initial model parameters of the initial long short-term memory model to the target model parameters, thereby obtaining the preset long short-term memory model.

[0097] In this embodiment, firstly, a first sample dataset including a first training sample set and a first validation sample set is obtained. For multiple raw materials, the first historical input parameters of the raw materials are input into an initial long short-term memory model for training, obtaining the first predicted output parameters for each raw material. Then, based on the first predicted output parameters and the first historical output parameters of the raw materials, the value of a first loss function is calculated, and the model parameters of the initial long short-term memory model are adjusted according to the value of the first loss function to obtain a preset long short-term memory model. By training the initial long short-term memory model using the first historical input parameters and the first historical output parameters from the first training sample set, a more accurate preset long short-term memory model can be generated.

[0098] In one embodiment, such as Figure 7 As shown, the second preset smelting process includes a refining process, an electrolysis process, and / or a slag beneficiation process; the preset smelting simulation model includes a preset Bayesian network model; the preset Bayesian network model includes a first preset Bayesian network model corresponding to the refining process, a second preset Bayesian network model corresponding to the electrolysis process, and / or a third preset Bayesian network model corresponding to the slag beneficiation process.

[0099] The intermediate simulation results are input into the preset smelting simulation model to simulate the second preset smelting process of the raw materials, generating simulation results of associated element output, including:

[0100] Step 720: Input the intermediate simulation results into the first preset Bayesian network model to simulate the refining process of raw materials and generate refining simulation results.

[0101] Step 740: Input the refining simulation results into the second preset Bayesian network model to perform electrolysis simulation on the raw material electrolysis process, generate electrolysis simulation results, and determine the simulation results of associated element production based on the electrolysis simulation results; and / or

[0102] The intermediate simulation results are input into the third preset Bayesian network model to simulate the slag beneficiation process of raw materials, generate slag beneficiation simulation results, and determine the simulation results of associated element production based on the slag beneficiation simulation results.

[0103] Among them, such as Figure 8 As shown, Figure 8 This is a schematic diagram of a preset smelting process in one embodiment. The preset smelting process includes a first preset smelting process where the raw material input parameters are greater than a preset threshold, and a second preset smelting process where the raw material input parameters are less than or equal to the preset threshold. The first preset smelting process includes a smelting process and a blowing process, and the second preset smelting process includes a refining process, an electrolysis process, and / or a slag beneficiation process. Optionally, the second preset smelting process may include a refining process and an electrolysis process, and may also include a slag beneficiation process, or may further include a refining process, an electrolysis process, and a slag beneficiation process. The first preset smelting process is represented by a dashed box, and the portion outside the dashed box represents the second preset smelting process. In the first preset smelting process, the vast majority of associated metallic elements can be generated and discharged.

[0104] In this embodiment, the preset smelting simulation model includes a preset Bayesian network model. The preset Bayesian network model includes a first preset Bayesian network model corresponding to the refining process, a second preset Bayesian network model corresponding to the electrolysis process, and / or a third preset Bayesian network model corresponding to the slag beneficiation process. A Bayesian network model is a probabilistic graphical model that can transform a problem into a joint probability distribution of random variables. The Bayesian network model primarily simulates network construction and parameter learning. First, the network structure of the Bayesian network model is constructed through theory and experience. The network structure of the Bayesian network model (referred to as a Bayesian network) is a directed acyclic graph representing causal relationships between variables. Nodes in a Bayesian network represent parameters or variables, edges represent causal relationships between parameters or variables, and an edge from a parent node to a child node represents the influence of a causal variable (or causal parameter) on an outcome variable (or outcome parameter). Connecting the nodes with multiple directed edges forms a Bayesian network. Then, parameters for each edge are learned using actual production data, thereby correcting the model parameters and generating the optimal parameters for the Bayesian network model.

[0105] Optionally, the first approach is as follows: First, server 104 can input some output parameters from the intermediate simulation results and parameters input into the refining process into a first preset Bayesian network model to simulate the refining process of raw materials and generate refining simulation results. Then, server 104 can input some output parameters from the refining simulation results and parameters input into the electrolysis process into a second preset Bayesian network model to simulate the electrolysis process of raw materials and generate electrolysis simulation results. Based on the electrolysis simulation results and the output parameters of other processes, the simulation results for the production of associated elements are determined. The second approach is as follows: Server 104 can also input some output parameters from the intermediate simulation results and parameters input into the slag beneficiation process into a third preset Bayesian network model to simulate the slag beneficiation process of raw materials and generate slag beneficiation simulation results. Based on the slag beneficiation simulation results and the output parameters of other processes, the simulation results for the production of associated elements are determined.

[0106] The third approach is as follows: First, server 104 can input some output parameters from intermediate simulation results and parameters input into the refining process into a first preset Bayesian network model to simulate the refining process of raw materials and generate refining simulation results. Second, server 104 can input some output parameters from the refining simulation results and parameters input into the electrolysis process into a second preset Bayesian network model to simulate the electrolysis process of raw materials and generate electrolysis simulation results. Furthermore, server 104 can also input some output parameters from intermediate simulation results and parameters input into the slag beneficiation process into a third preset Bayesian network model to simulate the slag beneficiation process of raw materials and generate slag beneficiation simulation results. Then, based on the electrolysis simulation results, the slag beneficiation simulation results, and the output parameters of other processes, the associated element output simulation results are determined. Here, the associated element output simulation results refer to the simulation results of associated elements in each output substance at each stage of the entire metal smelting process of raw materials. The entire process of metal smelting includes, but is not limited to, smelting, blowing, refining, electrolysis, and slag beneficiation.

[0107] The raw material input parameters include those for the refining process, the electrolysis process, and the slag beneficiation process. The refining process parameters include the amounts of cold feed, fuel, quartz sand, and coal added. The electrolysis process parameters include current density, sulfuric acid concentration, copper ion concentration, and chloride ion concentration in the electrolyte. The slag beneficiation process parameters include the dosage of reagents. The refining simulation results include the output parameters of the refining process. The electrolysis simulation results include the output parameters of the electrolysis process. The slag beneficiation simulation results include the output parameters of the slag beneficiation process.

[0108] In this embodiment, intermediate simulation results are input into a first preset Bayesian network model to simulate the refining process of raw materials, generating refining simulation results. These refining simulation results are then input into a second preset Bayesian network model to simulate the electrolysis process of raw materials, generating electrolysis simulation results. The simulation results for associated element production are then determined based on these electrolysis simulation results. Alternatively, intermediate simulation results are input into a third preset Bayesian network model to simulate the slag beneficiation process of raw materials, generating slag beneficiation simulation results. The simulation results for associated element production are then determined based on these slag beneficiation simulation results. By sequentially simulating the refining process, the electrolysis process, and the slag beneficiation process, the second preset smelting process can be simulated more accurately, thereby enabling a more accurate determination of the simulation results for associated element production.

[0109] In one embodiment, such as Figure 9 As shown, the smelting simulation method also includes:

[0110] Step 920: Obtain the second sample dataset; the second sample dataset includes a second training sample set and a second validation sample set. The second training sample set includes the second historical input parameters of multiple raw materials, and the second validation sample set includes the second historical output parameters of multiple raw materials.

[0111] Optionally, such as Figure 10 As shown, Figure 10 This is a schematic diagram of a Bayesian network structure in one embodiment. First, server 104 can obtain a second sample dataset from the historical production process. The second sample dataset includes a second training sample set and a second validation sample set. The second training sample set includes second historical input parameters for multiple raw materials, and the second validation sample set includes second historical output parameters for multiple raw materials. Taking the copper smelting process as an example, the second historical input parameters include operational input parameters, raw material input and weight parameters, and raw material input and elemental proportion parameters. For example, for the smelting process, the operational input parameters (…) Figure 10 The first column includes the blast volume, oxygen supply, quartz dosage, coal dosage, and return material dosage for the smelting process, as well as raw material input and weight parameters. Figure 10 The second column includes the following parameters: dry material content of calcium oxide, dry material content of silicon, dry material content of sulfur, dry material content of iron, total material content of water, dry material content of copper, and the amount of mixed ore added during the smelting process, as well as the parameters of raw material input and elemental proportions. Figure 10 The third column includes dry materials containing antimony, zinc, lead, and arsenic. The second set of historical output parameters includes output and weight parameters, as well as output and elemental percentage parameters. For example, for a smelting process, the output and weight parameters ( Figure 10The fourth column includes waste acid production, coarse dust weight, smelting boiler ash weight, smelting electrolytic dust weight, smelting slag weight, and matte weight, along with output and elemental composition parameters. Figure 10 The fifth column includes, but is not limited to, waste acid containing antimony, waste acid containing zinc, waste acid containing lead, waste acid containing arsenic, and waste acid containing copper.

[0112] Subsequently, based on the acquired second sample dataset and the causal relationships between various data or parameters, a Bayesian network is constructed for each specific process within the entire process (i.e., the preset smelting process). For example, regarding the smelting process in copper smelting, based on historical causal relationship experience in copper smelting production, the second historical input parameter is determined as the cause, and the second historical output parameter as the effect. Furthermore, since operational input parameters are the primary parameters regulating the overall production process, and raw material input and weight parameters are the parameters of the output substances and the source of element flow, it is determined that there is a causal relationship between operational input parameters and all second historical output parameters, and also a causal relationship between raw material input and weight parameters and all second historical output parameters. (See...) Figure 10 Because the second and fifth columns have a complex relationship, to avoid ambiguity on each side, Figure 10 (The causal relationship between the nodes in the second column and the nodes in the fifth column is not shown.) Furthermore, the raw material input and elemental proportion parameters refer to the content of each element in the raw materials. Copper is the main production element, and the other elements are associated metal elements. In this embodiment, only the impact of copper on the output of other associated elements is considered, ignoring the interactions between associated elements. Therefore, regarding the raw material input and elemental proportion parameters, copper affects all second historical output parameters, while other associated metal elements only affect the output parameters corresponding to their own associated elements. Additionally, there are also influence relationships between the output and weight parameters and the output and elemental proportion parameters in the second historical output parameters. Specifically, the output and weight parameters of each output substance affect the output and elemental proportion parameters of that output substance; that is, the weight of each output substance affects the content of each element in that output substance.

[0113] Step 940: For multiple raw materials, the second historical input parameters of the raw materials and the intermediate simulation results of the raw materials are input into the initial Bayesian network model for training to obtain the second predicted output parameters of each raw material.

[0114] Step 960: Calculate the value of the second loss function based on the second predicted output parameters and the second historical output parameters of the raw materials, and adjust the model parameters of the initial Bayesian network model based on the value of the second loss function to obtain the preset Bayesian network model.

[0115] Optionally, for multiple raw materials, server 104 can input the second historical input parameters of the raw materials and the intermediate simulation results of the raw materials into the initial Bayesian network model, and use the Naive Bayes ridge regression algorithm to train the model to obtain the second predicted output parameters of each raw material. Then, based on the second predicted output parameters (predicted values) and the second historical output parameters (true values) of the raw materials, server 104 can calculate the value of the second loss function and adjust the model parameters of the initial Bayesian network model according to the value of the second loss function to generate the target model parameters of the initial Bayesian network. Finally, server 104 can adjust the initial model parameters of the initial Bayesian network to the target model parameters, thereby obtaining the preset Bayesian network model. In this embodiment, the Naive Bayes ridge regression algorithm is used for parameter learning of the Bayesian network model. Ridge regression is an improved least squares model and belongs to the biased estimation regression method. Since the input parameters in this embodiment suffer from multicollinearity and heteroscedasticity, the ridge regression algorithm can effectively solve the problem of multicollinearity by adding a regularization term to the loss function and using gradient descent to solve it.

[0116] In this embodiment, firstly, a second sample dataset including a second training sample set and a second validation sample set is obtained. For multiple raw materials, the second historical input parameters of the raw materials and the intermediate simulation results of the raw materials are input into the initial Bayesian network model for training, obtaining the second predicted output parameters for each raw material. Then, based on the second predicted output parameters and the second historical output parameters of the raw materials, the value of the second loss function is calculated, and the model parameters of the initial Bayesian network model are adjusted according to the value of the second loss function to obtain the preset Bayesian network model. By training the initial Bayesian network model using the second historical input parameters and the second historical output parameters from the second training sample set, a more accurate preset Bayesian network model can be generated.

[0117] In an optional embodiment, such as Figure 11 As shown, a smelting simulation method is provided, which can be applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0118] Step 1102: Obtain the input parameters of the raw materials;

[0119] Step 1104: Obtain the first sample dataset; the first sample dataset includes a first training sample set and a first verification sample set. The first training sample set includes the first historical input parameters of multiple raw materials, and the first verification sample set includes the first historical output parameters of multiple raw materials.

[0120] Step 1106: For multiple raw materials, input the first historical input parameters of the raw materials into the initial long short-term memory model for training to obtain the first predicted output parameters of each raw material.

[0121] Step 1108: Calculate the value of the first loss function based on the first predicted output parameters of the raw materials and the first historical output parameters of the raw materials, and adjust the model parameters of the initial long short-term memory model based on the value of the first loss function to obtain the preset long short-term memory model.

[0122] Step 1110: Input the raw material input parameters into the preset long short-term memory model, perform smelting simulation on the raw material smelting process, and generate smelting simulation results; the smelting simulation results include the output parameters of the smelting process; the preset smelting process includes a first preset smelting process where the raw material input parameters are greater than a preset threshold and a second preset smelting process where the raw material input parameters are less than or equal to the preset threshold; the first preset smelting process includes a smelting process and a blowing process;

[0123] Step 1112: Input the output parameters of the smelting process into the preset long short-term memory model, perform a blowing simulation of the raw material blowing process, and generate intermediate simulation results of the raw material; the intermediate simulation results include the output parameters of the blowing process.

[0124] Step 1114: Obtain the second sample dataset; the second sample dataset includes a second training sample set and a second validation sample set. The second training sample set includes the second historical input parameters of multiple raw materials, and the second validation sample set includes the second historical output parameters of multiple raw materials.

[0125] Step 1116: For multiple raw materials, input the second historical input parameters of the raw materials and the intermediate simulation results of the raw materials into the initial Bayesian network model for training to obtain the second predicted output parameters of each raw material.

[0126] Step 1118: Calculate the value of the second loss function based on the second predicted output parameters and the second historical output parameters of the raw materials, and adjust the model parameters of the initial Bayesian network model based on the value of the second loss function to obtain the preset Bayesian network model.

[0127] Step 1120: Input the intermediate simulation results into the first preset Bayesian network model to perform refining simulation on the refining process of raw materials and generate refining simulation results; the second preset smelting process includes refining process, electrolysis process and / or slag beneficiation process.

[0128] Step 1122: Input the refining simulation results into the second preset Bayesian network model to perform electrolysis simulation on the raw material electrolysis process, generate electrolysis simulation results, and determine the simulation results of associated element production based on the electrolysis simulation results; and / or

[0129] Step 1124: Input the intermediate simulation results into the third preset Bayesian network model to simulate the slag beneficiation process of raw materials, generate slag beneficiation simulation results, and determine the simulation results of associated element production based on the slag beneficiation simulation results.

[0130] Step 1126: Output the simulation results of the associated element production.

[0131] In the aforementioned smelting simulation method, the input parameters of raw materials are obtained; these parameters are input into a preset smelting simulation model to simulate a preset smelting process for the raw materials, generating simulation results for the output of associated elements. The preset smelting process includes a first preset smelting process where the input parameters of raw materials are greater than a preset threshold and a second preset smelting process where the input parameters are less than or equal to the preset threshold; the simulation results for the output of associated elements are then output. Since the preset smelting process includes both the first and second preset smelting processes, representing the entire metal smelting process, this application, by inputting the obtained input parameters of raw materials into a preset smelting simulation model and simulating the preset smelting process for the raw materials, can simulate the entire metal smelting process corresponding to the raw materials, thereby generating simulation results for the output of associated elements more accurately. Therefore, the output and flow of associated elements can be simulated more accurately.

[0132] Furthermore, in terms of theoretical research, by using a pre-set smelting simulation model to conduct refined simulations of the production and flow of associated metal elements, it is possible to simulate the production of various substances and the production and flow of each element under different operating conditions. This allows for a detailed analysis of the production and flow patterns of associated metal elements during actual copper smelting production. In practical applications, the pre-set smelting simulation model of this application can simulate the production of various substances and the production and flow of each element under different operating conditions, thereby enabling the determination of the optimal production control scheme based on actual working conditions. Because this application can predict the production and flow of each element in advance, it can serve as an important reference for production adjustment, thereby improving the effectiveness of smelting production operations.

[0133] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0134] Based on the same inventive concept, this application also provides a smelting simulation apparatus for implementing the smelting simulation method described above. The solution provided by this apparatus is similar to the solution described in the above method; therefore, the specific limitations in one or more smelting simulation apparatus embodiments provided below can be found in the limitations of the smelting simulation method described above, and will not be repeated here.

[0135] In one embodiment, such as Figure 12 As shown, a smelting simulation device 1200 is provided, including: an acquisition module 1220, a smelting simulation module 1240, and an output module 1260, wherein:

[0136] The acquisition module 1220 is used to acquire the input parameters of raw materials.

[0137] The smelting simulation module 1240 is used to input the input parameters of raw materials into the preset smelting simulation model, simulate the preset smelting process of raw materials, and generate simulation results of associated element output. The preset smelting process includes a first preset smelting process where the input parameters of raw materials are greater than a preset threshold and a second preset smelting process where the input parameters of raw materials are less than or equal to the preset threshold.

[0138] Output module 1260 is used to output the simulation results of the associated element production.

[0139] In one embodiment, the smelting simulation module 1240 includes:

[0140] The intermediate simulation result generation unit is used to input the input parameters of raw materials into the preset smelting simulation model, simulate the first preset smelting process of raw materials, and generate intermediate simulation results of raw materials.

[0141] The associated element production simulation result generation unit is used to input the intermediate simulation results of raw materials into the preset smelting simulation model, simulate the second preset smelting process of raw materials, and generate associated element production simulation results.

[0142] In one embodiment, the first preset smelting process includes a smelting process and a blowing process; the preset smelting simulation model includes a preset long short-term memory model; and the intermediate simulation result generation unit includes:

[0143] The smelting simulation subunit is used to input the input parameters of raw materials into a preset long short-term memory model, simulate the smelting process of raw materials, and generate smelting simulation results; the smelting simulation results include the output parameters of the smelting process.

[0144] The blowing simulation subunit is used to input the output parameters of the smelting process into a preset long short-term memory model, perform blowing simulation on the blowing process of raw materials, and generate intermediate simulation results of raw materials; the intermediate simulation results include the output parameters of the blowing process.

[0145] In one embodiment, the smelting simulation apparatus 1200 further includes:

[0146] The first sample dataset acquisition module is used to acquire the first sample dataset. The first sample dataset includes a first training sample set and a first verification sample set. The first training sample set includes the first historical input parameters of multiple raw materials, and the first verification sample set includes the first historical output parameters of multiple raw materials.

[0147] The first training module is used to train the initial long short-term memory model by inputting the first historical input parameters of the raw materials into the model for training, and to obtain the first predicted output parameters of each raw material.

[0148] The preset long short-term memory model generation module is used to calculate the value of the first loss function based on the first predicted output parameters and the first historical output parameters of the raw materials, and adjust the model parameters of the initial long short-term memory model according to the value of the first loss function to obtain the preset long short-term memory model.

[0149] In one embodiment, the second preset smelting process includes a refining process, an electrolysis process, and / or a slag beneficiation process; the preset smelting simulation model includes a preset Bayesian network model; the preset Bayesian network model includes a first preset Bayesian network model corresponding to the refining process, a second preset Bayesian network model corresponding to the electrolysis process, and / or a third preset Bayesian network model corresponding to the slag beneficiation process; the associated element production simulation result generation unit includes:

[0150] The refining simulation subunit is used to input intermediate simulation results into the first preset Bayesian network model to simulate the refining process of raw materials and generate refining simulation results.

[0151] The electrolysis simulation subunit is used to input the refining simulation results into a second preset Bayesian network model, perform electrolysis simulation on the raw material electrolysis process, generate electrolysis simulation results, and determine the simulation results of associated element production based on the electrolysis simulation results; and / or

[0152] The slag beneficiation simulation subunit is used to input intermediate simulation results into the third preset Bayesian network model to simulate the slag beneficiation process of raw materials, generate slag beneficiation simulation results, and determine the simulation results of associated element output based on the slag beneficiation simulation results.

[0153] In one embodiment, the smelting simulation apparatus 1200 further includes:

[0154] The second sample dataset acquisition module is used to acquire the second sample dataset. The second sample dataset includes a second training sample set and a second validation sample set. The second training sample set includes the second historical input parameters of multiple raw materials, and the second validation sample set includes the second historical output parameters of multiple raw materials.

[0155] The second training module is used to train the initial Bayesian network model by inputting the second historical input parameters of the raw materials and the intermediate simulation results of the raw materials into the model for training, so as to obtain the second predicted output parameters of each raw material.

[0156] The preset Bayesian network model generation module is used to calculate the value of the second loss function based on the second predicted output parameters and the second historical output parameters of the raw materials, and to adjust the model parameters of the initial Bayesian network model based on the value of the second loss function to obtain the preset Bayesian network model.

[0157] Each module in the aforementioned smelting simulation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0158] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 13As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores smelting simulation data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a smelting simulation method.

[0159] Those skilled in the art will understand that Figure 13 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0160] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0161] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0162] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0163] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0164] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0165] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0166] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A smelting simulation method characterized by, The method includes: Obtain the input parameters of raw materials; The input parameters of the raw materials are input into a preset long short-term memory model to simulate the smelting process of the raw materials and generate smelting simulation results; the smelting simulation results include the output parameters of the smelting process. The output parameters of the smelting process are input into the preset long short-term memory model to simulate the blowing process of the raw material and generate intermediate simulation results of the raw material. The intermediate simulation results include the output parameters of the blowing process. The smelting process and the blowing process belong to the first preset smelting process, and the input parameters of the raw material in the first preset smelting process are greater than the preset threshold. The intermediate simulation results are input into the first preset Bayesian network model to simulate the refining process of the raw materials and generate refining simulation results. The refining simulation results are input into a second preset Bayesian network model to perform electrolysis simulation on the electrolysis process of the raw materials, generating electrolysis simulation results. Based on these results, the simulation results for the production of by-product elements are determined; and / or The intermediate simulation results are input into the third preset Bayesian network model to simulate the slag beneficiation process of the raw materials, generate slag beneficiation simulation results, and determine the simulation results of the associated element output based on the slag beneficiation simulation results; the refining process, the electrolysis process and / or the slag beneficiation process belong to the second preset smelting process, and the input parameters of the raw materials in the second preset smelting process are less than or equal to the preset threshold. Output the simulation results of the associated element production.

2. The method according to claim 1, characterized in that, The method further includes: Obtain a first sample dataset; the first sample dataset includes a first training sample set and a first verification sample set, the first training sample set includes first historical input parameters of multiple raw materials, and the first verification sample set includes first historical output parameters of the multiple raw materials. For multiple raw materials, the first historical input parameters of the raw materials are input into an initial long short-term memory model for training to obtain the first predicted output parameters of each raw material. Based on the first predicted output parameters of the raw materials and the first historical output parameters of the raw materials, the value of the first loss function is calculated, and the model parameters of the initial long short-term memory model are adjusted according to the value of the first loss function to obtain the preset long short-term memory model.

3. The method according to claim 1, characterized in that, The method further includes: Obtain a second sample dataset; the second sample dataset includes a second training sample set and a second validation sample set, the second training sample set includes second historical input parameters of multiple raw materials, and the second validation sample set includes second historical output parameters of the multiple raw materials. For multiple raw materials, the second historical input parameters of the raw materials and the intermediate simulation results of the raw materials are input into the initial Bayesian network model for training to obtain the second predicted output parameters of each raw material. Based on the second predicted output parameters of the raw materials and the second historical output parameters of the raw materials, the value of the second loss function is calculated, and the model parameters of the initial Bayesian network model are adjusted according to the value of the second loss function to obtain the preset Bayesian network model.

4. A smelting simulation device, characterized in that, The device includes: The acquisition module is used to acquire the input parameters of raw materials; The smelting simulation module includes a smelting simulation subunit, a blowing simulation subunit, a refining simulation subunit, an electrolysis simulation subunit, and / or a slag beneficiation simulation subunit; The smelting simulation subunit is used to input the input parameters of the raw materials into a preset long short-term memory model, perform smelting simulation on the smelting process of the raw materials, and generate smelting simulation results; the smelting simulation results include the output parameters of the smelting process. The blowing simulation subunit is used to input the output parameters of the smelting process into the preset long short-term memory model, perform blowing simulation on the blowing process of the raw material, and generate intermediate simulation results of the raw material; the intermediate simulation results include the output parameters of the blowing process; the smelting process and the blowing process belong to the first preset smelting process, and the input parameters of the raw material in the first preset smelting process are greater than a preset threshold. The refining simulation subunit is used to input the intermediate simulation results into the first preset Bayesian network model to perform refining simulation on the refining process of the raw materials and generate refining simulation results. The electrolysis simulation subunit is used to input the refining simulation results into a second preset Bayesian network model, perform electrolysis simulation on the electrolysis process of the raw materials, generate electrolysis simulation results, and determine the simulation results of associated element production based on the electrolysis simulation results; and / or The slag beneficiation simulation subunit is used to input the intermediate simulation results into the third preset Bayesian network model, perform slag beneficiation simulation on the slag beneficiation process of the raw materials, generate slag beneficiation simulation results, and determine the simulation results of the associated element output based on the slag beneficiation simulation results; the refining process, the electrolysis process and / or the slag beneficiation process belong to the second preset smelting process, and the input parameters of the raw materials in the second preset smelting process are less than or equal to the preset threshold. The output module is used to output the simulation results of the associated element production.

5. The apparatus according to claim 4, characterized in that, The device further includes: The first sample dataset acquisition module is used to acquire the first sample dataset; the first sample dataset includes a first training sample set and a first verification sample set, the first training sample set includes first historical input parameters of multiple raw materials, and the first verification sample set includes first historical output parameters of the multiple raw materials. The first training module is used to train an initial long short-term memory model by inputting the first historical input parameters of the raw materials into the model for training, thereby obtaining the first predicted output parameters of each raw material. A preset long short-term memory model generation module is used to calculate the value of a first loss function based on the first predicted output parameters of the raw materials and the first historical output parameters of the raw materials, and to adjust the model parameters of the initial long short-term memory model based on the value of the first loss function to obtain the preset long short-term memory model.

6. The apparatus according to claim 4, characterized in that, The device further includes: The second sample dataset acquisition module is used to acquire a second sample dataset; the second sample dataset includes a second training sample set and a second verification sample set, the second training sample set includes second historical input parameters of multiple raw materials, and the second verification sample set includes second historical output parameters of the multiple raw materials. The second training module is used to train the initial Bayesian network model by inputting the second historical input parameters of the raw materials and the intermediate simulation results of the raw materials into the model for training, so as to obtain the second predicted output parameters of each raw material. The preset Bayesian network model generation module is used to calculate the value of the second loss function based on the second predicted output parameters of the raw materials and the second historical output parameters of the raw materials, and adjust the model parameters of the initial Bayesian network model based on the value of the second loss function to obtain the preset Bayesian network model.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 3.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

Images