Deep learning-based batching model collaborative method, equipment and media for three-step stainless steel smelting

Abstract

The present invention discloses a deep learning-based batching model collaboration method, device, and medium for three-step stainless steel smelting. The method comprises the following steps: S1: acquiring historical raw material usage data, historical energy medium usage data, tapping molten steel composition, tapping molten steel temperature, raw material cost, and energy medium cost data, constructing a data set, and dividing the data set into a training set, a test set, and a validation set; S2: acquiring calculation data of raw materials and energy medium for a batching model; S3: constructing a batching model collaboration based on cost optimization; S4: inputting the calibrated batching model calculation data into the batching model collaboration based on cost optimization, and outputting predicted tapping molten steel composition, tapping molten steel temperature, and batching predicted cost; S5: real-time collection of current raw material and energy medium usage data and costs; and S6: operating the batching model collaboration based on deep learning to achieve cost optimization using the corrected raw material and energy medium calculation data.

Description

Technical Field

[0001] The present invention relates to a deep learning-based batching model collaboration method, system, equipment and medium for three-step stainless steel smelting. Background Art

[0002] The three-step stainless steel smelting process involves first melting ferronickel in an electric furnace and ferrochromium in an intermediate frequency furnace. Second, the molten ferronickel and ferrochromium mother liquors are blended in proportion and poured into an AOD furnace to produce stainless steel mother liquor. Third, the stainless steel mother liquor is poured into an LF or VOD furnace to produce qualified molten stainless steel. For the stainless steel smelting industry, steelmaking is the most costly process in steel production, consuming the highest amount of energy and emitting the highest carbon emissions. As the steel industry transitions to a green and low-carbon future, optimizing the cost-effective batching of steelmaking processes is crucial for promoting green economic operations. By rationally planning the batching of electric furnaces, intermediate frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, stainless steel smelters can achieve economical use of raw materials, auxiliary materials, and energy media, maximizing cost-effectiveness. Scientifically optimized batching planning and adjustments enable stainless steel smelters to achieve environmentally friendly and low-carbon production methods. This contributes to the establishment of a sustainable economic model and promotes the steel industry's development towards a more environmentally friendly, efficient, and sustainable direction.

[0003] In the existing technology, the electric furnace is equipped with an electric furnace charging model, the medium frequency furnace is equipped with an medium frequency furnace charging model, the AOD furnace is equipped with an AOD furnace charging model, the LF furnace is equipped with an LF furnace charging model, and the VOD furnace is equipped with a VOD furnace charging model. These five models operate independently and serve their respective processes. Moreover, these charging models are mainly used to smelt molten steel with qualified composition, and basically do not consider the collaborative costs between processes. Summary of the Invention

[0004] In order to overcome the above-mentioned defects, the purpose of the present invention is to provide a deep learning-based batching model collaborative method, system, equipment and medium for three-step stainless steel smelting.

[0005] To achieve the above-mentioned object, the present invention provides a deep learning-based batching model collaboration method for three-step stainless steel smelting, wherein the method comprises the following steps:

[0006] S1: Obtain historical raw material usage data for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, historical energy medium usage corresponding to the historical raw material usage data, molten steel composition, molten steel temperature, raw material costs, energy medium costs, and other data, construct a data set, and divide the data set into a training set, a test set, and a validation set;

[0007] S2: Obtain the calculation data of raw materials and energy media for the batching models of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and calibrate the calculation data of the batching models of raw materials and energy media using the historical raw materials and energy media usage data;

[0008] S3: Construct a cost-optimized batching model collaboration for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, divide the data set into a training set, a test set, and a validation set, input the data into the model collaboration, train, test, and validate the model, and determine whether the validation result meets the requirements. If so, execute step S4; if not, execute step S1 again.

[0009] S4: Inputting the calibrated raw materials and energy medium batching model calculation data into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and outputting the predicted molten steel composition, molten steel temperature, and batching prediction cost of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace;

[0010] S5: Real-time collection of the current raw and auxiliary materials, energy medium usage data and current raw and auxiliary materials, energy medium costs of electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, and correction of the predicted costs of raw and auxiliary materials and energy mediums based on the current raw and auxiliary materials, energy medium usage data and current raw and auxiliary materials, energy medium costs;

[0011] S6: The cost-optimal batching model for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces based on deep learning is coordinated with the calculated data of corrected raw materials, auxiliary materials, and energy media.

[0012] Furthermore, the raw and auxiliary materials, energy medium usage, and model calculation data include the usage, model calculation data of scrap steel, ferronickel, ferrochrome, iron-containing raw materials, carbon, flux, slag-forming agent, and the energy medium usage, model calculation data of water, electricity, oxygen, nitrogen, argon, and compressed air.

[0013] Furthermore, the raw material and energy medium cost data include cost data of scrap steel, ferronickel, ferrochrome, iron-containing raw materials, carbon, flux, slag-forming agent and energy medium cost data of water, electricity, oxygen, nitrogen, argon and compressed air.

[0014] Furthermore, the step S2 includes the following specific processes:

[0015] S21: Obtain the calculation data of raw materials, auxiliary materials, and energy media of the respective batching models of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace within a specified period;

[0016] S22: Extract the raw material and energy medium usage data of the same period from the historical raw material and energy medium usage data;

[0017] S23: Based on the extracted usage data of raw materials, auxiliary materials, and energy media of the same period, compensation calibration is performed on the calculated data of raw materials, auxiliary materials, and energy media of the corresponding period according to the specified ratio.

[0018] Furthermore, the step S3 includes the following specific processes:

[0019] S31: Construct a cost-optimized batching model for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces;

[0020] S32: Inputting the training set into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace to train the model;

[0021] S33: Input the test set into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, test the trained model, and determine whether the corresponding indicators of the test results meet the requirements. If so, execute step S34; if not, re-input a new training set to train the model;

[0022] S34: Input the verification set into the cost-optimized batching model collaboration of the tested electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace to verify the test results and determine whether the verification results meet the requirements. If so, execute step S4; if not, re-execute step S1.

[0023] Furthermore, in step S5, the process of correcting the predicted cost of raw materials and energy media by using the current raw materials and energy media usage data and the current raw materials and energy media costs is as follows:

[0024] S51: Calculate the cost of raw materials and energy media based on the calibrated calculation data of raw materials and energy media;

[0025] S52: Create a composite function of the predicted composition and temperature of the molten steel and the predicted cost of the batching for the electric furnace, medium frequency furnace, AOD furnace, LF furnace and VOD furnace;

[0026] S53: Calculate the optimal solution for the composite function of the composition, temperature and batching prediction cost of the tapping molten steel for the electric furnace, medium frequency furnace, AOD furnace, LF furnace and VOD furnace;

[0027] S54: Correcting the electric furnace batching model, the medium frequency furnace batching model, and the AOD furnace batching model of the electric furnace, the medium frequency furnace, the AOD furnace, the LF furnace, and the VOD furnace according to the specified ratio through the optimal solution.

[0028] Furthermore, in step S52, the formula for creating a composite function that ensures that the composition and temperature of the molten steel are qualified and the operating cost is optimal is:

[0029] f(x)=f 1EF (x)+f 1MF (x)+f 1AOD (x)+f 1LF (x)+f 1VOD (x)+f 2EF (x)+f 2MF (x)+f 2AOD (x)+f 2LF (x)+f 2VOD (x)+f 3EF (x)+f 3MF (x)+f 3AOD (x)+f 3LF (x)+f 3VOD (x)

[0030] Among them, f 1EF (x) is the composition function of the steel tapped from the electric furnace, f 1MF (x) is the steel composition function of the medium frequency furnace, f 1AOD (x) is the composition function of the steel tapped from the AOD furnace, f 1LF (x) is the composition function of the steel tapped from the LF furnace, f 1VOD (x) is the COD furnace tapping composition function, f 2EF (x) is the tapping temperature function of the electric furnace, f 2MF (x) is the tapping temperature function of the medium frequency furnace, f 2AOD (x) is the tapping temperature function of the AOD furnace, f 2LF (x) is the tapping temperature function of the LF furnace, f 2VOD (x) is the tapping temperature function of the VOD furnace, f 3EF (x) is the cost function of the electric furnace, f 3MF (x) is the cost function of the medium frequency furnace, f 3AOD (x) is the cost function of the AOD furnace, f 3LF (x) is the LF furnace cost function, f 3VOD (x) is the VOD furnace cost function.

[0031] To achieve the above-mentioned purpose, the terminal device of the present invention includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor loads and executes the computer program, it adopts the above-mentioned deep learning-based batching model collaborative method for three-step stainless steel smelting.

[0032] To achieve the above-mentioned objectives, the present invention provides a computer-readable storage medium, in which a computer program is stored. When the computer program is loaded and executed by a processor, the above-mentioned deep learning-based batching model collaborative method for three-step stainless steel smelting is adopted.

[0033] The present invention provides a cost-optimized batching model collaboration based on deep learning for three-step stainless steel smelting, which obtains data such as historical raw and auxiliary material usage, energy medium usage, molten steel composition, molten steel temperature, raw and auxiliary material costs, and energy medium costs of electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces to construct a data set; obtains the calculated data of raw and auxiliary materials and energy media of each batching model of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and calibrates the calculated data of raw and auxiliary materials and energy media through historical raw and auxiliary materials and energy medium usage; constructs a batching model collaboration based on cost optimization for the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, inputs the data set into the model in turn, and calibrates the model Training, testing and verification, and judging whether the verification results meet the requirements. If so, the calibrated raw and auxiliary materials, energy medium batching model calculation data are input into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and the predicted molten steel composition, molten steel temperature, and batching prediction cost of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace are output; the current raw and auxiliary materials, energy medium usage data and the current raw and auxiliary materials, energy medium costs of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace are collected in real time, and the cost-optimized batching model of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace based on deep learning is run in collaboration with the corrected calculation data of raw and auxiliary materials and energy medium.

[0034] On the one hand, the deep learning-based coordination of the cost-optimized batching model enables the complementary batching models of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace to achieve stable molten steel composition and temperature output. On the other hand, it improves the cost-effectiveness of the three-step stainless steel smelting process, effectively reducing smelting costs and increasing economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of a collaborative method for achieving cost-optimal batching models based on deep learning for three-step stainless steel smelting according to the present invention. DETAILED DESCRIPTION

[0036] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0037] In the description of the present invention, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention.

[0038] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0039] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood broadly. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0040] The three-step stainless steel smelting method of the present invention refers to the following steps: first, an electric furnace is used to melt nickel iron and a medium frequency furnace is used to melt chromium iron; second, the molten nickel iron and chromium iron mother liquors are blended in proportion and poured into an AOD furnace to smelt into stainless steel mother liquor; third, the stainless steel mother liquor is poured into an LF furnace or a VOD furnace to smelt into qualified stainless steel liquid. Depending on the raw materials, the electric furnace can independently provide molten steel mother liquor to the AOD furnace, and the medium frequency furnace can also independently provide molten steel mother liquor to the AOD furnace, that is, the first step can be only an electric furnace, or only a medium frequency furnace, or both an electric furnace and a medium frequency furnace. According to the composition and temperature requirements of the final molten steel, the stainless steel mother liquor smelted in the AOD furnace can be provided to either the LF furnace or the VOD furnace, that is, the third step can be either the LF furnace or the VOD furnace.

[0041] This invention aims to achieve cost-optimal batching model collaboration by leveraging deep learning technology to achieve cost-optimal batching model collaboration among the existing EAF batching models, the IF batching models, the AOD batching models, the LF batching models, and the VOD batching models. This allows for cost data sharing and association, thereby coupling the batching models of the EAF, IF, AOD, LF, and VOD furnaces based on a cost-optimal algorithm, achieving cost-optimal batching model collaboration for three-step stainless steel smelting. The aforementioned EAF batching model and IF batching model are optional.

[0042] See attached Figure 1 As shown, the present invention is used for the three-step stainless steel smelting process. The method of achieving the optimal cost-effective batching model based on deep learning includes the following steps:

[0043] S1: Obtain historical raw material usage data for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, historical energy medium usage corresponding to the historical raw material usage data, molten steel composition, molten steel temperature, raw material costs, energy medium costs, and other data, construct a data set, and divide the data set into a training set, a test set, and a validation set;

[0044] S2: Obtain the calculation data of raw materials and energy media for the batching models of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and calibrate the calculation data of the batching models of raw materials and energy media using the historical raw materials and energy media usage data;

[0045] S3: Construct a cost-optimized batching model for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, divide the data set into a training set, a test set, and a validation set, and input them into the model. The model is trained, tested, and validated, and it is determined whether the validation result meets the requirements. If so, execute step S4; if not, execute step S1 again.

[0046] S4: Inputting the calibrated raw materials and energy medium batching model calculation data into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and outputting the predicted molten steel composition, molten steel temperature, and batching prediction cost of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace;

[0047] S5: Real-time collection of the current raw and auxiliary materials, energy medium usage data and current raw and auxiliary materials, energy medium costs of electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, and correction of the predicted costs of raw and auxiliary materials and energy mediums based on the current raw and auxiliary materials, energy medium usage data and current raw and auxiliary materials, energy medium costs;

[0048] S6: The cost-optimal batching model for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces based on deep learning is coordinated with the calculated data of corrected raw materials, auxiliary materials, and energy media.

[0049] Step S1 is mainly to provide a data basis for subsequent model training, testing and verification, so that the model has the learning ability to meet the corresponding requirements. In the data preparation process, it is necessary to obtain comprehensive historical raw material usage data of electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces in various periods, historical energy medium usage corresponding to the historical raw material usage data, molten steel composition, molten steel temperature, raw material cost, energy medium cost and other data. After obtaining the relevant data, it is necessary to pre-process the acquired data, and the pre-processing process includes data cleaning, data conversion and data normalization. The process of data cleaning is mainly to remove invalid data, abnormal data, duplicate data or noise data in the data. The purpose of data conversion is to convert data that is inconvenient to analyze into data that is more convenient to analyze. The purpose of data normalization is to limit the processed data to a certain range and summarize the statistical distribution of unified data samples.

[0050] In step S2, the calculated data for the raw materials, auxiliary materials, and energy media for each batching model of the electric furnace, intermediate frequency furnace, AOD furnace, LF furnace, and VOD furnace is obtained. This calculated data is crucial to the smelting system. The smelting system performs smelting based on the calculated data of the batching model, enabling the smelting system to operate stably and produce molten steel with qualified composition and temperature. However, the calculated data is not completely accurate, and the batching data of the same workshop over a long period of time and within the same cycle has a certain regularity. Therefore, it is of great significance to correct the calculated data of the batching model by obtaining historical raw materials, auxiliary materials, and energy media usage.

[0051] In steps S3 and S4, a machine learning model that collaborates with the cost-optimized batching model is constructed, and the constructed model is trained, tested, and verified through the acquired relevant historical data. This allows the model to have a learning ability that meets certain requirements, can be based on cost optimization, and can achieve the production of qualified products at lower operating costs, thereby achieving lower operating costs and higher economic benefits.

[0052] The cost-optimized batching model constructed above works together to form a convolutional neural network model. The process of building a convolutional neural network model includes constructing convolutional layers, pooling layers, and fully connected layers. At the same time, the corresponding parameters of each constructed layer are set, including the size and step size of the convolution kernel. At the same time, the activation function and loss function must be selected, and the optimizer must be set.

[0053] In one embodiment, the first convolutional layer has 4 input channels and 8 output channels. The filter / kernel sliding step size in the convolutional layer is 1. During the convolution operation, extra data of 0 is added to the edges of the input data. The maximum pooling size is 5. The second convolutional layer has 8 input channels and 18 output channels. The filter / kernel sliding step size in the convolutional layer is 1. During the convolution operation, extra data of 0 is added to the edges of the input data. The maximum pooling size is 5. The third fully connected layer has 324 input channels and 135 output channels. The fourth fully connected layer has 135 input channels and 96 output channels. The output layer has 96 fully connected channels and 12 output channels. The Sigmoid activation function is used as the activation function, and the Manhattan distance loss function is used as the loss function.

[0054] During model training of a convolutional neural network, the model output is calculated using forward propagation, and the error between the predicted value and the true value is calculated using a loss function. Then, the gradient is calculated using a backpropagation algorithm, and the network parameters are updated using an optimizer to minimize the loss function. This process is repeated multiple times until the preset number of iterations is reached or the loss function reaches a low value. During training, hyperparameters such as the learning rate need to be adjusted to balance the model's training speed and generalization ability. The optimizer uses the Adam optimization algorithm to update the model parameters.

[0055] Step S2 includes the following specific processes:

[0056] S21: Obtain the calculation data of raw materials, auxiliary materials, and energy media of the respective batching models of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace within a specified period;

[0057] S22: Extract the raw material and energy medium usage data of the same period from the historical raw material and energy medium usage data;

[0058] S23: Based on the extracted usage data of raw materials, auxiliary materials, and energy media of the same period, compensation calibration is performed on the calculated data of raw materials, auxiliary materials, and energy media of the corresponding period according to the specified ratio.

[0059] Step S3 includes the following specific processes:

[0060] S31: Construct a cost-optimized batching model for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces;

[0061] S32: Inputting the training set into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace to train the model;

[0062] S33: Input the test set into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, test the trained model, and determine whether the corresponding indicators of the test results meet the requirements. If so, execute step S34; if not, re-input a new training set to train the model;

[0063] S34: Input the verification set into the cost-optimized batching model collaboration of the tested electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace to verify the test results and determine whether the verification results meet the requirements. If so, execute step S4; if not, re-execute step S1.

[0064] In step S5, the process of correcting the predicted cost of raw materials and energy media by using the current raw materials and energy media usage data and the current raw materials and energy media costs is as follows:

[0065] S51: Calculate the cost of raw materials and energy media based on the calibrated calculation data of raw materials and energy media;

[0066] S52: Create a composite function of the predicted composition and temperature of the molten steel and the predicted cost of the batching for the electric furnace, medium frequency furnace, AOD furnace, LF furnace and VOD furnace;

[0067] S53: Calculate the optimal solution for the composite function of the composition, temperature and batching prediction cost of the tapping molten steel for the electric furnace, medium frequency furnace, AOD furnace, LF furnace and VOD furnace;

[0068] S54: Correcting the electric furnace batching model, the medium frequency furnace batching model, and the AOD furnace batching model of the electric furnace, the medium frequency furnace, the AOD furnace, the LF furnace, and the VOD furnace according to the specified ratio through the optimal solution.

[0069] Preferably, the formula for creating a composite function in step S52 to ensure that the composition and temperature of the molten steel are qualified and the operating cost is optimal is:

[0070] f(x)=f 1EF (x)+f 1MF (x)+f 1AOD (x)+f 1LF (x)+f 1VOD (x)+f 2EF (x)+f 2MF (x)+f 2AOD (x)+f 2LF (x)+f 2VOD (x)+f 3EF (x)+f 3MF (x)+f 3AOD (x)+f 3LF (x)+f3VOD (x)

[0071] Among them, f 1EF (x) is the composition function of the steel tapped from the electric furnace, f 1MF (x) is the steel composition function of the medium frequency furnace, f 1AOD (x) is the composition function of the steel tapped from the AOD furnace, f 1LF (x) is the composition function of the steel tapped from the LF furnace, f 1VOD (x) is the COD furnace tapping composition function, f 2EF (x) is the tapping temperature function of the electric furnace, f 2MF (x) is the tapping temperature function of the medium frequency furnace, f 2AOD (x) is the tapping temperature function of the AOD furnace, f 2LF (x) is the tapping temperature function of the LF furnace, f 2VOD (x) is the tapping temperature function of the VOD furnace, f 3EF (x) is the cost function of the electric furnace, f 3MF (x) is the cost function of the medium frequency furnace, f 3AOD (x) is the cost function of the AOD furnace, f 3LF (x) is the LF furnace cost function, f 3VOD (x) is the VOD furnace cost function.

[0072] In summary, the present invention provides a cost-optimal batching model collaboration method based on deep learning for three-step stainless steel smelting, which obtains data such as historical raw and auxiliary material usage, energy medium usage, molten steel composition, molten steel temperature, raw and auxiliary material costs, and energy medium costs of electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces to construct a data set; obtains the calculation data of raw and auxiliary materials and energy media of each batching model of electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, and calibrates the calculation data of raw and auxiliary materials and energy media through historical raw and auxiliary materials and energy medium usage; constructs a batching model collaboration based on cost optimization for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, inputs the data sets into the model in turn, and calibrates the model The model is trained, tested and verified, and it is judged whether the verification result meets the requirements. If so, the calibrated raw and auxiliary materials, energy medium batching model calculation data are input into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and the predicted molten steel composition, molten steel temperature, and batching prediction cost of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace are output; the current raw and auxiliary materials, energy medium usage data and the current raw and auxiliary materials, energy medium costs of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace are collected in real time, and the cost-optimized batching model of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace based on deep learning is run with the corrected calculation data of raw and auxiliary materials and energy medium.

[0073] An embodiment of the present application also discloses a terminal device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor adopts the deep learning-based ingredient model collaborative method for three-step stainless steel smelting of the above embodiment when executing the computer program.

[0074] Among them, the terminal device can be a computer device such as a desktop computer, a laptop computer or a cloud server, and the terminal device includes but is not limited to a processor and a memory. For example, the terminal device can also include input and output devices, network access devices and buses, etc.

[0075] Among them, the processor can adopt a central processing unit (CPU). Of course, according to actual usage, other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be adopted. The general-purpose processor can adopt a microprocessor or any conventional processor, etc., and this application does not impose any restrictions on this.

[0076] Among them, the memory can be an internal storage unit of the terminal device, such as the hard disk or memory of the terminal device, or it can be an external storage device of the terminal device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD) or flash memory card (FC) equipped on the terminal device, etc., and the memory can also be a combination of the internal storage unit and the external storage device of the terminal device. The memory is used to store computer programs and other programs and data required by the terminal device. The memory can also be used to temporarily store data that has been output or is to be output. This application does not impose any restrictions on this.

[0077] Among them, through this terminal device, the deep learning-based ingredient model collaborative method for three-step stainless steel smelting of the above embodiment is stored in the memory of the terminal device, and is loaded and executed on the processor of the terminal device for user convenience.

[0078] An embodiment of the present application further discloses a computer-readable storage medium, and the computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the deep learning-based batching model collaborative method for three-step stainless steel smelting of the above embodiment is adopted.

[0079] Among them, the computer program can be stored in a computer-readable medium, the computer program includes computer program code, the computer program code can be in the form of source code, object code, executable file or certain middleware, etc. The computer-readable medium includes any entity or device that can carry computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that computer-readable medium includes but is not limited to the above-mentioned components.

[0080] Among them, through this computer-readable storage medium, the deep learning-based ingredient model collaborative method for three-step stainless steel smelting of the above embodiment is stored in the computer-readable storage medium, and is loaded and executed on the processor to facilitate the storage and application of the deep learning-based ingredient model collaborative method for three-step stainless steel smelting.

[0081] The present invention has been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above-described embodiments. Various modifications may be made within the scope of knowledge of those skilled in the art without departing from the spirit of the present invention. Many other changes and modifications that do not depart from the spirit and scope of the present invention should be considered within the scope of protection of the present invention.

[0082] In the description of this specification, specific features, structures, materials or characteristics may be combined in an appropriate manner in any one or more embodiments or examples.

[0083] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A deep learning-based batching model collaboration method for three-step stainless steel smelting, characterized by: The method comprises the following steps: S1: Obtain historical raw material usage data for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, historical energy medium usage corresponding to the historical raw material usage data, molten steel composition, molten steel temperature, raw material costs, and energy medium costs, construct a dataset, and divide the dataset into a training set, a test set, and a validation set; S2: Obtain the calculation data of raw materials and energy media for the batching models of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and calibrate the calculation data of the batching models of raw materials and energy media using the historical raw materials and energy media usage data; S3: Construct a cost-optimized batching model collaboration for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, divide the data set into a training set, a test set, and a validation set, input the data into the model collaboration, train, test, and validate the model, and determine whether the validation result meets the requirements. If so, execute step S4; if not, execute step S1 again. S4: Inputting the calibrated raw materials and energy medium batching model calculation data into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, and outputting the predicted molten steel composition, molten steel temperature, and batching prediction cost of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace; S5: Real-time collection of the current raw and auxiliary materials, energy medium usage data and current raw and auxiliary materials, energy medium costs of electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces, and correction of the predicted costs of raw and auxiliary materials and energy mediums based on the current raw and auxiliary materials, energy medium usage data and current raw and auxiliary materials, energy medium costs; S6: The cost-optimal batching model for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces based on deep learning works in conjunction with the calculated data of the corrected raw materials, auxiliary materials, and energy media; Wherein, the step S2 includes the following specific processes: S21: Obtain the calculation data of raw materials, auxiliary materials, and energy media of the respective batching models of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace within a specified period; S22: Extract the raw material and energy medium usage data of the same period from the historical raw material and energy medium usage data; S23: Based on the extracted usage data of raw materials, auxiliary materials, and energy media of the same period, compensation calibration is performed on the calculated data of raw materials, auxiliary materials, and energy media of the corresponding period according to the specified ratio.

2. The deep learning-based batching model collaborative method for three-step stainless steel smelting according to claim 1, characterized in that: The cost data of raw materials and energy media include the cost data of scrap steel, nickel iron, chromium iron, iron-containing raw materials, carbon, flux, slag-forming agent and the cost data of energy media such as water, electricity, oxygen, nitrogen, argon and compressed air.

3. The deep learning-based batching model collaborative method for three-step stainless steel smelting according to claim 1, characterized in that: The step S3 includes the following specific processes: S31: Construct a cost-optimized batching model for electric furnaces, medium frequency furnaces, AOD furnaces, LF furnaces, and VOD furnaces; S32: Inputting the training set into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace to train the model; S33: Input the test set into the cost-optimized batching model collaboration of the electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace, test the trained model, and determine whether the corresponding indicators of the test results meet the requirements. If so, execute step S34; if not, re-input a new training set to train the model; S34: Input the verification set into the cost-optimized batching model collaboration of the tested electric furnace, medium frequency furnace, AOD furnace, LF furnace, and VOD furnace to verify the test results and determine whether the verification results meet the requirements. If so, execute step S4; if not, re-execute step S1.

4. The deep learning-based batching model collaborative method for three-step stainless steel smelting according to claim 1, characterized in that: In step S5, the process of correcting the predicted cost of raw materials and energy media by using the current raw materials and energy media usage data and the current raw materials and energy media costs is as follows: S51: Calculate the cost of raw materials and energy media based on the calibrated calculation data of raw materials and energy media; S52: Create a composite function of the predicted composition and temperature of the molten steel and the predicted cost of the batching for the electric furnace, medium frequency furnace, AOD furnace, LF furnace and VOD furnace; S53: Calculate the optimal solution for the composite function of the composition, temperature and batching prediction cost of the tapping molten steel for the electric furnace, medium frequency furnace, AOD furnace, LF furnace and VOD furnace; S54: Correcting the electric furnace batching model, the medium frequency furnace batching model, and the AOD furnace batching model of the electric furnace, the medium frequency furnace, the AOD furnace, the LF furnace, and the VOD furnace according to the specified ratio through the optimal solution.

5. The deep learning-based batching model collaborative method for three-step stainless steel smelting according to claim 4, characterized in that: The formula for creating a composite function in step S52 to ensure that the composition and temperature of the molten steel are qualified and the operating cost is optimal is: f(x)=f 1EF (x)+f 1MF (x)+f 1AOD (x)+f 1LF (x)+f 1VOD (x)+f 2EF (x)+f 2MF (x)+f 2AOD (x)+f 2LF (x)+f 2VOD (x)+f 3EF (x)+f 3MF (x)+f 3AOD (x)+f 3LF (x)+f 3VOD (x) Among them, f 1EF (x) is the composition function of the steel tapped from the electric furnace, f 1MF (x) is the steel composition function of the medium frequency furnace, f 1AOD (x) is the composition function of the steel tapped from the AOD furnace, f 1LF (x) is the composition function of the steel tapped from the LF furnace, f 1VOD (x) is the COD furnace tapping composition function, f 2EF (x) is the tapping temperature function of the electric furnace, f 2MF (x) is the tapping temperature function of the medium frequency furnace, f 2AOD (x) is the tapping temperature function of the AOD furnace, f 2LF (x) is the tapping temperature function of the LF furnace, f 2VOD (x) is the tapping temperature function of the VOD furnace, f 3EF (x) is the cost function of the electric furnace, f 3MF (x) is the cost function of the medium frequency furnace, f 3AOD (x) is the cost function of the AOD furnace, f 3LF (x) is the LF furnace cost function, f 3VOD (x) is the VOD furnace cost function.

6. The deep learning-based batching model collaborative method for three-step stainless steel smelting according to claim 1, characterized in that: The three-step stainless steel smelting method includes: the first step of melting ferronickel in an electric furnace and melting ferrochrome in a medium frequency furnace; the second step of blending the molten ferronickel and ferrochrome mother liquors in proportion and pouring them into an AOD furnace to smelt stainless steel mother liquor; and the third step of pouring the stainless steel mother liquor into an LF furnace or a VOD furnace to smelt qualified stainless steel liquid.

7. A terminal device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: When the processor loads and executes the computer program, the deep learning-based batching model collaborative method for three-step stainless steel smelting described in claim 1 is adopted.

8. A computer-readable storage medium storing a computer program, wherein: When the computer program is loaded and executed by the processor, the deep learning-based batching model collaborative method for three-step stainless steel smelting described in claim 1 is adopted.

Images