A Rhythm-Aware Co-prediction Method for Steam Generation and Consumption in Steelmaking and Refining Processes

Abstract

This application relates to the field of iron and steel metallurgical technology and provides a rhythm-aware method for collaborative prediction of steam production and consumption in steelmaking-refining processes. The method includes: firstly, acquiring historical production data of the steelmaking-refining processes; secondly, constructing a spatiotemporal graph network model to extract dynamic logistics constraint features of equipment such as charging cranes, and predicting the start time and operation duration of converter blowing and subsequent refining processes; subsequently, constructing production and consumption quantification models: inputting metallurgical reaction physical features such as carbon-temperature balance factors into a multi-scale attention prediction network to calculate the total steam recovery of a single converter furnace; simultaneously, calculating the steam consumption of the refining process based on the stage rated steam consumption rate of the refining vacuum pump group operation procedure; mapping the total converter steam recovery and the refining process steam consumption to minute-level steam production flow curves and minute-level steam consumption flow curves, respectively, and performing algebraic superposition on a unified time axis to generate a minute-level net steam flux prediction curve.

Description

Technical Field

[0001] This application belongs to the field of iron and steel metallurgy technology, and in particular relates to a method for predicting steam production and consumption in steelmaking and refining processes based on rhythm perception. Background Technology

[0002] The steel industry is a typical process manufacturing industry, with its production processes involving the transformation of material and energy flows. In the converter steelmaking process, recovering heat from high-temperature flue gas to produce steam via waste heat boilers is a key technology for achieving energy conservation and emission reduction in enterprises. Converter steelmaking exhibits significant intermittent and batch production characteristics. A single smelting cycle typically includes multiple discrete stages such as ferroalloy addition, scrap steel addition, blowing, and tapping. Only during the blowing stage, due to the exothermic reaction of carbon and oxygen, is a large amount of recoverable steam generated. In other non-blowing stages, steam production drops sharply or even to zero. Furthermore, the subsequent refining process, as the main steam consumption unit in the region, operates closely following the converter, exhibiting intermittent steam demand.

[0003] Currently, for the prediction of such discontinuous energy flows, industrial sites mainly adopt the following two types of technical solutions:

[0004] The first type is static statistical estimation: dispatchers estimate the total daily steam recovery based on the number of smelting furnaces planned for the shift, combined with empirical steam production coefficients per ton of steel, and simply assume a uniform production rhythm. This method only provides a macroscopic total reference, cannot reflect fluctuations in individual furnace runs, and often ignores the dynamic consumption of the refining process. The second type is short-term forecasting based on time series: with the popularization of industrial big data, some technical solutions have begun to utilize historical data collected by flow meters, employing intelligent algorithms such as Time Series Models (ARIMA) or Long Short-Term Memory Neural Networks (LSTM) for sliding window or time series data forecasting. This method essentially uses the inertia of historical data to extrapolate future flow values, and cannot accurately describe the intermittent characteristics of converter steam recovery and consumption caused by production rhythm.

[0005] However, the above solutions lack an understanding of production rhythm: converter production is not an isolated operation; its production rhythm is rigidly constrained by multiple logistical factors, such as the supply of molten iron at the front end and the casting speed of the continuous casting machine at the back end. Existing technologies often treat each process as an independent time series, ignoring the dynamic spatiotemporal topology and queuing constraints between the converter and the charging trolley, molten iron ladle, and other transfer equipment. This neglect of logistical coupling results in the model being unable to perceive why the previous heat was delayed or when the current heat is ready for blowing, thus failing to accurately predict the precise start time of the next blowing cycle, often leading to prediction lag. Furthermore, the decisive influence of metallurgical mechanisms on total production and consumption is ignored: steam generation is essentially an exothermic process of carbon and silicon oxidation in the converter pool. The total amount of steam recovered is closely related to the physicochemical sensible heat of the molten iron entering the furnace, the proportion of scrap steel added, and the oxygen supply intensity. Existing technologies mostly use black-box models, focusing only on the numerical patterns of historical flow data, lacking in-depth exploration of the aforementioned metallurgical mechanism characteristics. When raw material conditions fluctuate, such as a high carbon content in a batch of molten iron or a temporary increase in the scrap ratio, simple time-series models cannot detect changes in operating conditions, and the output predictions often deviate significantly from the actual recovery volume. There is a lack of a holistic regional balance perspective: production scheduling based on daily output and order information is often affected by fluctuations in operating conditions, adjustments to raw materials and fuels, and the operational efficiency of production personnel, resulting in time deviations and severe lags in communication with the energy system. This physical and temporal disconnect between business systems means that existing energy forecasting schemes often focus only on the steam production of converters in isolation, failing to align and synchronously predict the steam consumption of refining processes. Because the dispatch center cannot grasp the real net demand of the region on the main pipeline network in real time, it is difficult to formulate optimal pipeline scheduling and accumulator charging / discharging strategies. There is a lack of regional-level source-load collaborative analysis capabilities: existing technologies typically only predict steam recovery or steam consumption separately, lacking the ability to uniformly model and collaboratively analyze the relationship between steam generation, steam consumption, and net flux within the production area, making it difficult to achieve forward-looking predictions of dynamic source-load balance within the region. Summary of the Invention

[0006] This application provides a method and related equipment for predicting the combined steam production and consumption of steelmaking and refining processes based on rhythm perception, which can solve the problems of rhythm prediction lag, lack of mechanism characteristics, and difficulty in quantifying production and consumption balance in the prior art.

[0007] In a first aspect, embodiments of this application provide a method for predicting the coordinated steam production and consumption of steelmaking-refining processes based on rhythm perception, including: Historical production performance data of the process corresponding to the converter in the steelmaking-refining process is obtained, and based on the historical production performance data, a mechanism characteristic group including the carbon temperature balance factor, an operating condition characteristic group, and a time sequence characteristic group including the logistics queuing delay characteristic are determined. The logistics queuing delay characteristic is the time difference between the blowing start time of any manufacturing order and the tapping end time of the previous manufacturing order. The transportation equipment corresponding to the converter in the steelmaking-refining process is taken as nodes. The operating status data of each node is obtained to construct a dynamic feature matrix based on the operating status data. The operating time data of each node is also obtained to construct a dynamic adjacency matrix based on the operating time data and the logistics queuing delay characteristics. The dynamic feature matrix and the dynamic adjacency matrix are input into a preset prediction model to obtain the converter blowing time window and the refining operation time window; Based on the mechanism characteristic group, operating condition characteristic group and time series characteristic group, calculate the total steam recovery of a single converter furnace. The steam consumption of the refining process is calculated based on the stage rated steam consumption rate of the refining vacuum pump unit according to the operating procedure and the refining steam consumption time mask based on the refining operation time window. The total amount of steam recovered is mapped to the converter blowing time window to obtain a minute-level steam production flow curve, and the steam consumption is mapped to the refining operation time window to obtain a minute-level steam consumption flow curve, thereby generating a minute-level net steam flux prediction curve for future scheduling cycles.

[0008] Optionally, the step of determining the mechanistic characteristic group including the carbon-temperature balance factor based on the historical production performance data includes: Based on the principle of heat balance in the converter molten pool, a carbon-temperature balance factor reflecting the steam generation potential is constructed using iron composition and temperature data. The mechanistic characteristic set of the carbon-temperature balance factor is obtained, and its mathematical characterization is as follows:

[0009] In the formula, where This indicates the amount of molten iron added, in kg; , These represent the carbon content and silicon content of molten iron, respectively. , These represent the exothermic enthalpy values ​​of the oxidation reactions of carbon and silicon, respectively, in kJ / kg; Specific heat capacity of molten iron, kJ / (kg·℃); , These represent the initial temperature of molten iron entering the furnace and the reference temperature for the temperature drop of molten iron specified in the process, respectively, in °C; This indicates the amount of scrap steel loaded, expressed in kg.

[0010] Optionally, the step of calculating the total steam recovery of a single converter cycle based on the mechanism characteristic group, operating condition characteristic group, and time series characteristic group includes: The mechanism feature group, the working condition feature group, and the time series feature group are concatenated to construct a multidimensional input sequence; The multidimensional input sequence is transformed into a feature vector matrix through a linear projection layer and a position encoding module; The feature vector matrix is ​​input into a Transformer encoder containing multiple cascaded layers to map the feature vector matrix into a query matrix, a key matrix, and a value matrix; Based on the query matrix, key matrix, and value matrix, encoded data is obtained through a self-attention module for executing the self-attention mechanism with residual connections and a feedforward neural network with a linear layer of ReLU activation function; The aggregated feature representation of the top layer of the Transformer encoder is taken, and the total steam recovery of a single converter furnace is output through a linear regression mapping layer.

[0011] Optionally, the step of calculating the steam consumption of the refining process based on the rated steam consumption rate of the refining vacuum pump unit according to the stage of the operating procedure and the refining steam consumption time mask based on the refining operation time window includes: The single refining operation is divided into standard process stages. Based on the equipment's factory parameters and preset operating procedures, the rated steam consumption rate of different stages is extracted, and the relative standard operating time intervals of different stages are set. Based on the rated steam consumption rate at different stages and the relative standard operating time interval at different stages, a piecewise function of steam consumption rate based on the relative operating time of refining is established. Based on the refining steam consumption time mask of the refining operation time window and the start time in the refining operation time window, a minute-level refining steam consumption curve is calculated and generated in the future scheduling cycle to determine the steam consumption of the refining process.

[0012] Optionally, the step of mapping the total steam recovery to a converter blowing time window includes: Based on the principle of energy conservation, the total amount of steam recovered is used as the waveform integral area constraint. The benchmark steam production waveform function is obtained by statistical analysis of high-frequency performance data based on historical real operating conditions. The converter steam production time mask based on the converter blowing time window is used as a gating switch. The instantaneous steam production flow rate containing the fluctuation of real operating conditions is back-reassured and reconstructed through area normalization. The total amount of steam recovered is then mapped to the converter blowing time window to obtain the minute-level steam production flow rate curve.

[0013] Optionally, the step of mapping the total steam recovery to a converter blowing time window and the steam consumption to a refining operation time window to generate a minute-level net steam flux prediction curve for future scheduling cycles includes: The steam consumption is mapped to the refining operation time window to obtain a minute-level steam flow rate curve; On a unified global time axis, the minute-level steam production flow curve and the minute-level steam consumption flow curve are algebraically superimposed and offset to determine the net flux demand, so as to obtain the minute-level net steam flux prediction curve for future scheduling cycles.

[0014] Optionally, the method further includes: Based on the minute-level net steam flux prediction curve for the future scheduling cycle, a dynamic routing interaction strategy between the converter area and the main pipeline network is dynamically generated.

[0015] On the other hand, this application provides a rhythm-aware steam production and consumption co-prediction device for steelmaking-refining processes, characterized in that the rhythm-aware steam production and consumption co-prediction device for steelmaking-refining processes includes: The feature extraction module is used to acquire historical production performance data of the corresponding process of the converter in the steelmaking-refining process, and to determine the mechanism feature group, the working condition feature group, and the time sequence feature group including the carbon temperature balance factor, based on the historical production performance data. The logistics queuing delay feature is the time difference between the blowing start time of any manufacturing order and the tapping end time of the previous manufacturing order. The dynamic graph construction module is used to take the transportation equipment corresponding to the converter process in the steelmaking-refining process as nodes, obtain the operating status data of each node, construct a dynamic feature matrix based on the operating status data, and obtain the operating time data of each node, construct a dynamic adjacency matrix based on the operating time data and the logistics queuing delay characteristics. The time window prediction module is used to input the dynamic feature matrix and the dynamic adjacency matrix into a preset prediction model to obtain the converter blowing time window and the refining operation time window; The recovery prediction module is used to calculate the total steam recovery of a single converter furnace based on the mechanism characteristic group, operating condition characteristic group and time series characteristic group. The consumption prediction module is used to calculate the steam consumption of the refining process based on the rated steam consumption rate of the refining vacuum pump group according to the stage of the operating procedure and the refining steam consumption time mask based on the refining operation time window. The mapping module is used to map the total steam recovery to the converter blowing time window to obtain a minute-level steam production flow curve, and to map the steam consumption to the refining operation time window to obtain a minute-level steam consumption flow curve, thereby generating a minute-level net steam flux prediction curve for the future scheduling cycle.

[0016] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the rhythm-aware steam production and consumption co-prediction method for steelmaking-refining processes as described in any of the first aspects above.

[0017] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the rhythm-aware steam production and consumption co-prediction method for steelmaking-refining processes as described in any one of the first aspects above.

[0018] Fifthly, embodiments of this application provide a computer program product that, when run on a terminal device, causes the terminal device to execute the rhythm-aware steam production and consumption co-prediction method for steelmaking-refining processes as described in any of the first aspects above.

[0019] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0020] The beneficial effects of this application's embodiments compared to existing technologies are as follows: First, the introduction of a spatiotemporal graph neural network fully considers the dynamic topological constraints of logistics equipment, effectively overcoming the time lag of production cycle fluctuations and achieving advanced and accurate positioning of the operation time windows of each process. Second, metallurgical mechanism constraints such as carbon-temperature balance factors are explicitly embedded into the Transformer multi-scale attention network, forming a dual-driven mode of metallurgical mechanism and data, which can maintain high predictive stability and generalization ability even under extreme disturbance conditions. More importantly, this application innovatively proposes a production consumption model based on time-domain gating. The waveform reconstruction mechanism performs precise spatiotemporal alignment and integral mapping of the macroscopic total production and consumption forecast, successfully reconstructing a minute-level high-frequency instantaneous flow curve that restores the fluctuations of the actual operating conditions. This solves the technical defect of traditional models that can only output scalar totals but cannot characterize continuous flow states. Finally, by performing algebraic superposition calculations on a unified time axis for production and consumption, the region-level net energy boundary is accurately quantified. The output dynamic net steam flux curve provides extremely reliable panoramic feedforward decision support for the plant's EMS system to proactively execute operations such as accumulator charging and discharging, peak shaving and valley filling, and pressure stabilization and supply assurance. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of the method for predicting steam production and consumption in steelmaking-refining processes based on rhythm perception, provided in an embodiment of this application. Figure 2 This is a flowchart of another method for predicting steam production and consumption in the steelmaking-refining process based on rhythm perception, provided in an embodiment of this application. Figure 3 This is a schematic diagram of the spatiotemporal graph neural network prediction architecture and temporal mask generation in the rhythm-aware-based steam production and consumption collaborative prediction method for steelmaking-refining processes provided in the embodiments of this application; Figure 4 This is a schematic diagram of the Transformer-based total steam output prediction model structure in the rhythm-aware steam production and consumption co-prediction method for steelmaking-refining processes provided in the embodiments of this application. Figure 5 This is a schematic diagram of the generation of production and consumption waveform reconstruction and interaction strategy based on time-domain gating mechanism in the rhythm-aware steam production and consumption collaborative prediction method for steelmaking-refining process provided in the embodiments of this application; Figure 6 This is a schematic diagram of the structure of the rhythm-aware steam production and consumption co-prediction device for steelmaking-refining processes provided in the embodiments of this application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application; Figure 8 This is a schematic diagram of the structure of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation

[0023] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0024] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0025] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0026] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [the described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [the described condition or event] is detected," or "in response to detection of [the described condition or event]."

[0027] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0029] like Figure 1 and Figure 2 As shown in the embodiments of this application, a method for predicting the combined steam production and consumption of steelmaking and refining processes based on rhythm perception is provided, including: S101. Obtain historical production performance data of the process corresponding to the converter in the steelmaking-refining process, and determine the mechanism characteristic group, the working condition characteristic group, and the time sequence characteristic group including the carbon temperature balance factor, based on the historical production performance data. The logistics queuing delay characteristic is the time difference between the blowing start time of any manufacturing order and the tapping end time of the previous manufacturing order. For example, multi-source heterogeneous production data from the steelmaking-refining process is acquired, cleaned, and aligned. Historical production performance data for the converter and related processes is collected through data interfaces between the steel enterprise's manufacturing execution system and basic automation system. This historical production performance data includes, but is not limited to: Production rhythm characteristics include data such as the start time of molten iron mixing, the start time of blowing, and the end time of tapping. Raw material characteristics include: scrap steel weight, molten iron addition, oxygen blowing volume during the process, and steel grade number. Composition and state characteristics: data such as molten iron content, molten iron temperature, scrap steel composition characteristics, target tapping temperature, and target molten steel composition constraints; Target tag data: The actual total amount of steam recovered and the actual amount of refining consumption corresponding to the manufacturing command number.

[0030] The collected data is cleaned to remove missing and abnormal extreme values ​​from the sensors. Using the manufacturing command number as the primary key, the multi-source heterogeneous data with different sampling frequencies are correlated and aligned to construct a standardized model training dataset.

[0031] S102. Take the transportation equipment corresponding to the converter process in the steelmaking-refining process as nodes, obtain the operating status data of each node, construct a dynamic feature matrix based on the operating status data, and obtain the operating time data of each node, construct a dynamic adjacency matrix based on the operating time data and the logistics queuing delay characteristics. For example, the converter workshop is modeled as a dynamic graph structure. To predict the start and end times of processes from the next manufacturing order, including: Graph Node Topology and Feature Matrix Construction: Key logistics equipment such as converters, charging trolleys, molten iron ladles, and torpedo ladle cars are defined as sets of graph nodes. For each node, not only are equipment attributes assigned, but component data and current production condition data are also extracted and concatenated to construct a dynamic feature matrix for each node. .

[0032] Definition of dynamic adjacency matrix: edge set Characterize the interaction relationships between devices. Extract the duration of molten iron mixing, the torpedo ladle transportation time, and calculate the logistics queuing delay. To accurately quantify the real-time changes in logistics constraints between equipment, this application innovatively constructs a dynamic adjacency matrix. To represent the set of edges The weighting is determined by the timing constraints between devices. Higher-order devices are assigned weights when there are tight timing constraints; weights decay when devices are in a loose waiting state. The specific selection logic is as follows: Strong connection state ( =1): When node With nodes When physical interaction is taking place, such as when a crane is feeding iron into a converter or when a ladle is pouring molten iron into a converter, it is considered a strong logistics constraint and its weight is set to 1. Weak connection state ( =0.5): When node With nodes When there is no physical interaction, but it is in the process waiting sequence, such as when a crane is queuing up to transport molten iron in the converter span, it is judged as a potential constraint and the weight is set to 0.5. No connection ( =0): When the device is idle or serving other unrelated processes, the weight is set to 0.

[0033] Through a dynamic mapping mechanism, the model can capture the logistics congestion topology of the workshop in real time at every time t, rather than relying solely on static physical connections.

[0034] S103. Input the dynamic feature matrix and the dynamic adjacency matrix into the preset prediction model to obtain the converter blowing time window and the refining operation time window; For example, such as Figure 3 As shown, the model inference calculation involves the node feature matrix. With dynamic adjacency matrix Input a pre-trained ST-GNN model. First, use spatial graph convolutional layers to aggregate the collaborative influence features of multiple device nodes at the current time along the topological graph; second, use a temporal convolutional network to capture the sequential evolution pattern of historical production beats along the time axis.

[0035] Prediction output: The output target of the fully connected layer of the model is strictly limited to the time dimension, that is, to predict the start time of the converter blowing under the next manufacturing command. Duration of blowing Based on this, and combined with the standard timing of the refining process, the operation time window for subsequent refining processes is deduced.

[0036] Time mask generation: Based on the predicted start and end times, on a unified continuous time axis. Construct a step function. Generate converter steam production time masks respectively. Refining steam consumption time mask Within the predicted operation time window, the mask function value is 1, and the value is 0 for other time periods, thus realizing the mathematical serialization expression of discrete production cycle time.

[0037] S104. Calculate the total steam recovery of a single converter furnace based on the mechanism characteristic group, operating condition characteristic group, and time sequence characteristic group. S105. Calculate the steam consumption of the refining process based on the stage rated steam consumption rate of the refining vacuum pump group according to the operating procedure and the refining steam consumption time mask based on the refining operation time window. S106. Map the total amount of steam recovered to the converter blowing time window to obtain a minute-level steam production flow curve, and map the steam consumption to the refining operation time window to obtain a minute-level steam consumption flow curve, thereby generating a minute-level net steam flux prediction curve for the future scheduling cycle.

[0038] For example, based on the minute-level steam production flow curve and the minute-level steam consumption flow curve, the steam production flow value and steam consumption flow value at the same moment are obtained respectively. The net steam flux at the current moment is calculated based on the steam production flow value and the steam consumption flow value. As time changes, the net steam flux at different moments is obtained, and then the minute-level net steam flux prediction curve for the future scheduling cycle is generated.

[0039] In one possible implementation, the step of determining the mechanistic characteristic set including the carbon-temperature balance factor based on the historical production performance data includes: Based on the principle of heat balance in the converter molten pool, a carbon-temperature balance factor reflecting the steam generation potential is constructed using iron composition and temperature data. The mechanistic characteristic set of the carbon-temperature balance factor is obtained, and its mathematical characterization is as follows:

[0040] In the formula, where This indicates the amount of molten iron added, in kg; , These represent the carbon content and silicon content of molten iron, respectively. , These represent the exothermic enthalpy values ​​of the oxidation reactions of carbon and silicon, respectively, in kJ / kg; Specific heat capacity of molten iron, kJ / (kg·℃); , These represent the initial temperature of molten iron entering the furnace and the reference temperature for the temperature drop of molten iron specified in the process, respectively, in °C; This indicates the amount of scrap steel loaded, expressed in kg.

[0041] For example, a composite feature engineering approach is constructed to extract production cycle characteristics and metallurgical thermodynamic mechanism characteristics. Based on the acquired historical production performance data, the following key features are constructed to enhance the model's ability to perceive physical laws: Logistics queuing delay characteristics: Calculate the time difference between the start time of the current manufacturing order i and the end time of the tapping of the previous manufacturing order i-1. :

[0042] This feature reflects the constraints that logistical factors such as train scheduling congestion and delayed molten iron supply place on the production rhythm.

[0043] A carbon-temperature balance factor is constructed for total volume prediction: Based on the principle of converter molten pool heat balance, a carbon-temperature balance factor reflecting the steam generation potential is constructed using molten iron composition and temperature data. This factor can be calculated using, but is not limited to, the following formula:

[0044] In the formula, the numerator represents the initial total heat carried into the converter by the molten iron. This indicates the amount of molten iron added, in kg; , These represent the carbon content and silicon content of molten iron, respectively. , These represent the exothermic enthalpy values ​​of the oxidation reactions of carbon and silicon, respectively, in kJ / kg; Specific heat capacity of molten iron, kJ / (kg·℃); , Represent the initial temperature of molten iron entering the furnace and the reference temperature for the temperature drop of molten iron specified by the process, respectively, in °C; denominator This represents the amount of scrap steel loaded, in kg. Energy equivalence transformation of heterogeneous variables was achieved in this dimension. The larger the factor value, the more excess heat is represented in the molten pool, and the greater the theoretical potential for it to escape with the high-temperature flue gas and be converted into steam.

[0045] In one possible implementation, the step of calculating the total steam recovery of a single converter cycle based on the mechanism characteristic group, operating condition characteristic group, and time series characteristic group includes: The mechanism feature group, the working condition feature group, and the time series feature group are concatenated to construct a multidimensional input sequence; The multidimensional input sequence is transformed into a feature vector matrix through a linear projection layer and a position encoding module; The feature vector matrix is ​​input into a Transformer encoder containing multiple cascaded layers to map the feature vector matrix into a query matrix, a key matrix, and a value matrix; Based on the query matrix, key matrix, and value matrix, encoded data is obtained through a self-attention module for executing the self-attention mechanism with residual connections and a feedforward neural network with a linear layer of ReLU activation function; The aggregated feature representation of the top layer of the Transformer encoder is taken, and the total steam recovery of a single converter furnace is output through a linear regression mapping layer.

[0046] For example, such as Figure 4As shown, the self-attention mechanism in the Transformer model is used to capture global dependencies, establishing a nonlinear mapping from complex feedstock conditions to total steam recovery. The specific calculation process is as follows: Feature serialization and embedding representation: Mechanistic feature groups (C / Si content in molten iron, temperature, carbon-temperature balance factor), operating condition feature groups (oxygen blowing rate, scrap weight), and time-series feature groups are concatenated to construct a multi-dimensional input sequence. Through a linear projection layer and a position encoding module, the input sequence is transformed into a sequence with dimensions of... eigenvector matrix This preserves the physical attributes and relative position information of the input features.

[0047] Transformer encoder feature extraction: extracting the feature vector matrix The input consists of a Transformer encoder with multiple cascaded layers. The core of the encoder is a multi-head self-attention mechanism, which maps the input features to a query matrix Q, a key matrix K, and a value matrix V through different linear transformation matrices, and calculates the attention weights:

[0048] In the formula, denoted as the dimension of the bond vector. This formula can dynamically calculate the correlation weights within and between heterogeneous feature groups such as carbon-temperature balance factor and oxygen blowing rate. For example, when operating conditions fluctuate drastically, the model can adaptively improve the attention score of mechanistic features.

[0049] Feedforward computation and total prediction output: The output of the self-attention module, after residual connection and layer normalization processing, is input into the feedforward neural network. The feedforward neural network consists of two linear layers with ReLU activation functions to enhance the nonlinear expressive power of the model.

[0050] Finally, the aggregated feature representation of the top layer of the Transformer encoder is taken and, through a linear regression mapping layer, the predicted total steam recovery for a single converter furnace under the current manufacturing command is output.

[0051] In one possible implementation, the step of calculating the steam consumption of the refining process based on the rated steam consumption rate of the refining vacuum pump group according to the stage of the operating procedure and the refining steam consumption time mask based on the refining operation time window includes: The single refining operation is divided into standard process stages. Based on the equipment's factory parameters and preset operating procedures, the rated steam consumption rate of different stages is extracted, and the relative standard operating time intervals of different stages are set. Based on the rated steam consumption rate at different stages and the relative standard operating time interval at different stages, a piecewise function of steam consumption rate based on the relative operating time of refining is established. Based on the refining steam consumption time mask of the refining operation time window and the start time in the refining operation time window, a minute-level refining steam consumption curve is calculated and generated in the future scheduling cycle to determine the steam consumption of the refining process.

[0052] For example, the rated steam consumption rate for a stage is defined as follows: A single refining operation is divided into K standard process stages (such as the light vacuum pumping stage, the deep vacuum holding stage, and the vacuum breaking stage). Based on the equipment's factory parameters and on-site operating procedures, the rated steam consumption rate for the k-th stage is extracted. (Unit: t / h), and set the relative standard operating time interval for this stage as [ ), where the initial relative time .

[0053] Constructing a piecewise constant waveform function: Establishing a function based on refined relative operation time Piecewise function of steam consumption rate :

[0054] Global timeline mapping calculation: Extract the start time of the refining operation for the next manufacturing command. Relative working time Mapped to the global prediction time axis t, the conversion relationship is as follows: Combined with refining steam consumption time mask Calculate and generate minute-level refining steam consumption curves for future scheduling cycles. :

[0055] In the formula, the mask This ensures that the system's steam consumption is strictly forced to zero during non-refining operation periods, eliminating noise interference during non-operational cycles.

[0056] In one possible implementation, the step of mapping the total steam recovery to a converter blowing time window includes: Based on the principle of energy conservation, the total amount of steam recovered is used as the waveform integral area constraint. The benchmark steam production waveform function is obtained by statistical analysis of high-frequency performance data based on historical real operating conditions. The converter steam production time mask based on the converter blowing time window is used as a gating switch. The instantaneous steam production flow rate containing the fluctuation of real operating conditions is back-reassured and reconstructed through area normalization. The total amount of steam recovered is then mapped to the converter blowing time window to obtain the minute-level steam production flow rate curve.

[0057] For example, such as Figure 5 As shown, the converter steam production waveform reconstruction based on total quantity constraints: According to the principle of energy conservation, the total steam production of a single furnace is reconstructed. As a strict waveform integral area constraint, a benchmark steam production waveform function obtained from statistical analysis of high-frequency performance data under historical real-world operating conditions is introduced. Using converter steam generation time mask As a gated switch, the instantaneous steam production flow rate, which includes fluctuations in actual operating conditions, is derived and reconstructed through area normalization:

[0058] This ensures that the integral area of ​​the reconstructed curve is strictly equal to the total predicted amount, and that the timing of its occurrence is controlled by the rhythm prediction results. In the formula... Represents the predicted instantaneous steam flow rate at time t, in t / h; This represents the predicted total steam recovery per furnace, in tons (t). It is a predefined function, such as a trapezoidal function or a beta distribution function, whose amplitude range is usually normalized to [0,1].

[0059] In one possible implementation, the step of mapping the total steam recovery to a converter blowing time window and the steam consumption to a refining operation time window to generate a minute-level net steam flux prediction curve for future scheduling cycles includes: The steam consumption is mapped to the refining operation time window to obtain a minute-level steam flow rate curve; On a unified global time axis, the minute-level steam production flow curve and the minute-level steam consumption flow curve are algebraically superimposed and offset to determine the net flux demand, so as to obtain the minute-level net steam flux prediction curve for future scheduling cycles.

[0060] For example, the constructed refining consumption waveform By refining the steam consumption time mask Gating is implemented and mapped to a globally unified time axis. When time t falls within a non-refining operation period, the steam consumption at that station is forcibly zeroed to eliminate noise interference from non-operational cycles and generate a continuous minute-level refining steam consumption curve. .

[0061] Dynamic offsetting calculation of net steam production and consumption: On a unified global time axis t, the production flow and consumption flow at the same moment are algebraically superimposed and offset to accurately quantify the net flow demand of the converter steelmaking area on the entire plant's main steam pipeline network. = -

[0062] In one possible implementation, the method further includes: Based on the minute-level net steam flux prediction curve for the future scheduling cycle, a dynamic routing interaction strategy between the converter area and the main pipeline network is dynamically generated.

[0063] For example, the pipeline interaction strategy is dynamically generated. Based on the net flux calculation results, a dynamic routing interaction strategy between the converter area and the main pipeline network is generated: when When >0: A steam surplus is determined in the current area. The system generates an external transmission command to integrate the excess steam into the plant's high-pressure pipeline network or deliver it to the steam accumulator; when <0: A steam shortage is detected in the current area. The system generates a replenishment command, requiring the introduction of high-grade steam from the main pipeline network to meet the power requirements for refining vacuum pumping; when When =0: Determines the current state as a quiescent state during a production gap, and logically isolates the region from the entire plant's piping network. The final output includes a panoramic, multi-dimensional collaborative prediction curve encompassing output flow, consumption flow, and net flux. In one possible implementation, the minute-level net steam flux forecast curve and pipeline interaction strategy for the future scheduling cycle are transmitted in real time to the enterprise energy management center via a data interface. Based on the received feedforward forecast data, the EMS scheduling system performs dynamic balancing operations for the steelmaking-refining area: when a continuous surplus peak in the net steam flux of a section is predicted, the opening of the charging valve of the steam accumulator is adjusted in advance to store energy or the grid-connected power generation load of the low-pressure steam turbine generator unit is increased, thereby effectively reducing the steam venting rate; when a large load steam consumption gap is predicted in the subsequent refining process, the entire plant's high-pressure main pipeline network is scheduled in advance or the bypass desuperheating and pressure reducing station is opened to stabilize the pressure and supplement steam, strictly preventing a sudden drop in pipeline pressure and ensuring the safety of the power medium supply for the refining vacuum pump unit.

[0064] On the other hand, such as Figure 6 As shown, this application provides a rhythm-aware steam production and consumption prediction device for steelmaking-refining processes, comprising: The feature extraction module 201 is used to acquire historical production performance data of the process corresponding to the converter in the steelmaking-refining process, and to determine the mechanism feature group, the working condition feature group, and the time sequence feature group including the carbon temperature balance factor, based on the historical production performance data. The logistics queuing delay feature is the time difference between the blowing start time of any manufacturing order and the steel tapping end time of the previous manufacturing order. The dynamic graph construction module 202 is used to take the transportation equipment corresponding to the converter process in the steelmaking-refining process as nodes, obtain the operating status data of each node, construct a dynamic feature matrix based on the operating status data, and obtain the operating time data of each node, construct a dynamic adjacency matrix based on the operating time data and the logistics queuing delay characteristics. The time window prediction module 203 is used to input the dynamic feature matrix and the dynamic adjacency matrix into a preset prediction model to obtain the converter blowing time window and the refining operation time window; The recovery prediction module 204 is used to calculate the total steam recovery of a single furnace in the converter based on the mechanism characteristic group, the operating condition characteristic group and the time series characteristic group. The consumption prediction module 205 is used to calculate the steam consumption of the refining process based on the stage rated steam consumption rate of the refining vacuum pump group according to the operating procedure and the refining steam consumption time mask based on the refining operation time window. The mapping module 206 is used to map the total amount of steam recovery to the converter blowing time window to obtain a minute-level steam production flow curve, and to map the steam consumption to the refining operation time window to obtain a minute-level steam consumption flow curve, thereby generating a minute-level net steam flux prediction curve for the future scheduling cycle.

[0065] In one possible implementation, such as Figure 7 As shown, this application embodiment provides a terminal device 300, including: a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it realizes: acquiring historical production performance data of the process corresponding to the converter in the steelmaking-refining process, and determining a mechanism feature group including a carbon temperature balance factor, a working condition feature group, and a timing feature group including a logistics queuing delay feature based on the historical production performance data. The logistics queuing delay feature is the time difference between the blowing start time of any manufacturing order and the tapping end time of the previous manufacturing order. The transportation equipment corresponding to the converter in the steelmaking-refining process is taken as nodes. The operating status data of each node is obtained to construct a dynamic feature matrix based on the operating status data. The operating time data of each node is also obtained to construct a dynamic adjacency matrix based on the operating time data and the logistics queuing delay characteristics. The dynamic feature matrix and the dynamic adjacency matrix are input into a preset prediction model to obtain the converter blowing time window and the refining operation time window; Based on the mechanism characteristic group, operating condition characteristic group and time series characteristic group, calculate the total steam recovery of a single converter furnace. The steam consumption of the refining process is calculated based on the stage rated steam consumption rate of the refining vacuum pump unit according to the operating procedure and the refining steam consumption time mask based on the refining operation time window. The total amount of steam recovered is mapped to the converter blowing time window to obtain a minute-level steam production flow curve, and the steam consumption is mapped to the refining operation time window to obtain a minute-level steam consumption flow curve, thereby generating a minute-level net steam flux prediction curve for future scheduling cycles.

[0066] In one possible implementation, such as Figure 8As shown, this application embodiment provides a computer-readable storage medium 400, on which a computer program 411 is stored. When the computer program 411 is executed by a processor, it performs the following: acquiring historical production performance data of the process corresponding to the converter in the steelmaking-refining process, and determining a mechanistic feature group including a carbon-temperature balance factor, a working condition feature group, and a temporal feature group including a logistics queuing delay feature based on the historical production performance data. The logistics queuing delay feature is the time difference between the blowing start time of any manufacturing order and the tapping end time of the previous manufacturing order. The transportation equipment corresponding to the converter in the steelmaking-refining process is taken as nodes. The operating status data of each node is obtained to construct a dynamic feature matrix based on the operating status data. The operating time data of each node is also obtained to construct a dynamic adjacency matrix based on the operating time data and the logistics queuing delay characteristics. The dynamic feature matrix and the dynamic adjacency matrix are input into a preset prediction model to obtain the converter blowing time window and the refining operation time window; Based on the mechanism characteristic group, operating condition characteristic group and time series characteristic group, calculate the total steam recovery of a single converter furnace. The steam consumption of the refining process is calculated based on the stage rated steam consumption rate of the refining vacuum pump unit according to the operating procedure and the refining steam consumption time mask based on the refining operation time window. The total amount of steam recovered is mapped to the converter blowing time window to obtain a minute-level steam production flow curve, and the steam consumption is mapped to the refining operation time window to obtain a minute-level steam consumption flow curve, thereby generating a minute-level net steam flux prediction curve for future scheduling cycles.

[0067] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0068] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0069] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0070] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0071] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0072] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0073] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0074] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for predicting steam production and consumption in steelmaking-refining processes based on rhythm perception, characterized in that, include: Historical production performance data of the process corresponding to the converter in the steelmaking-refining process is obtained, and based on the historical production performance data, a mechanism characteristic group including the carbon temperature balance factor, an operating condition characteristic group, and a time sequence characteristic group including the logistics queuing delay characteristic are determined. The logistics queuing delay characteristic is the time difference between the blowing start time of any manufacturing order and the tapping end time of the previous manufacturing order. The transportation equipment corresponding to the converter in the steelmaking-refining process is taken as nodes. The operating status data of each node is obtained to construct a dynamic feature matrix based on the operating status data. The operating time data of each node is also obtained to construct a dynamic adjacency matrix based on the operating time data and the logistics queuing delay characteristics. The dynamic feature matrix and the dynamic adjacency matrix are input into a preset prediction model to obtain the converter blowing time window and the refining operation time window; Based on the mechanism characteristic group, operating condition characteristic group and time series characteristic group, calculate the total steam recovery of a single converter furnace. Based on the stage rated steam consumption rate in the refining vacuum pump unit operation procedure and the refining steam consumption time mask based on the refining operation time window, the steam consumption of the refining process is calculated. The total amount of steam recovered is mapped to the converter blowing time window to obtain a minute-level steam production flow curve, and the steam consumption is mapped to the refining operation time window to obtain a minute-level steam consumption flow curve, thereby generating a minute-level net steam flux prediction curve for future scheduling cycles.

2. The method for predicting steam production and consumption in steelmaking-refining processes based on rhythm perception as described in claim 1, characterized in that, The step of determining the mechanistic characteristic group including the carbon-temperature balance factor based on the historical production performance data includes: Based on the principle of heat balance in the converter molten pool, a carbon-temperature balance factor reflecting the steam generation potential is constructed using iron composition and temperature data. The mechanistic characteristic set of the carbon-temperature balance factor is obtained, and its mathematical characterization is as follows: In the formula, This indicates the amount of molten iron added, in kg; , These represent the carbon content and silicon content of molten iron, respectively. , These represent the exothermic enthalpy values ​​of the oxidation reactions of carbon and silicon, respectively, in kJ / kg; Specific heat capacity of molten iron, kJ / (kg·℃); , These represent the initial temperature of molten iron entering the furnace and the reference temperature for the temperature drop of molten iron specified in the process, respectively, in °C; This indicates the amount of scrap steel loaded, expressed in kg.

3. The method for predicting steam production and consumption in steelmaking-refining processes based on rhythm perception as described in claim 1, characterized in that, The step of calculating the total steam recovery of a single converter cycle based on the mechanism characteristic group, operating condition characteristic group, and time series characteristic group includes: The mechanism feature group, the working condition feature group, and the time series feature group are concatenated to construct a multidimensional input sequence; The multidimensional input sequence is transformed into a feature vector matrix through a linear projection layer and a position encoding module; The feature vector matrix is ​​input into a Transformer encoder containing multiple cascaded layers to map the feature vector matrix into a query matrix, a key matrix, and a value matrix; Based on the query matrix, key matrix, and value matrix, encoded data is obtained through a self-attention module for executing the self-attention mechanism with residual connections and a feedforward neural network with a linear layer of ReLU activation function; The aggregated feature representation of the top layer of the Transformer encoder is taken, and the total steam recovery of a single converter furnace is output through a linear regression mapping layer.

4. The method for coordinated prediction of steam production and consumption in steelmaking-refining processes based on rhythm perception as described in claim 1, characterized in that, The step of calculating the steam consumption of the refining process based on the rated steam consumption rate of the refining vacuum pump unit according to the stage of the operating procedure and the refining steam consumption time mask based on the refining operation time window includes: The single refining operation is divided into standard process stages. Based on the equipment's factory parameters and preset operating procedures, the rated steam consumption rate of different stages is extracted, and the relative standard operating time intervals of different stages are set. Based on the rated steam consumption rate at different stages and the relative standard operating time interval at different stages, a piecewise function of steam consumption rate based on the relative operating time of refining is established. Based on the refining steam consumption time mask of the refining operation time window and the start time in the refining operation time window, a minute-level refining steam consumption curve is calculated and generated in the future scheduling cycle to determine the steam consumption of the refining process.

5. The method for predicting steam production and consumption in steelmaking-refining processes based on rhythm perception as described in claim 1, characterized in that, The step of mapping the total steam recovery to the converter blowing time window includes: Based on the principle of energy conservation, the total amount of steam recovered is used as the waveform integral area constraint. The benchmark steam production waveform function is obtained by statistical analysis of high-frequency performance data based on historical real operating conditions. The converter steam production time mask based on the converter blowing time window is used as a gating switch. The instantaneous steam production flow rate containing the fluctuation of real operating conditions is back-reassured and reconstructed through area normalization. The total amount of steam recovered is then mapped to the converter blowing time window to obtain the minute-level steam production flow rate curve.

6. The method for predicting steam production and consumption in steelmaking-refining processes based on rhythm perception as described in claim 1, characterized in that, The step of mapping the total steam recovery to a converter blowing time window and the steam consumption to a refining operation time window to generate a minute-level net steam flux prediction curve for the future scheduling cycle includes: The steam consumption is mapped to the refining operation time window to obtain a minute-level steam flow rate curve; On a unified global time axis, the minute-level steam production flow curve and the minute-level steam consumption flow curve are algebraically superimposed and offset to determine the net flux demand, so as to obtain the minute-level net steam flux prediction curve for future scheduling cycles.

7. The method for predicting steam production and consumption in steelmaking-refining processes based on rhythm perception as described in claim 1, characterized in that, The method further includes: Based on the minute-level net steam flux prediction curve for the future scheduling cycle, a dynamic routing interaction strategy between the converter area and the main pipeline network is dynamically generated.

8. A rhythm-sensing-based steam production and consumption prediction device for steelmaking-refining processes, characterized in that, The rhythm-aware-based steam production and consumption prediction device for steelmaking-refining processes includes: The feature extraction module is used to acquire historical production performance data of the corresponding process of the converter in the steelmaking-refining process, and to determine the mechanism feature group, the working condition feature group, and the time sequence feature group including the carbon temperature balance factor, based on the historical production performance data. The logistics queuing delay feature is the time difference between the blowing start time of any manufacturing order and the tapping end time of the previous manufacturing order. The dynamic graph construction module is used to take the transportation equipment corresponding to the converter process in the steelmaking-refining process as nodes, obtain the operating status data of each node, construct a dynamic feature matrix based on the operating status data, and obtain the operating time data of each node, construct a dynamic adjacency matrix based on the operating time data and the logistics queuing delay characteristics. The time window prediction module is used to input the dynamic feature matrix and the dynamic adjacency matrix into a preset prediction model to obtain the converter blowing time window and the refining operation time window; The recovery prediction module is used to predict the total amount of steam recovered in a single furnace of the converter based on the mechanism characteristic group, the operating condition characteristic group, and the time series characteristic group. The consumption prediction module is used to calculate the steam consumption of the refining process based on the stage rated steam consumption rate in the refining vacuum pump group operation procedure and the refining steam consumption time mask based on the refining operation time window. The mapping module is used to map the total steam recovery to the converter blowing time window to obtain a minute-level steam production flow curve, and to map the steam consumption to the refining operation time window to obtain a minute-level steam consumption flow curve, thereby generating a minute-level net steam flux prediction curve for the future scheduling cycle.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the rhythm-aware steam production and consumption co-prediction method for steelmaking-refining processes as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the rhythm-aware steam production and consumption co-prediction method for steelmaking-refining processes as described in any one of claims 1 to 7.

Images