Control methods, computer storage media, and terminal equipment for converter blowing systems

Abstract

This invention discloses a control method, computer storage medium, and terminal equipment for a converter blowing system, relating to the field of blowing control technology. The method includes: coupling auxiliary lance measurement data with a decarburization rate model and a heating rate model in real time; generating a lance position correction factor using lance position changes and reaction rate changes; dynamically correcting the basic decarburization rate to obtain a more accurate target decarburization rate; and generating multi-dimensional control commands in conjunction with the heating rate. This method can significantly improve the accuracy of the endpoint temperature and carbon content prediction, avoiding supplementary blowing or subsequent blowing due to decarburization rate prediction errors. Simultaneously, by coordinating the adjustment of oxygen supply, coolant, and lance position, it ensures the stability and production efficiency of the blowing process.

Description

Technical Field

[0001] This invention relates to the field of blowing control technology, and in particular to a control method for a converter blowing system, as well as a computer storage medium and terminal equipment. Background Technology

[0002] Currently, during converter blowing, the data measured by the secondary lance is usually used for static analysis and simple threshold judgment after the blowing is completed. For example, after the secondary lance measurement, the difference between the current temperature and the target temperature is manually referenced to roughly estimate the subsequent oxygen supply or coolant addition. It cannot be adjusted in real time during the converter blowing process.

[0003] The above scheme cannot dynamically adjust the control parameters according to the actual state deviation during the blowing process, thus it cannot control the converter blowing process, resulting in low endpoint temperature and composition hit rate, and insufficient stability of the blowing process, which easily leads to problems such as splashing and re-drying, affecting production efficiency and product quality. This results in low stability and low production efficiency of converter blowing. Summary of the Invention

[0004] To solve the above-mentioned technical problems, the present invention provides a control method for a converter blowing system, comprising: Acquire the measurement parameters at the current moment, including the molten pool temperature and carbon content, as well as the corresponding status parameters, including the cumulative oxygen consumption, the cumulative amount of auxiliary materials added, the real-time gun position, and the oxygen supply flow rate; Generate a correlation dataset between the measured parameters and the state parameters, and input the correlation dataset into the decarbonization rate model and the heating rate model respectively; The gun position correction factor is calculated based on the gun position change and reaction rate change output by the decarbonization rate model, and the base decarbonization rate is corrected based on the gun position correction factor to determine the target decarbonization rate. Obtain the heating rate output by the heating rate model; Based on the target decarbonization rate and the heating rate, control commands are generated and executed, including oxygen supply adjustment commands, coolant adjustment commands, and gun position adjustment commands.

[0005] In one embodiment, the step of calculating a gun position correction factor based on the gun position change and reaction rate change output by the decarburization rate model, and correcting the base decarburization rate based on the gun position correction factor to determine the target decarburization rate includes: The associated dataset is input into the pre-trained decarburization rate model, and the decarburization rate model extracts the shape features of the gun position waveform and outputs the gun position change coefficient. The carbon diffusion driving force is determined, and the basic decarbonization rate is generated based on the carbon diffusion driving force, reaction rate constant, and oxygen supply, and corrected by a penalty factor, wherein the penalty factor is calculated based on current measurement data. Based on the change in reaction rate and the change coefficient of gun position, the gun position correction factor is calculated according to a preset formula. The base decarburization rate is adjusted according to the gun position correction factor to generate the target decarburization rate.

[0006] In one embodiment, the step of inputting the associated dataset into the pre-trained decarburization rate model, extracting the shape features of the gun position waveform from the decarburization rate model, and outputting the gun position change coefficient includes: Input the associated dataset into the decarbonization rate model; The decarbonization rate model extracts local waveform features through at least one convolutional kernel and generates correction coefficients through dimensionality reduction using a pooling layer. The correction coefficients are fused with the local waveform features extracted by each of the convolution kernels to obtain fused features; The target waveform features are extracted from the fused features through global max pooling, and the extracted target waveform features are mapped to generate the gun position change coefficient based on the fully connected layer.

[0007] In one embodiment, the step of determining the carbon diffusion driving force, calculating based on the carbon diffusion driving force, the reaction rate constant, and the oxygen supply, and generating the basic decarbonization rate based on a penalty factor, wherein the penalty factor is calculated based on the current sputtering pulse data, includes: The reaction rate constant is determined based on the molten pool temperature measured by the current sub-lance, and the carbon diffusion driving force is determined based on the current carbon content; Obtain the ratio of the current oxygen supply flow rate to the weight of molten steel, and determine the penalty factor calculated based on the current measurement data; The basic decarburization rate is obtained by multiplying the reaction rate constant, the carbon diffusion driving force, the penalty factor, and the ratio of oxygen supply flow rate to molten steel weight.

[0008] In one embodiment, the penalty factor includes a molten pool penalty factor and a splash penalty factor, and the step of determining the penalty factor calculated based on the current measurement data further includes: Obtain the current gun position measurement value from the current measurement data, and calculate the degree of gun position fluctuation; Based on the preset mapping relationship between gun position fluctuation and molten pool penalty factor, the molten pool penalty factor corresponding to the degree of gun position fluctuation is determined; Obtain the current number of splash pulses from the current measurement data, and determine the splash penalty factor according to the preset mapping relationship between the number of splash pulses and the splash penalty factor.

[0009] In one embodiment, the step of generating and executing control commands based on the target decarbonization rate and the heating rate, wherein the control commands include oxygen supply adjustment commands, coolant adjustment commands, and gun position adjustment commands, includes: If the target decarbonization rate is higher than the preset target range, a first oxygen supply adjustment command and a first gun position adjustment command are generated. Execute the first oxygen supply adjustment command and the first gun position adjustment command to reduce the oxygen supply flow rate of the first preset value through the first oxygen supply adjustment command, and to increase the gun position of the second preset value through the first gun position adjustment command. If the target decarbonization rate is lower than the preset target range, a second oxygen supply adjustment command and a second gun position adjustment command are generated. The second oxygen supply adjustment command and the second gun position adjustment command are executed to increase the oxygen supply flow rate of the first preset value through the second oxygen supply adjustment command and to decrease the gun position of the second preset value through the second gun position adjustment command.

[0010] In one embodiment, the step of obtaining the heating rate output by the heating rate model includes: Determine the molten pool temperature sequence in the associated dataset and input the molten pool temperature sequence into the heating rate model, wherein the heating rate model includes a first branch and a second branch; The first branch is used to extract feature maps from the molten pool temperature sequence and generate heat loss correction coefficients. The molten pool temperature sequence is processed by the second branch to generate a slag layer penalty factor; The heating rate correction coefficient is obtained by adding the heat loss correction coefficient to the slag layer penalty factor and then adding a unit value. The heating rate is determined based on the heating rate correction factor.

[0011] In one embodiment, after the step of determining the heating rate based on the heating rate correction coefficient, the method further includes: Determine the target rate, and determine the corresponding difference between the heating rate and the target rate; The parameter adjustment amount is determined based on the difference and the preset coefficient, and the target instruction is generated based on the parameter adjustment amount.

[0012] In one embodiment, the step of generating and executing control commands based on the target decarbonization rate and the heating rate, wherein the control commands include oxygen supply adjustment commands, coolant adjustment commands, and gun position adjustment commands, further includes: The predicted temperature corresponding to the target time point is determined based on the heating rate. If the predicted temperature is greater than the preset temperature at the target time point, a first coolant adjustment command is generated. Execute the first coolant adjustment command to add the corresponding coolant via the first coolant adjustment command; If the predicted temperature is less than or equal to the preset temperature, a target control command is generated and executed to pause coolant delivery and increase oxygen supply or lower the gun position.

[0013] In addition, to solve the above-mentioned technical problems, the present invention also provides a computer storage medium storing executable program code; the executable program code is used to execute the control method of the converter blowing system as described above.

[0014] In addition, to solve the above-mentioned technical problems, the present invention also provides a terminal device, including a memory and a processor; the memory stores program code that can be executed by the processor; the program code is used to execute the control method of the converter blowing system as described above. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the first embodiment of the control method for the converter blowing system of the present invention; Figure 2 This is a schematic diagram of the decarbonization rate calculation process involved in the embodiments of the present invention; Figure 3 This is a schematic diagram of the heating rate model involved in the embodiments of the present invention; Figure 4 This is a schematic diagram of the hardware operating environment of the terminal device involved in the embodiments of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0017] It should be noted that if the embodiments of the present invention involve directional indications, such as up, down, left, right, front, back, etc., these directional indications are only used to explain the relative positional relationships and movement of the components in a specific posture. If the specific posture changes, the directional indications will also change accordingly. Furthermore, if the embodiments of the present invention involve descriptions such as "first," "second," "S1," "S2," "step one," "step two," etc., these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance, or implicitly indicating the number of technical features indicated or the execution order of the method. Those skilled in the art will understand that anything that does not violate the inventive concept should be included within the scope of protection of the present invention.

[0018] Example Please refer to Figure 1 , Figure 1 This is a schematic flowchart of the first embodiment of the control method for the converter blowing system of the present invention, as shown below. Figure 1 As shown, the present invention provides a control method for a converter blowing system, comprising the following steps: Step S10: Obtain the measurement parameters at the current moment, including the molten pool temperature and carbon content, as well as the corresponding status parameters, including the cumulative oxygen consumption, the cumulative amount of auxiliary materials added, the real-time gun position, and the oxygen supply flow rate. In this embodiment, the processing actions can be performed by the converter blowing system, with the auxiliary lance measuring TSC (Temperature, Sampling, Carbon) to obtain relevant parameters. The auxiliary lance TSC measurement is a core automated detection technology that involves inserting a measuring probe into the molten pool during the converter blowing process for rapid online temperature measurement, carbon determination, and sampling. The auxiliary lance probe contains a thermocouple and a sampler, allowing for rapid insertion of the lance at a preset time point during blowing, automatic lifting after measurement, and delivery of the measurement results. The molten pool temperature is the current actual temperature of the molten steel, expressed in degrees Celsius. Carbon content refers to the percentage by mass of dissolved carbon in the molten steel. Cumulative oxygen consumption refers to the total volume of oxygen supplied to the molten pool through the oxygen lance from the start of the current blowing process to the current moment. Cumulative auxiliary material addition refers to the total mass of auxiliary materials such as lime, dolomite, and coolant added from the start of the blowing process to the current moment. Real-time lance position refers to the vertical distance between the oxygen lance nozzle end face and the static surface of the molten pool. Oxygen supply flow rate refers to the volume of oxygen supplied to the molten pool through the oxygen lance per unit time.

[0019] Specifically, the converter blowing control system first sends a measurement command to the auxiliary lance measurement system, which automatically completes the actions of lowering the lance, measuring, and raising it. The thermocouple in the auxiliary lance probe continuously collects temperature signals after being inserted into the molten pool, and generates a molten pool temperature value after analog-to-digital conversion. Simultaneously, a sampler collects a sample of molten steel, which is then analyzed by a spectrometer to obtain the carbon content value. These two measurement parameters are read from the interface of the auxiliary lance measurement system. Next, the converter blowing control system reads the cumulative oxygen consumption from the programmable logic controller (PLC), a value obtained by integrating the oxygen supply flow rate over time from the oxygen flow meter; it reads the cumulative auxiliary material addition from the hopper weighing sensor; it reads the real-time lance position from the encoder feedback of the oxygen lance servo driver; and it reads the current oxygen supply flow rate from the flow transmitter of the oxygen regulating valve. After all data acquisition is completed, the control system temporarily stores these parameters in memory according to a unified sampling timestamp.

[0020] Step S20: Generate a correlation dataset between the measured parameters and the state parameters, and input the correlation dataset into the decarbonization rate model and the heating rate model respectively; In this embodiment, the decarburization rate model is a neural network model used to calculate the rate at which the carbon content of molten steel decreases per unit time. This model comprehensively considers the effects of multiple factors such as temperature, oxygen supply intensity, lance position fluctuation, and splashing degree. The heating rate model is a neural network model used to calculate the rate at which the temperature of the molten pool rises per unit time. Both models run on the model server of the converter blowing system.

[0021] Specifically, the six parameters obtained—molten pool temperature, carbon content, cumulative oxygen consumption, cumulative auxiliary material addition, real-time nozzle position, and oxygen supply flow rate—are first packaged into a correlated dataset according to their timestamps. This dataset is stored in memory as structured variables, with each parameter accompanied by a unit identifier and a validity flag. Then, the application programming interface (API) of the decarburization rate model is called, passing the molten pool temperature, carbon content, oxygen supply flow rate, cumulative oxygen consumption, and real-time nozzle position from the correlated dataset as input parameters. Simultaneously, the control system calls the interface of the heating rate model, passing the molten pool temperature, oxygen supply flow rate, and cumulative oxygen consumption from the correlated dataset as input parameters. The two models perform parallel calculations, outputting intermediate results for subsequent steps.

[0022] As an optional implementation, after generating the associated dataset, the current converter status can be quantitatively assessed based on this dataset, and core deviation indicators can be calculated. These core deviation indicators include temperature deviation, carbon content deviation, and blowing progress deviation. Specifically, temperature deviation ΔT = current TSC temperature - target endpoint temperature; composition deviation ΔC = current carbon content - target endpoint carbon content; ΔP = current phosphorus content - target endpoint phosphorus content; and progress deviation is based on the ratio of cumulative oxygen consumption to theoretical oxygen demand, assessing the blowing progress. For example, if the theoretical oxygen demand is Qtheoretical and the cumulative oxygen consumption is Qcumulative, then the progress deviation ΔPprogress = Qcumulative / Qtheoretical × 100%. After this, the current converter status can be assessed using the core deviation indicators, allowing maintenance personnel to adjust equipment parameters based on the assessment report.

[0023] Step S30: Calculate the gun position correction factor using the gun position change coefficient and reaction rate change output by the decarburization rate model, and correct the basic decarburization rate based on the gun position correction factor to determine the target decarburization rate. In this embodiment, the lance position variation coefficient is a scalar value output by the decarburization rate model after analyzing the real-time lance position sequence through a one-dimensional convolutional neural network. It characterizes the degree of influence of the current lance position fluctuation mode on decarburization efficiency. The lance position correction factor is a multiplicative coefficient, obtained by adding the lance position variation coefficient and the reaction rate change to the unit value, and is used to comprehensively correct the basic decarburization rate. The basic decarburization rate is the theoretical decarburization rate calculated considering only the four core physical quantities: temperature, carbon content, oxygen supply flow rate, and molten steel weight, expressed as mass percentage per minute or kilogram per minute. The target decarburization rate is the actual effective decarburization rate after correction by the lance position correction factor.

[0024] The change in reaction rate refers to the change in the reaction rate constant caused by temperature variations. Specifically, it represents the increment or adjustment of the reaction rate constant when the current temperature T changes relative to a reference temperature (such as the previous temperature T0, or the laboratory calibration temperature). In practical control systems, this quantity is often used to dynamically correct predictions of the decarburization rate. When fluctuations are detected in the converter blowing system, the change in reaction rate is used for rapid compensation calculations without needing to recalculate the complete exponential term. Essentially, it reflects the marginal impact of temperature disturbances on reaction capacity and is a practical simplification in engineering.

[0025] Specifically, firstly, the lance position variation coefficient output by the decarburization rate model is obtained. This coefficient is derived from the shape characteristics of the lance position waveform extracted by the decarburization rate model. Then, based on the pre-generated carbon diffusion driving force, the basic decarburization rate is calculated by combining the reaction rate constant, penalty factor, and oxygen supply flow rate. Next, the difference between the current molten pool temperature and the preset reference temperature is calculated, and multiplied by an empirical coefficient to obtain the reaction rate variation. Finally, the lance position variation and the reaction rate variation are calculated to obtain the lance position correction factor. Finally, the basic decarburization rate is multiplied by the lance position correction factor to obtain the target decarburization rate. The converter blowing control system then reads this target decarburization rate from the output interface of the decarburization rate model.

[0026] Optionally, the decarbonization rate model can simultaneously extract the gun position measurement values ​​from the most recent few seconds to form a time series, and input this series into a pre-trained one-dimensional convolutional neural network. This network extracts the shape features of the gun position waveform through multiple convolutional kernels, including jitter frequency and glide slope, processes them, and outputs the gun position change coefficient. Furthermore, the decarbonization rate model can also calculate the amount of decarbonization per unit time in the subsequent period, and combine this with ΔC to calculate the required "remaining decarbonization time". For example, if the amount of decarbonization per unit time in the subsequent period is v, then the remaining decarbonization time t_remaining = |ΔC| / v_decarbonization. A preliminary derivation of the theoretical remaining oxygen supply is given: oxygen supply = decarbonization amount × oxygen-carbon reaction coefficient. If the oxygen-carbon reaction coefficient is k, then the theoretical remaining oxygen supply Q_remaining = |ΔC| × k.

[0027] Optionally, in this embodiment, step S30 includes: Step S31: Input the associated dataset into the pre-trained decarburization rate model, extract the shape features of the gun position waveform from the decarburization rate model, and output the gun position change coefficient; Step S32: Determine the carbon diffusion driving force, calculate the basic decarbonization rate based on the carbon diffusion driving force, reaction rate constant and oxygen supply, and correct it based on the penalty factor, wherein the penalty factor is calculated based on the current measurement data; Step S33: Calculate the gun position correction factor according to the preset formula based on the change in reaction rate and the gun position change coefficient. Step S34: Correct the basic decarburization rate according to the gun position correction factor to generate the target decarburization rate.

[0028] Specifically, refer to Figure 2 , Figure 2 This is a schematic diagram of the decarbonization rate calculation process involved in an embodiment of the present invention. The gun position correction factor G = 1 + Δk_T + Δk_L, where Δk_T is the change in reaction rate, Δk_L is the gun position change coefficient, and Δk_T = δ·(T - T_target). For example, δ = 0.01 ℃. -1T_target = 1600 ℃. If the secondary lance measures 1630 ℃, then Δk_T = +0.3, the rate increases by 30%, meaning the actual temperature inside the furnace is higher than the model assumption. Next, the target decarburization rate dC / dt = dC / dt_mech · G is calculated.

[0029] Furthermore, step S32 also includes: The reaction rate constant is determined based on the molten pool temperature measured by the current sublance, and the carbon diffusion driving force is determined based on the current carbon content; the ratio of the current oxygen supply flow rate to the weight of molten steel is obtained, and the penalty factor calculated based on the current measurement data is determined; the reaction rate constant, the carbon diffusion driving force, the penalty factor, and the ratio of the oxygen supply flow rate to the weight of molten steel are multiplied together to obtain the basic decarburization rate.

[0030] Specifically, the basic decarbonization rate is expressed as: in, is the reaction rate constant, which is temperature-dependent. It is a parameter characterizing the inherent speed of the reaction. It is independent of concentration and depends only on temperature and the nature of the reaction. When the temperature increases, k will increase according to the Arrhenius exponential law, thereby driving up the overall reaction rate. However, k itself is not the final reaction rate value. A is the carbon diffusion driving force, which is related to the carbon content. This refers to the oxygen supply flow rate; This refers to the weight of the molten steel. as well as As a penalty factor, where As the molten pool penalty factor, This is a splash penalty factor.

[0031] More specifically, Reaction rate constant, optional ;in, It is a frequency factor, a constant determined by the properties of the reaction / diffusion itself; The activation energy is the minimum energy required for carbon diffusion / decarbonization reactions. It is the gas constant; The TSC temperature is given by the thermocouple in the secondary gun. The higher the temperature, the faster the CO bubbles decarbonize, so an exponential term needs to be included later. For example, the reaction rate constant increases by approximately 8% for every 10°C increase in temperature. The laboratory performs calibration once.

[0032] A represents the driving force for carbon diffusion, characterizing the magnitude of the tendency for carbon to diffuse from the interior of molten steel to the interface. ;in, The initial carbon content can be selected as the instantaneous carbon content of the molten pool when the secondary lance is just pulled up and the sparks have not yet extinguished. This initial carbon content is also the starting point for the model to deduct from the initial carbon content. An example is: 0.27%; Critical carbon content (Csat) is the saturated solubility of carbon in molten iron under specific conditions. In other words, it represents the theoretically lowest possible carbon content in the molten pool under current temperature, oxygen potential, and furnace atmosphere conditions. For example, at 1600 °C, the critical carbon content in molten iron is approximately Csat ≈ 0.04%. The correction index is an empirical coefficient that fits the actual diffusion pattern; for example, This will make the slope of the high-carbon zone steeper and the low-carbon zone flatter.

[0033] More specifically, The oxygen supply flow rate is the sole energy input for decarbonization. The larger the volume, the faster the decarburization; the example shows 480 Nm based on feedback from the oxygen lance flow valve. 3 / min; This refers to the weight of the molten steel. How much oxygen is supplied per kilogram of molten steel per minute?

[0034] In a preferred embodiment, the penalty factor includes a molten pool penalty factor and a splash penalty factor. The step of determining the penalty factor calculated based on the current measurement data further includes: Obtain the current gun position measurement value from the current measurement data and calculate the gun position fluctuation degree; determine the molten pool penalty factor corresponding to the gun position fluctuation degree according to the preset mapping relationship between gun position fluctuation and molten pool penalty factor; obtain the current splash pulse count from the current measurement data and determine the splash penalty factor according to the preset mapping relationship between splash count and splash penalty factor.

[0035] Specifically, the current gun position measurement value is obtained from the current measurement data, and the degree of gun position fluctuation is calculated using the formula. Determine the molten pool penalty factor; among which, This is the molten pool penalty factor; Adjust the weights for the molten pool; This represents the average number of gun positions. Additionally... Characterizes the degree to which molten pool fluctuations suppress the decarburization rate; optionally, =3 m -1 ; That is, when there is no gun position shake. ; That is, the gun position shakes violently. hour, When it drops to 0.5, the decarbonization rate is halved, which is consistent with the visual observation that "the area of ​​the impact zone is shaken apart".

[0036] The current number of splash pulses in the current measurement data is obtained using the formula. Determine the splash penalty factor; among which, As a splash penalty factor; Adjust the weights for splashing; Number of splashes; More specifically, Number of splashes It can be obtained through a high-speed camera at the furnace opening, sound energy, etc. For example, if four splash pulses were detected within the past 30 seconds, then... Because the splashing carries away CO bubbles prematurely, the effective decarbonization rate is reduced. Example: This indicates no splashing; Second-rate, This indicates that there was too much splashing in that minute, and only 80% of the decarburization was effective.

[0037] Next, according to the formula Multiplying by the molten pool penalty factor and the splashing penalty factor (two penalty factors), the decarburization rate after decay is obtained as the final basic decarburization rate.

[0038] Further, in this embodiment, step S31 includes: The associated dataset is input into the decarbonization rate model, which extracts local waveform features through at least one convolutional kernel and reduces the dimensionality through a pooling layer to generate correction coefficients. The correction coefficients are then fused with the local waveform features extracted by each convolutional kernel to obtain fused features. Target waveform features are extracted from the fused features through global max pooling, and the extracted target waveform features are mapped to generate the gun position change coefficients based on a fully connected layer.

[0039] Specifically, such as Figure 2 The feature extraction layer includes at least one convolutional layer (Conv8→Pool→Conv4→GlobalMax). Its core function is to use eight convolutional kernels to first extract local waveform features (such as jitter frequency and glide slope), then use pooling layers to reduce dimensionality and remove redundancy, and finally use Conv4 to further fuse the features, thereby achieving feature differentiation of different gun position change patterns. In other words, the CNN uses eight convolutional kernels to quantize the "waveform shape" into a correction coefficient, which is the local waveform feature. Then, dimensionality reduction and fusion are performed to obtain the fused features. GlobalMax (global max pooling) extracts the core information of key features based on the fused features, avoiding local noise interference, and obtains the final waveform features. Then, the output layer (Dense8→Dense1) maps the final waveform features extracted by the convolution to the final scalar Δk_L (gun position change) through a fully connected layer (Dense), completing the "feature → quantized value" conversion.

[0040] Step S40: Obtain the heating rate output by the heating rate model; Optionally, in this embodiment, step S40 includes: The molten pool temperature sequence in the associated dataset is determined and input into the heating rate model, which includes a first branch and a second branch. The first branch is used to extract feature maps from the molten pool temperature sequence and generate a heat loss correction coefficient. The second branch is used to process the molten pool temperature sequence and generate a slag layer penalty factor. The heat loss correction coefficient and the slag layer penalty factor are added together and a unit value is added to obtain the heating rate correction coefficient. The heating rate is determined based on the heating rate correction coefficient.

[0041] Specifically, the heating rate model internally comprises two parallel and functionally distinct computational branches. The first branch, the main branch, consists of multiple convolutional layers, pooling layers, a global max-pooling layer, and a fully connected layer. It is specifically designed to extract local dynamic features from the molten pool temperature sequence, such as the temperature drop slope and fluctuation frequency, which are directly related to furnace lining heat loss. The second branch, the auxiliary branch, consists of a 1-to-1 convolutional layer, residual connection paths, a global average pooling layer, and a fully connected layer. It is specifically designed to extract long-range smoothing trend features from the temperature sequence, which are directly related to the impact of slag thickness on heat dissipation. After independent computation, the outputs of the two branches are fused at the end of the model. The feature map is constructed by sliding convolutional kernels across the sequence, calculating the dot product of the kernel weights and the corresponding sequence values ​​at each position, and outputting a new numerical value; the outputs at all positions constitute the feature map.

[0042] The heat loss correction factor is a scalar output from the first branch after global max pooling and fully connected layers. It quantifies the degree to which furnace lining heat loss inhibits the heating rate. Its physical meaning is: under the same unit weight oxygen supply input, the discount on the heating rate corresponding to the additional heat loss caused by furnace lining aging, erosion, or nodulation; it represents the reduction in the theoretical heating rate. The slag layer penalty factor is a scalar output from the second branch, used to quantify the negative impact of the current slag layer thickness and its degree of foaming on the heating rate.

[0043] When the slag layer is thick or severely foamed, the slag layer covering the molten pool surface enhances heat dissipation, reducing the heating rate. In this case, the slag layer penalty factor is negative, and a larger absolute value indicates stronger heat dissipation. When the slag layer is thin, heat dissipation is weaker, and the slag layer penalty factor is close to zero or positive. The generation of this factor depends on the combination of residual connections and one-to-one convolution. The residual connections bypass the convolution operation and directly pass the original temperature sequence to subsequent layers to preserve the long-range average temperature information, while the one-to-one convolution is equivalent to performing a weighted average of the sequence to extract the overall temperature level.

[0044] Specifically, after acquiring and preprocessing the molten pool temperature sequence, the interface of the heating rate model is invoked. The normalized molten pool temperature sequence is passed as an input tensor to the model. The heating rate model first performs a dimensionality transformation on the input tensor, adjusting it to the data format required by a one-dimensional convolutional neural network, namely, batch size, number of channels, and sequence length. Subsequently, the model simultaneously copies the input tensor twice, sending them to the computation graphs of the first and second branches respectively. The two branches execute in parallel without interference. The control system waits for the computation of both branches to complete and then obtains the output results of each branch.

[0045] For example, such as Figure 3 As shown, Figure 3 This is a schematic diagram of the heating rate model architecture involved in the embodiment of the present invention. The main tower CNN (C1→P1→C2→C3→G1→D1→D2) – the first branch, which is the furnace lining heat loss branch. Input: Temperature sequence of the past 64 points (downsampled at 0.64 s 200 Hz). Feature extraction: Three convolutional layers C1→P1→C2→C3 extract the "temperature decrease slope" and "fluctuation frequency" into 8-dimensional features; Output: Δk_loss heat loss correction coefficient is output through GlobalMax→Dense16→Dense1.

[0046] The second branch is (R1→G2→D3→D4)Conv1-8ch ReLU→GlobalAvgPool→Dense4 ReLU→Dense1, where 1×1 Conv is equivalent to "weighted averaging of the temperature sequence," jumping to global average pooling to avoid the main convolution washing away long-range smoothing information. The output Δk_slag (slag layer penalty factor): negative values ​​represent "thick slag layer / severe foaming," enhancing heat dissipation; positive values ​​represent "thin slag layer," weakening heat dissipation. The value range is ±0.1.

[0047] Furthermore, in this embodiment, after the step of determining the heating rate based on the heating rate correction coefficient, the method further includes: Determine the target rate, and then determine the corresponding difference between the heating rate and the target rate. The parameter adjustment amount is determined based on the difference and preset coefficients, and the target instruction is generated based on the parameter adjustment amount.

[0048] Specifically, the converter blowing control system reads the target rate corresponding to the current blowing stage from the process parameter table. The target rate changes with the blowing process. Then, it determines the current stage based on the current cumulative oxygen consumption or blowing time and obtains the corresponding target value. Subsequently, it reads the heating rate from the output variables, subtracts the target rate from the heating rate, and obtains the difference. Then, it determines whether the difference is positive or negative. If the difference is positive, it means that the actual heating rate is higher than the target heating rate, and cooling measures need to be taken. This can be done by increasing the amount of coolant added, reducing the oxygen supply intensity, or a combination of both. By reading the preset cooling coefficient, which contains two sub-parameters: the oxygen supply cooling coefficient and the coolant cooling coefficient. The oxygen supply cooling coefficient represents the reduction in the heating rate that can be achieved by reducing the oxygen supply intensity by one unit (e.g., per standard cubic meter per minute per ton). The coolant cooling coefficient represents the reduction in the heating rate that can be achieved by increasing the amount of coolant by one ton (or the number of kilograms of coolant added per ton of molten steel).

[0049] Divide the difference by the oxygen supply cooling coefficient to obtain the amount of oxygen supply intensity adjustment to be reduced; or divide the difference by the coolant cooling coefficient to obtain the amount of coolant addition adjustment to be increased. If the difference is negative, it indicates that the actual heating rate is lower than the target heating rate, and heating measures are required. Read the preset heating coefficient, which represents the heating rate increase that can be achieved by increasing the oxygen supply intensity by one unit. Divide the absolute value of the difference by the heating coefficient to obtain the amount of oxygen supply intensity adjustment to be increased. Alternatively, the amount of coolant addition can be reduced. After determining the adjustment amount, the control system adds the adjustment amount to the current oxygen supply intensity setpoint or coolant addition setpoint to generate the corresponding control command. The oxygen supply adjustment command is sent to the positioner of the oxygen regulating valve in the form of current or digital signal via fieldbus, and the valve opening changes accordingly. The coolant adjustment command triggers the frequency converter of the vibrating feeder through the programmable logic controller to control the feeder's running time or speed, thereby achieving quantitative feeding. At the same time, the adjustment amount and adjustment time are recorded for subsequent closed-loop feedback correction.

[0050] Step S50: Based on the target decarburization rate and heating rate, generate and execute control commands, including oxygen supply adjustment commands, coolant adjustment commands, and gun position adjustment commands; In this embodiment, the heating rate model is a neural network model used to predict the rate of temperature rise in the molten pool. This model first calculates the theoretical heating rate based on the heat released from the reaction of oxygen and carbon and the specific heat capacity of the molten steel. Then, it uses a neural network to analyze historical temperature sequences to estimate the heat loss of the furnace lining and the slag layer, finally outputting a corrected actual heating rate. The heating rate refers to the numerical increase in the molten pool temperature per unit time. Control commands are operation commands sent by the control system to the executing equipment. The oxygen supply adjustment command is used to change the opening of the oxygen regulating valve, thereby changing the oxygen supply flow rate. The coolant adjustment command is used to control the start and stop of the vibrating feeder, thereby adding a predetermined amount of ore or scrap steel to the molten pool to absorb heat. The lance position adjustment command is used to control the raising and lowering of the oxygen lance servo motor, thereby changing the distance between the oxygen lance nozzle and the molten pool surface. These commands are sent to the corresponding actuators in the form of electrical signals via a fieldbus or programmable logic controller.

[0051] Optionally, in this embodiment, step S50 includes: If the target decarbonization rate is higher than the preset target range, a first oxygen supply adjustment command and a first gun position adjustment command are generated. Execute the first oxygen supply adjustment command and the first gun position adjustment command to reduce the oxygen supply flow rate of the first preset value through the first oxygen supply adjustment command, and to raise the gun position of the second preset value through the first gun position adjustment command; If the target decarbonization rate is lower than the preset target range, a second oxygen supply adjustment command and a second gun position adjustment command are generated. Execute the second oxygen supply adjustment command and the second gun position adjustment command to increase the oxygen supply flow rate of the first preset value through the second oxygen supply adjustment command, and decrease the gun position of the second preset value through the second gun position adjustment command.

[0052] Specifically, the converter blowing system compares the obtained target decarburization rate with a preset target range. If the target decarburization rate is higher than the upper limit of the range, a first oxygen supply adjustment command is generated, requiring the oxygen supply flow rate to decrease by a first preset value. Simultaneously, a first lance position adjustment command is generated, requiring the lance position to be raised to a second preset value to reduce the oxygen utilization rate in the impact zone, thereby slowing down decarburization. If the target decarburization rate is lower than the lower limit of the range, commands to increase the oxygen supply flow rate and decrease the lance position are generated. For example, if the target decarburization rate dC / dt is higher than the target range, the oxygen supply flow rate is immediately reduced by 3-5% via PLC, and the lance position is raised stepwise by 0.1-0.2 m to reduce the oxygen utilization rate in the impact zone and slow down decarburization; if dC / dt is lower than the range, the oxygen supply is increased and the lance position is lowered to enhance stirring.

[0053] Optionally, in this embodiment, step S50 further includes: Determine the predicted temperature corresponding to the target time point based on the heating rate; If the predicted temperature is higher than the preset temperature at the target time, generate the first coolant adjustment command; Execute the first coolant adjustment command to add the corresponding coolant. If the predicted temperature is less than or equal to the preset temperature, a target control command is generated and executed to pause coolant delivery and increase oxygen supply or lower the gun position.

[0054] Specifically, the predicted temperature corresponding to the target time point is first calculated based on the heating rate, and then compared with the preset temperature at the target time point. If the predicted temperature will exceed the target endpoint temperature during subsequent blowing processes, the control system generates a coolant adjustment command, requiring the vibrating feeder to add a preset amount of coolant, such as lightly calcined dolomite or iron ore, and to complete the addition within a specified time. If the heating rate is insufficient, the coolant addition is paused or a command to increase the oxygen supply intensity is generated. All commands are output through the programmable logic controller to the oxygen regulating valve, the oxygen lance servo driver, and the motor controller of the vibrating feeder, and the actuators complete the corresponding actions according to the command values.

[0055] For example, if the predicted temperature is >10 °C higher than the target endpoint, the vibrating feeder is immediately started to add coolant such as lightly calcined dolomite or iron ore at a rate of 0.8-1.2 kg per ton of molten steel, and this is completed within 30 seconds; if dT / dt is insufficient, i.e. the predicted temperature is less than or equal to the preset temperature, the coolant is suspended, the oxygen supply intensity is appropriately increased, or the gun position is lowered to increase the stirring heat.

[0056] Furthermore, in this embodiment, after the step of generating a first oxygen supply adjustment command and a first gun position adjustment command if the target decarbonization rate is higher than a preset target range, the method further includes: When the target decarbonization rate is detected to have entered the target range and the heating rate is close to the preset rate, a preset coolant dosing command is generated and executed to dosing the corresponding target coolant.

[0057] Specifically, regarding excipients, the dC / dt ratio is 0.10-0.08%·min - 1 window and dT / dt approaching 4 °C·min - At 1 hour, aluminum shot or ferrosilicon is added 10-15 seconds in advance via a rotary chute for pre-deoxidation to reduce post-blowing. Furthermore, the entire converter blowing system is adjusted in a closed-loop cycle of 1 second: model output → L2 system → synchronized operation of oxygen valve, feeder, and servo cylinder, with exhaust gas analysis and secondary lance measurements written back for correction, ensuring that the final carbon and temperature are simultaneously reached, achieving dynamic and precise control.

[0058] In the technical solution provided in this embodiment, the secondary lance measurement data is coupled in real time with the decarburization rate model and the heating rate model. A lance position correction factor is generated using the lance position change and reaction rate change to dynamically correct the basic decarburization rate, thereby obtaining a more accurate target decarburization rate. This is combined with the heating rate to generate multi-dimensional control commands. The above method can significantly improve the accuracy of the endpoint temperature and carbon content prediction, avoiding supplementary blowing or subsequent blowing caused by decarburization rate prediction deviations. Simultaneously, by coordinating the adjustment of oxygen supply, coolant, and lance position, the stability and production efficiency of the blowing process are ensured.

[0059] On the other hand, the present invention also provides a computer storage medium storing executable program code; the executable program code is used to execute the control method of any of the above-mentioned converter blowing systems.

[0060] On the other hand, the present invention also provides a terminal device, including a memory and a processor; the memory stores program code that can be executed by the processor; the program code is used to execute the control method of any of the above-mentioned converter blowing systems.

[0061] For example, the program code can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the program code in the terminal device.

[0062] The terminal device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the terminal device may also include input / output devices, network access devices, buses, etc.

[0063] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), 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. A general-purpose processor can be a microprocessor or any conventional processor.

[0064] The memory can be an internal storage unit of the terminal device, such as a hard drive or RAM. The memory can also be an external storage device of the terminal device, such as a plug-in hard drive, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory can include both internal and external storage units of the terminal device. The memory is used to store the program code 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 will be output.

[0065] The aforementioned computer storage medium and terminal equipment are created based on the control method of the aforementioned converter blowing system. Their technical functions and beneficial effects will not be elaborated here. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0066] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A control method for a converter blowing system, characterized in that, include: Acquire the measurement parameters at the current moment, including the molten pool temperature and carbon content, as well as the corresponding status parameters, including the cumulative oxygen consumption, the cumulative amount of auxiliary materials added, the real-time gun position, and the oxygen supply flow rate; Generate a correlation dataset between the measured parameters and the state parameters, and input the correlation dataset into the decarbonization rate model and the heating rate model respectively; The gun position correction factor is calculated based on the gun position change and reaction rate change output by the decarbonization rate model, and the base decarbonization rate is corrected based on the gun position correction factor to determine the target decarbonization rate. Obtain the heating rate output by the heating rate model; Based on the target decarbonization rate and the heating rate, control commands are generated and executed, including oxygen supply adjustment commands, coolant adjustment commands, and gun position adjustment commands.

2. The control method for the converter blowing system according to claim 1, characterized in that, The step of calculating the gun position correction factor based on the gun position change and reaction rate change output by the decarburization rate model, and correcting the basic decarburization rate based on the gun position correction factor to determine the target decarburization rate includes: The associated dataset is input into the pre-trained decarburization rate model, and the decarburization rate model extracts the shape features of the gun position waveform and outputs the gun position change coefficient. The carbon diffusion driving force is determined, and the basic decarbonization rate is generated based on the carbon diffusion driving force, reaction rate constant, and oxygen supply, and corrected by a penalty factor, wherein the penalty factor is calculated based on current measurement data. Based on the change in reaction rate and the change coefficient of gun position, the gun position correction factor is calculated according to a preset formula. The base decarburization rate is adjusted according to the gun position correction factor to generate the target decarburization rate.

3. The control method for the converter blowing system according to claim 2, characterized in that, The step of inputting the associated dataset into the pre-trained decarburization rate model, extracting the shape features of the gun position waveform from the decarburization rate model, and outputting the gun position change coefficient includes: Input the associated dataset into the decarbonization rate model; The decarbonization rate model extracts local waveform features through at least one convolutional kernel and generates correction coefficients through dimensionality reduction using a pooling layer. The correction coefficients are fused with the local waveform features extracted by each of the convolution kernels to obtain fused features; The target waveform features are extracted from the fused features through global max pooling, and the extracted target waveform features are mapped to generate the gun position change coefficient based on the fully connected layer.

4. The control method for the converter blowing system according to claim 2, characterized in that, The step of determining the carbon diffusion driving force, calculating based on the carbon diffusion driving force, reaction rate constant, and oxygen supply, and generating the basic decarbonization rate based on a penalty factor, wherein the penalty factor is calculated based on the current sputtering pulse data, includes: The reaction rate constant is determined based on the molten pool temperature measured by the current sub-lance, and the carbon diffusion driving force is determined based on the current carbon content; Obtain the ratio of the current oxygen supply flow rate to the weight of molten steel, and determine the penalty factor calculated based on the current measurement data; The basic decarburization rate is obtained by multiplying the reaction rate constant, the carbon diffusion driving force, the penalty factor, and the ratio of oxygen supply flow rate to molten steel weight.

5. The control method for the converter blowing system according to claim 4, characterized in that, The penalty factor includes a molten pool penalty factor and a splash penalty factor. The step of determining the penalty factor calculated based on the current measurement data further includes: Obtain the current gun position measurement value from the current measurement data, and calculate the degree of gun position fluctuation; Based on the preset mapping relationship between gun position fluctuation and molten pool penalty factor, the molten pool penalty factor corresponding to the degree of gun position fluctuation is determined; Obtain the current number of splash pulses from the current measurement data, and determine the splash penalty factor according to the preset mapping relationship between the number of splash pulses and the splash penalty factor.

6. The control method for the converter blowing system according to claim 1, characterized in that, The step of generating and executing control commands based on the target decarbonization rate and the heating rate, wherein the control commands include oxygen supply adjustment commands, coolant adjustment commands, and gun position adjustment commands, includes: If the target decarbonization rate is higher than the preset target range, a first oxygen supply adjustment command and a first gun position adjustment command are generated. Execute the first oxygen supply adjustment command and the first gun position adjustment command to reduce the oxygen supply flow rate of the first preset value through the first oxygen supply adjustment command, and to increase the gun position of the second preset value through the first gun position adjustment command. If the target decarbonization rate is lower than the preset target range, a second oxygen supply adjustment command and a second gun position adjustment command are generated. The second oxygen supply adjustment command and the second gun position adjustment command are executed to increase the oxygen supply flow rate of the first preset value through the second oxygen supply adjustment command and to decrease the gun position of the second preset value through the second gun position adjustment command.

7. The control method for the converter blowing system according to claim 1, characterized in that, The step of obtaining the heating rate output by the heating rate model includes: Determine the molten pool temperature sequence in the associated dataset and input the molten pool temperature sequence into the heating rate model, wherein the heating rate model includes a first branch and a second branch; The first branch is used to extract feature maps from the molten pool temperature sequence and generate heat loss correction coefficients. The molten pool temperature sequence is processed by the second branch to generate a slag layer penalty factor; The heating rate correction coefficient is obtained by adding the heat loss correction coefficient to the slag layer penalty factor and then adding a unit value. The heating rate is determined based on the heating rate correction factor.

8. The control method for the converter blowing system as described in claim 7, characterized in that, After the step of determining the heating rate based on the heating rate correction coefficient, the method further includes: Determine the target rate, and determine the corresponding difference between the heating rate and the target rate; The parameter adjustment amount is determined based on the difference and the preset coefficient, and the target instruction is generated based on the parameter adjustment amount.

9. The control method for the converter blowing system according to claim 1, characterized in that, The step of generating and executing control commands based on the target decarbonization rate and the heating rate, wherein the control commands include oxygen supply adjustment commands, coolant adjustment commands, and gun position adjustment commands, further includes: The predicted temperature corresponding to the target time point is determined based on the heating rate. If the predicted temperature is greater than the preset temperature at the target time point, a first coolant adjustment command is generated. Execute the first coolant adjustment command to add the corresponding coolant via the first coolant adjustment command; If the predicted temperature is less than or equal to the preset temperature, a target control command is generated and executed to pause coolant delivery and increase oxygen supply or lower the gun position.

10. A terminal device, characterized in that, It includes a memory and a processor; the memory stores program code that can be executed by the processor; the program code is used to execute the control method of the converter blowing system according to any one of claims 1 to 9.

Images