A method and system for intelligent control of gas flow in a continuous annealing furnace

Abstract

This invention provides an intelligent control method and system for gas flow in a continuous annealing furnace, comprising: collecting and preprocessing key process parameter data during the operation of the continuous annealing furnace; inputting the key process parameter data into a trained gas flow prediction model to predict and output the target gas flow in each zone of the continuous annealing furnace; the gas flow prediction model is constructed by combining a heat transfer physics model and a data-driven AI model; the heat transfer physics model calculates the optimal furnace temperature setpoint based on the target strip temperature, and the data-driven AI model predicts the target gas flow based on the optimal furnace temperature setpoint; the prediction result conforms to physical laws and also learns the complex nonlinear relationship between actual combustion and control through the data-driven model; correcting the target gas flow based on production process parameters; when the difference between the corrected target gas flow and the real-time detected gas flow exceeds a preset threshold, feedback adjustment of the gas flow in each furnace zone is performed based on the actual strip temperature, target temperature, and furnace temperature using a PID control algorithm. This invention enables intelligent control of the gas flow in a continuous annealing furnace.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation / heat treatment equipment control technology, and specifically refers to a method and system for intelligent control of gas flow in a continuous annealing furnace. Background Technology

[0002] Continuous annealing furnaces are core equipment in the heat treatment of metal strips. Their function is to heat the metal strips to the target temperature, achieving phase transformation and performance optimization. They are widely used in the production of stainless steel, aluminum, and other strips. The core challenges in controlling continuous annealing furnaces lie in the following: when operating conditions such as strip specifications, belt speed, and furnace calorific value change, the gas flow adjustment lags, resulting in significant furnace temperature fluctuations. Furthermore, they cannot effectively handle high-specification, complex belt speed scenarios. These problems lead to poor consistency in annealed product quality, low production efficiency, and an inability to meet the modern industrial demand for high-precision strip heat treatment.

[0003] Existing technologies for controlling the gas flow in continuous annealing furnaces mainly rely on manual experience-based adjustments or fixed-rule algorithms. However, these methods have significant limitations: manual adjustments cannot respond in real time to complex operating conditions and exhibit fluctuations under various conditions; fixed-rule algorithms have poor adaptability, cannot accurately adjust the flow in high-specification change scenarios, and lack intelligent prediction and optimization mechanisms; existing technologies generally suffer from unstable control and insufficient precision, failing to meet the modern industrial demands for high-precision, high-efficiency, rapid-response, stable, and reliable control of gas flow in continuous annealing furnaces, thus limiting annealing quality and production efficiency. Summary of the Invention

[0004] To address the technical problems existing in the prior art, the present invention provides a method and system for intelligent control of gas flow in a continuous annealing furnace, the technical solution of which is as follows: On the one hand, a method for intelligent control of gas flow in a continuous annealing furnace is provided, the method comprising: S1. Collect and preprocess data of key process parameters during the operation of the continuous annealing furnace; S2. Input the key process parameter data into the trained gas flow prediction model to predict and output the target gas flow of each zone of the continuous annealing furnace. The construction of the gas flow prediction model combines a heat transfer physics model and a data-driven AI model. The heat transfer physics model calculates the optimal furnace temperature setpoint based on the target zone temperature, and the data-driven AI model predicts the target gas flow based on the optimal furnace temperature setpoint. The prediction results not only conform to physical laws, but also learn the complex nonlinear relationship between actual combustion and control through the data-driven model. S3. Correct the target gas flow rate by combining the production process parameters to obtain the corrected target gas flow rate; S4. When the difference between the corrected target gas flow rate and the real-time detected gas flow rate exceeds a preset threshold, the gas flow rate of each furnace zone is adjusted by feedback through a PID control algorithm based on the actual temperature of the strip, the target temperature, and the furnace temperature.

[0005] On the other hand, a continuous annealing furnace gas flow intelligent control system is provided, the system comprising: The data acquisition and preprocessing module is used to acquire and preprocess key process parameter data during the operation of the continuous annealing furnace; The prediction module is used to input the key process parameter data into the trained gas flow prediction model and predict the target gas flow of each zone of the continuous annealing furnace. The construction of the gas flow prediction model combines a heat transfer physics model and a data-driven AI model. The heat transfer physics model calculates the optimal furnace temperature setpoint based on the target zone temperature, and the data-driven AI model predicts the target gas flow based on the optimal furnace temperature setpoint. The prediction results not only conform to physical laws, but also learn the complex nonlinear relationship between actual combustion and control through the data-driven model. The correction module is used to correct the target gas flow rate by combining the production process parameters to obtain the corrected target gas flow rate. The feedback adjustment module is used to adjust the gas flow rate of each furnace zone based on the actual temperature of the strip, the target temperature, and the furnace temperature when the difference between the corrected target gas flow rate and the real-time detected gas flow rate exceeds a preset threshold.

[0006] On the other hand, an electronic device is provided, comprising a processor and a memory, wherein the memory stores at least one instruction, which is loaded and executed by the processor to implement the above-described intelligent control method for gas flow in a continuous annealing furnace.

[0007] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the above-described intelligent control method for gas flow in a continuous annealing furnace.

[0008] The beneficial effects of the technical solution provided by this invention include at least the following: By constructing flow-parameter correlation models (such as process tables and AI models) for different strip specifications through a historical data processing module, the target gas flow rate can be accurately predicted when a change in strip specifications is detected. Compared with existing technologies, this module overcomes the problem of relying solely on fixed rules or single parameter correlation, effectively solving the mismatch in gas flow rate adjustment when specifications change, and significantly improving the accuracy of flow prediction and the adaptability of system operating conditions. The target gas flow rate predicted by the model is dynamically corrected in conjunction with production process parameters such as belt speed, furnace calorific value, and combustion coefficient. Unlike existing technologies that mostly rely on fixed parameters or simple manual adjustments, this invention achieves efficient adaptation of gas flow rate to current production conditions through intelligent adjustment of process parameters, improving the accuracy and efficiency of flow control. The PID control algorithm used enables real-time feedback adjustment of gas flow rate, combining the target gas flow rate with the real-time collected flow rate to achieve real-time closed-loop control of flow rate and temperature. Existing technologies often employ fixed control logic or lack feedback regulation. This module effectively solves the control lag problem, significantly improving system response speed and long-term stability. An emergency temperature over-limit handling strategy is designed; when the furnace temperature exceeds a preset threshold, the corresponding furnace flow rate is adjusted according to the degree of over-limit. Compared to existing technologies lacking emergency mechanisms or experiencing delayed emergency responses, this module achieves intelligent response to furnace temperature, effectively preventing equipment damage from overheating and ensuring equipment operational safety. It integrates multi-dimensional parameters (temperature, strip specifications, gas flow rate, furnace process parameters, etc.) to achieve intelligent decision-making, enabling intelligent and precise control of the continuous annealing furnace gas flow rate. This effect overcomes the limitations of existing technologies relying on single-parameter correlation and control methods, achieving intelligent regulation under complex operating conditions and improving system operational quality and production efficiency. Through the collaborative work of various functional modules, the system possesses good stability and robustness, maintaining stable operation even under complex conditions such as changes in strip specifications, furnace temperature fluctuations, and sudden changes in process parameters. Compared to the weak control stability of existing technologies, this system significantly improves equipment reliability, ensuring the continuous and efficient operation of the production process. These beneficial effects collectively enable intelligent, precise, efficient, safe, stable, and reliable control of the gas flow in continuous annealing furnaces, solving various problems existing in current technologies, meeting the modern industrial demand for high-precision and high-efficiency heat treatment equipment control, significantly improving production quality and efficiency, and ensuring equipment operation safety. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1This is a flowchart of an intelligent control method for gas flow in a continuous annealing furnace provided in an embodiment of the present invention; Figure 2 This is a block diagram of an intelligent control system for the gas flow of a continuous annealing furnace provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0011] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0012] This invention provides an intelligent control method for the gas flow of a continuous annealing furnace. This method can be implemented by an electronic device, which can be a terminal or a server. Figure 1 The diagram shown is a flowchart of the method. The processing flow may include the following steps:

[0013] S1. Collect and preprocess data of key process parameters during the operation of the continuous annealing furnace; Optionally, the key process parameter data include: Furnace temperature parameters: Measured values ​​of thermocouple temperatures in each zone of the annealing furnace; Strip specifications: width, thickness, steel grade; Belt operating parameters: belt speed; Gas flow parameters: Measured gas flow values ​​for each zone of the annealing furnace; Auxiliary process parameters: calorific value of fuel gas, combustion coefficient of each furnace zone; The key process parameter data cover the key indicators affecting the gas flow control of the continuous annealing furnace, and the collection scope covers all stages of the continuous annealing furnace from inlet to outlet.

[0014] Optionally, the data acquisition method for the key process parameters includes: An automated data acquisition method using PLC interface and parameter mapping is adopted. By using a predefined parameter mapping table in the program, PLC control signals are accurately mapped to parameter values ​​such as temperature and flow rate, ensuring the accuracy and consistency of the acquired data. Real-time data is continuously collected at set time intervals to ensure the timeliness of data collection.

[0015] The collected real-time data is continuously stored in the cache to form a dynamic historical dataset containing the most recent N process parameter sequences.

[0016] S2. Input the key process parameter data into the trained gas flow prediction model to predict and output the target gas flow of each zone of the continuous annealing furnace. The construction of the gas flow prediction model combines a heat transfer physics model and a data-driven AI model. The heat transfer physics model calculates the optimal furnace temperature setpoint based on the target zone temperature, and the data-driven AI model predicts the target gas flow based on the optimal furnace temperature setpoint. The prediction results not only conform to physical laws, but also learn the complex nonlinear relationship between actual combustion and control through the data-driven model. Optionally, the construction of the gas flow prediction model includes: S21. Establish a physical model for predicting strip temperature based on the heat transfer physical mechanism, and calibrate the key physical coefficients in the model using historical data to form a heat transfer physical model that can accurately reflect the heat transfer behavior under different working conditions. S22. Using the heat transfer physical model, the optimal furnace temperature setpoint required to meet the target temperature is calculated in reverse. S23. Based on a large number of "operating condition-target zone temperature-optimal furnace temperature setpoint" sample pairs, train a data-driven AI model. After training, predict the gas flow rate of each zone according to the real-time operating condition and target zone temperature.

[0017] Optionally, S21 specifically includes: S211, Establishment of physical equations; Based on the heat exchange mechanism dominated by radiation and convection in open-flame heating furnaces, a comprehensive heat transfer differential equation is established for the temperature rise of strip steel within the furnace zone. The core of this comprehensive heat transfer differential equation is energy balance: the net heat absorbed by the strip steel equals the change in its internal energy. This balance relationship is described by the following comprehensive heat transfer differential equation: in, For the quality of the micro-element strip steel, The specific heat capacity related to the strip temperature, For strip temperature, For time; Radiative heat transfer Based on the Stefan-Boltzmann law, it is expressed as , It is the Stefan-Boltzmann constant. The effective emissivity of the strip surface to be solved is... The heated surface area of ​​the micro-element segment. The ambient temperature of the furnace; Convection heat transfer : expressed as , Let be the convective heat transfer coefficient to be solved; The comprehensive heat transfer differential equation is applied to the discrete furnace zone length. Integrate on the above, and consider belt speed. The impact, The comprehensive calculation equations for single-area simulation are obtained as follows: in, , These are the temperatures at which the strip enters and leaves the furnace zone, respectively. For multiple furnace zones, because the furnace zones are interconnected, the temperature of the previous furnace zone... Directly used as the current furnace area The first furnace area The furnace is already warm when it is put in; ambient temperature is used as a substitute, and is expressed as follows: The last furnace area That is, the outlet is warm, which is expressed as ; S212, Solving for coefficients; Based on historical stable production data, an optimization algorithm is used to perform parameter inversion and solve for the effective emissivity ε and the convective heat transfer coefficient h. The specific process is as follows: Collect historical stable production data covering multiple specifications and operating conditions, including: furnace temperature in each zone. strip speed Strip specifications, used for calculating quality. With specific heat capacity and ambient temperature and actual measured value of outlet temperature ; Substituting the historical stable production data as a sample into the comprehensive calculation equation, we obtain the predicted outlet temperature under the corresponding operating conditions. Based on this, an objective function for the sum of squared errors is constructed using the difference between the predicted and measured outlet temperatures: in Given the sample size, this problem is a typical nonlinear least squares optimization problem. Since ε and h both have definite physical meanings, feasible boundary ranges for the parameters need to be set during the solution process: ε belongs to the surface radiation characteristic parameter, and its value should be within the range of... The convective heat transfer coefficient h must be a positive value, and based on empirical data, the range is further limited to... The feasible boundary range is used to avoid non-physical solutions or numerical divergences; Under the constraints of the feasible boundary range, the objective function is solved using the numerical optimization algorithm L-BFGS-B, which supports boundary constraints. The solution process includes: (1) Select initial values ​​within the parameter boundaries; (2) Calculate the sum of squared errors and gradients under the current parameters; (3) Update parameters along the descent direction and automatically project them to the boundary range; (4) Terminate when the objective function converges or the parameter change is below the threshold; This yields the optimal heat transfer coefficient combination that minimizes prediction error. ; By traversing all historical operating conditions, a system is established that maps operating parameters to the optimal combination of heat transfer coefficients. The discrete mapping relationship constitutes a heat transfer physical model that can automatically match coefficients according to real-time operating conditions, also known as the strip temperature prediction physical model.

[0018] Optionally, S22 specifically includes: S221, Problem Definition; Temperature at a given target outlet (Usually set based on process requirements) and under the premise of current operating conditions (belt speed, specifications, etc.), the furnace temperature to be set for each zone is calculated in reverse. The optimization objective of the problem is to adjust the furnace temperature according to the current operating conditions and zone settings. The final strip temperature prediction value calculated using the physical model for strip temperature prediction infinitely close ; S222, Optimization solution; The problem is constructed as a nonlinear programming problem with the objective function being the mean square error: in Set a vector for the furnace temperature to be optimized. The number of samples; Its constraints are the upper and lower limits of the process for each furnace temperature, namely: ; Based on this, the objective function is solved using the gradient descent method, and the specific process is as follows: (1) Initialize furnace temperature setpoint: Set the initial vector based on historical stable operating conditions or the median value of the interval. ; (2) Error and gradient calculation: The predicted temperature of the outlet zone is calculated using the comprehensive calculation equation, and the partial derivative of the objective function with respect to the set temperature of each furnace zone is obtained to obtain the gradient direction; (3) Iterative update: ,in Step size (can be selected as fixed step size or adaptive mode); (4) Boundary correction: If the updated furnace temperature exceeds the process range, it is projected back to the feasible range; (5) Convergence judgment: When the decrease of the objective function or the change in furnace temperature is lower than the set threshold, the iteration terminates and the optimal furnace temperature setpoint that satisfies the objective temperature is obtained. This transforms the final process quality objective—the outlet temperature—into the process control objective—the furnace temperature setpoint for each zone, thereby providing a direct basis for optimizing furnace temperature settings.

[0019] Optionally, S23 specifically includes: S231. Construct the training dataset; Construct a training dataset covering the entire operating range: For various typical operating condition combinations and corresponding stable outlet temperatures in historical data, generate a large number of "operating condition-target temperature-optimal furnace temperature setpoint" sample pairs to form a training dataset; Extract the actual gas flow rate of each zone corresponding to the dataset samples as the output label; The model's input feature vector integrates information from multiple sources, including: Furnace temperature characteristics: measured temperature values ​​of each thermocouple in each zone of the annealing furnace, and the optimal furnace temperature setpoint calculated in reverse by S22; Strip specifications: width, thickness, steel type; Strip operating characteristics: belt speed; Auxiliary process characteristics: calorific value of fuel gas, combustion coefficient of each furnace zone; Comprehensive interaction features: Transformation and interaction features constructed based on the above basic features; S232, Model Training; The AI ​​models include: multi-output regression tree models (such as XGBoost), deep neural networks; The model takes feature vectors as input and gas flow rates in each zone as output; The model is trained using mean squared error as the loss function. It learns to meet the optimal furnace temperature setting under given operating conditions and objectives, implicitly satisfying the actual gas flow rate of each zone corresponding to the target temperature. It automatically captures complex factors that are difficult to describe with physical formulas: operating conditions, combustion efficiency, and equipment characteristics.

[0020] S3. Correct the target gas flow rate by combining the production process parameters to obtain the corrected target gas flow rate; Optionally, S3 specifically includes: S31, belt speed correction; Based on the belt speed parameter, the target gas flow rate is adjusted according to the ratio of the belt speed change to the target gas flow rate. The calculation formula is as follows: in The target gas flow rate is V1 and V2, which are the process speeds before and after the production line speed adjustment, respectively, and α is the adjustment coefficient (usually around 1.0). S32, Furnace area calorific value correction; Based on the calorific value parameters of the gas, the target gas flow rate is adjusted according to the proportion of the change in calorific value relative to the target gas flow rate. The calculation formula is as follows: Where q1 and q2 are the actual values ​​before and after the change in calorific value, respectively, and β is the adjustment coefficient (usually around 1.0, which needs to be adjusted according to the specific scenario). Based on the above corrections, the actual gas flow rate that can be used for control issuance is... for: .

[0021] S4. When the difference between the corrected target gas flow rate and the real-time detected gas flow rate exceeds a preset threshold, the gas flow rate of each furnace zone is adjusted by feedback through a PID control algorithm based on the actual temperature of the strip, the target temperature, and the furnace temperature.

[0022] Optionally, S4 specifically includes: Target temperature of strip T aim With real-time collected strip temperature T real As input, the adjustment value of the output gas flow rate is determined by the core control formula: in K p , K i , K d These are the proportional gain coefficient, integral gain coefficient, and differential gain coefficient; T error1 It represents the cumulative temperature difference over a recent period; T error2 It is the rate of change of temperature difference over a recent period of time; The trigger condition for feedback control is that the difference between the corrected target gas flow rate and the real-time detected gas flow rate exceeds a preset threshold. This setting aims to reduce system fluctuations and improve system stability.

[0023] Optionally, the method further includes: designing an emergency temperature over-limit handling strategy, wherein when the furnace zone temperature T2 exceeds a preset threshold T1 (e.g., 1185 degrees), the corresponding furnace zone gas flow rate is adjusted according to the degree of over-limit to achieve emergency furnace temperature stabilization; The emergency temperature exceedance handling strategy includes: First time exceeding the limit: Appropriately reduce the gas flow rate in the corresponding furnace area (e.g., reduce the flow rate by 100 m³ / h) to prevent the temperature from rising further; Continuous over-limit: The flow rate decrease is gradually increased to accelerate the cooling process. The calculation formula is as follows: in The coefficients are set based on experience; Large exceedance: Adjust the flow rate proportionally to quickly stabilize the temperature and ensure equipment safety. The calculation formula is as follows: Where n is the sequence number of the collected data, the sequence number of the data collected at the current moment is n, and the sequence number of the data collected at the previous moment is n-1.

[0024] like Figure 2 As shown in the figure, this embodiment of the invention also provides an intelligent control system for the gas flow of a continuous annealing furnace, the system comprising: The data acquisition and preprocessing module 210 is used to acquire and preprocess various key process parameter data during the operation of the continuous annealing furnace. The prediction module 220 is used to input the key process parameter data into the trained gas flow prediction model and predict and output the target gas flow of each zone of the continuous annealing furnace. The construction of the gas flow prediction model combines a heat transfer physics model and a data-driven AI model. The heat transfer physics model calculates the optimal furnace temperature setpoint based on the target zone temperature, and the data-driven AI model predicts the target gas flow based on the optimal furnace temperature setpoint. The prediction results not only conform to physical laws, but also learn the complex nonlinear relationship between actual combustion and control through the data-driven model. The correction module 230 is used to correct the target gas flow rate in combination with the production process parameters to obtain the corrected target gas flow rate; The feedback adjustment module 240 is used to adjust the gas flow rate of each furnace zone by means of a PID control algorithm when the difference between the corrected target gas flow rate and the real-time detected gas flow rate exceeds a preset threshold, based on the actual temperature of the strip, the target temperature and the furnace temperature.

[0025] The intelligent control system for gas flow in a continuous annealing furnace provided in this embodiment of the invention has a functional structure that corresponds to the intelligent control method for gas flow in a continuous annealing furnace provided in this embodiment of the invention, and will not be described again here.

[0026] Figure 3This is a schematic diagram of the structure of an electronic device 300 provided in an embodiment of the present invention. The electronic device 300 may vary considerably due to different configurations or performance. It may include one or more central processing units (CPUs) 301 and one or more memories 302. The memory 302 stores at least one instruction, which is loaded and executed by the processor 301 to implement the steps of the above-described intelligent control method for the gas flow of a continuous annealing furnace.

[0027] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including instructions that can be executed by a processor in a terminal to complete the aforementioned intelligent control method for the gas flow of a continuous annealing furnace. For example, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device.

[0028] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0029] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent control of gas flow in a continuous annealing furnace, characterized in that, The method includes: S1. Collect and preprocess data of key process parameters during the operation of the continuous annealing furnace; S2. Input the key process parameter data into the trained gas flow prediction model to predict and output the target gas flow of each zone of the continuous annealing furnace. The construction of the gas flow prediction model combines a heat transfer physics model and a data-driven AI model. The heat transfer physics model calculates the optimal furnace temperature setpoint based on the target zone temperature, and the data-driven AI model predicts the target gas flow based on the optimal furnace temperature setpoint. The prediction results not only conform to physical laws, but also learn the complex nonlinear relationship between actual combustion and control through the data-driven model. S3. Correct the target gas flow rate by combining the production process parameters to obtain the corrected target gas flow rate; S4. When the difference between the corrected target gas flow rate and the real-time detected gas flow rate exceeds a preset threshold, the gas flow rate of each furnace zone is adjusted by feedback through a PID control algorithm based on the actual temperature of the strip, the target temperature, and the furnace temperature.

2. The method according to claim 1, characterized in that, The key process parameter data include: Furnace zone temperature parameters: Measured values ​​of thermocouple temperatures in each zone of the annealing furnace; Strip specifications: width, thickness, steel grade; Belt operating parameters: belt speed; Gas flow parameters: Measured gas flow values ​​for each zone of the annealing furnace; Auxiliary process parameters: calorific value of fuel gas, combustion coefficient of each furnace zone; The key process parameter data cover the key indicators affecting the gas flow control of the continuous annealing furnace, and the collection scope covers all stages of the continuous annealing furnace from inlet to outlet.

3. The method according to claim 1, characterized in that, The construction of the gas flow prediction model includes: S21. Establish a physical model for predicting strip temperature based on the heat transfer physical mechanism, and calibrate the key physical coefficients in the model using historical data to form a heat transfer physical model that can accurately reflect the heat transfer behavior under different working conditions. S22. Using the heat transfer physical model, the optimal furnace temperature setpoint required to meet the target temperature is calculated in reverse. S23. Based on a large number of "operating condition-target zone temperature-optimal furnace temperature setpoint" sample pairs, train a data-driven AI model. After training, predict the gas flow rate of each zone according to the real-time operating condition and target zone temperature.

4. The method according to claim 3, characterized in that, S21 specifically includes: S211, Establishment of physical equations; Based on the heat exchange mechanism dominated by radiation and convection in open-flame heating furnaces, a comprehensive heat transfer differential equation is established for the temperature rise of strip steel within the furnace zone. The core of this comprehensive heat transfer differential equation is energy balance: the net heat absorbed by the strip steel equals the change in its internal energy. This balance relationship is described by the following comprehensive heat transfer differential equation: in, For the quality of the micro-element strip steel, The specific heat capacity related to the strip temperature, For strip temperature, For time; Radiative heat transfer Based on the Stefan-Boltzmann law, it is expressed as , It is the Stefan-Boltzmann constant. The effective emissivity of the strip surface to be solved is... The heated surface area of ​​the micro-element segment. The ambient temperature of the furnace; Convection heat transfer : expressed as , Let be the convective heat transfer coefficient to be solved; The comprehensive heat transfer differential equation is applied to the discrete furnace zone length. Integrate on the above, and consider the belt speed. The impact, The comprehensive calculation equations for single-area simulation are obtained as follows: in, , These are the temperatures at which the strip enters and leaves the furnace zone, respectively. For multiple furnace zones, because the furnace zones are interconnected, the temperature of the previous furnace zone... Directly used as the current furnace area The first furnace area The furnace is already warm when it is put in; ambient temperature is used as a substitute, and is expressed as follows: The last furnace area That is, the outlet is warm, which is expressed as ; S212, Solving for coefficients; Based on historical stable production data, an optimization algorithm is used to perform parameter inversion and solve for the effective emissivity ε and the convective heat transfer coefficient h. The specific process is as follows: Collect historical stable production data covering multiple specifications and operating conditions, including: furnace temperature in each zone. strip speed Strip specifications, used for calculating quality. With specific heat capacity and ambient temperature and actual measured value of outlet temperature ; Substituting the historical stable production data as a sample into the comprehensive calculation equation, we obtain the predicted outlet temperature under the corresponding operating conditions. Based on this, an objective function for the sum of squared errors is constructed using the difference between the predicted and measured outlet temperatures: in Given the sample size, this problem is a typical nonlinear least squares optimization problem. Since ε and h both have definite physical meanings, feasible boundary ranges for the parameters need to be set during the solution process: ε belongs to the surface radiation characteristic parameter, and its value should be within the range of... The convective heat transfer coefficient h must be a positive value, and based on empirical data, the range is further limited to... The feasible boundary range is used to avoid non-physical solutions or numerical divergences; Under the constraints of the feasible boundary range, the objective function is solved using the numerical optimization algorithm L-BFGS-B, which supports boundary constraints. The solution process includes: (1) Select initial values ​​within the parameter boundaries; (2) Calculate the sum of squared errors and gradients under the current parameters; (3) Update parameters along the descent direction and automatically project them to the boundary range; (4) Terminate when the objective function converges or the parameter change is below the threshold; This yields the optimal heat transfer coefficient combination that minimizes prediction error. ; By traversing all historical operating conditions, a system is established that maps operating parameters to the optimal combination of heat transfer coefficients. The discrete mapping relationship constitutes a heat transfer physical model that can automatically match coefficients according to real-time operating conditions, also known as the strip temperature prediction physical model.

5. The method according to claim 4, characterized in that, S22 specifically includes: S221, Problem Definition; Temperature at a given target outlet Under the premise of the current operating conditions, reverse the calculation to determine the furnace temperature to be set for each zone. The optimization objective of the problem is to adjust the furnace temperature according to the current operating conditions and zone settings. The final strip temperature prediction value calculated using the physical model for strip temperature prediction infinitely close ; S222, Optimization solution; The problem is constructed as a nonlinear programming problem with the objective function being the mean square error: in Set a vector for the furnace temperature to be optimized. The number of samples; Its constraints are the upper and lower limits of the process for each furnace temperature, namely: ; Based on this, the objective function is solved using the gradient descent method, and the specific process is as follows: (1) Initialize furnace temperature setpoint: Set the initial vector based on historical stable operating conditions or the median value of the interval. ; (2) Error and gradient calculation: The predicted temperature of the outlet zone is calculated using the comprehensive calculation equation, and the partial derivative of the objective function with respect to the set temperature of each furnace zone is obtained to obtain the gradient direction; (3) Iterative update: ,in Step size; (4) Boundary correction: If the updated furnace temperature exceeds the process range, it is projected back to the feasible range; (5) Convergence judgment: When the decrease of the objective function or the change in furnace temperature is lower than the set threshold, the iteration terminates and the optimal furnace temperature setpoint that satisfies the objective temperature is obtained. This transforms the final process quality objective—the outlet temperature—into the process control objective—the furnace temperature setpoint for each zone, thereby providing a direct basis for optimizing furnace temperature settings.

6. The method according to claim 5, characterized in that, S23 specifically includes: S231. Construct the training dataset; Construct a training dataset covering the entire operating range: For various typical operating condition combinations and corresponding stable outlet temperatures in historical data, generate a large number of "operating condition-target temperature-optimal furnace temperature setpoint" sample pairs to form a training dataset; Extract the actual gas flow rate of each zone corresponding to the dataset samples as the output label; The model's input feature vector integrates information from multiple sources, including: Furnace temperature characteristics: measured temperature values ​​of each thermocouple in each zone of the annealing furnace, and the optimal furnace temperature setpoint calculated in reverse by S22; Strip specifications: width, thickness, steel type; Strip operating characteristics: belt speed; Auxiliary process characteristics: calorific value of fuel gas, combustion coefficient of each furnace zone; Comprehensive interaction features: Transformation and interaction features constructed based on the above basic features; S232, Model Training; The AI ​​model includes: a multi-output regression tree model and a deep neural network; The model takes feature vectors as input and gas flow rates in each zone as output; The model is trained using mean squared error as the loss function. It learns to meet the optimal furnace temperature setting under given operating conditions and objectives, implicitly satisfying the actual gas flow rate of each zone corresponding to the target temperature. It automatically captures complex factors that are difficult to describe with physical formulas: operating conditions, combustion efficiency, and equipment characteristics.

7. The method according to claim 1, characterized in that, S3 specifically includes: S31, belt speed correction; Based on the belt speed parameter, the target gas flow rate is adjusted according to the ratio of the belt speed change to the target gas flow rate. The calculation formula is as follows: in The target gas flow rate is V1 and V2, which are the process speeds before and after the production line speed adjustment, respectively, and α is the adjustment coefficient. S32, Furnace area calorific value correction; Based on the calorific value parameters of the gas, the target gas flow rate is adjusted according to the proportion of the change in calorific value relative to the target gas flow rate. The calculation formula is as follows: Where q1 and q2 are the actual values ​​before and after the change in calorific value, respectively, and β is the adjustment coefficient; Based on the above corrections, the actual gas flow rate that can be used for control issuance is... for: .

8. The method according to claim 1, characterized in that, S4 specifically includes: Target temperature of the strip T aim With real-time collected strip temperature T real As input, the adjustment value of the output gas flow rate is determined by the core control formula: in K p , K i , K d These are the proportional gain coefficient, integral gain coefficient, and differential gain coefficient; T error1 It represents the cumulative temperature difference over a recent period; T error2 It is the rate of change of temperature difference over a recent period of time; The trigger condition for feedback control is that the difference between the corrected target gas flow rate and the real-time detected gas flow rate exceeds a preset threshold. This setting aims to reduce system fluctuations and improve system stability.

9. The method according to claim 1, characterized in that, The method further includes: designing an emergency temperature over-limit handling strategy, whereby when the furnace zone temperature T2 exceeds a preset threshold T1, the corresponding furnace zone gas flow rate is adjusted according to the degree of over-limit to achieve emergency furnace temperature stabilization. The emergency temperature exceedance handling strategy includes: First instance of exceeding the limit: Appropriately reduce the gas flow rate in the corresponding furnace area to prevent the temperature from rising further; Continuous over-limit: The flow rate decrease is gradually increased to accelerate the cooling process. The calculation formula is as follows: in The coefficients are set based on experience; Large exceedance: Adjust the flow rate proportionally to quickly stabilize the temperature and ensure equipment safety. The calculation formula is as follows: Where n is the sequence number of the collected data, the sequence number of the data collected at the current moment is n, and the sequence number of the data collected at the previous moment is n-1.

10. An intelligent control system for the gas flow rate of a continuous annealing furnace, characterized in that, The system includes: The data acquisition and preprocessing module is used to acquire and preprocess key process parameter data during the operation of the continuous annealing furnace; The prediction module is used to input the key process parameter data into the trained gas flow prediction model and predict the target gas flow of each zone of the continuous annealing furnace. The construction of the gas flow prediction model combines a heat transfer physics model and a data-driven AI model. The heat transfer physics model calculates the optimal furnace temperature setpoint based on the target zone temperature, and the data-driven AI model predicts the target gas flow based on the optimal furnace temperature setpoint. The prediction results not only conform to physical laws, but also learn the complex nonlinear relationship between actual combustion and control through the data-driven model. The correction module is used to correct the target gas flow rate by combining the production process parameters to obtain the corrected target gas flow rate. The feedback adjustment module is used to adjust the gas flow rate of each furnace zone based on the actual temperature of the strip, the target temperature, and the furnace temperature when the difference between the corrected target gas flow rate and the real-time detected gas flow rate exceeds a preset threshold.

Images