A stainless steel annealing furnace online process control method and system based on UKF-MPC
Patent Information
- Application Number
- CN202611047593.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-15
- Publication Date
- 2026-10-30
AI Technical Summary
传统PID/模糊PID控制：仅能实现单回路事后反馈，无法有效克服退火炉大滞后特性，炉温波动大、超调明显，对钢种切换、带钢规格变化、燃料热值波动、炉气黑度漂移等工况的自适应能力差，控制精度难以满足高端产品要求
1、本发明通过总括热吸收率法构建分段动态模型，结合无迹卡尔曼滤波实现带钢温度与总括热吸收率的联合估计，一方面可有效滤除现场高温计的测量噪声，提升温度反馈信号的可靠性；另一方面可在线辨识时变换热参数，实时修正模型偏差，显著增强了控制模型对退火炉多工况扰动的适配能力。
Smart Images
Figure CN123010390A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent control technology for the thermal process of stainless steel continuous annealing furnaces, specifically to an online process control method and system for stainless steel annealing furnaces based on unscented Kalman filtering (UKF) and model predictive control (MPC). Background Technology
[0002] Continuous annealing of stainless steel is a core process for improving the microstructure of cold-rolled strip, eliminating work hardening, and controlling mechanical properties and surface quality. The precision and uniformity of strip temperature control directly determine the grain size, yield strength, elongation, and surface finish of the product, and are key indicators for quality grading of high-end stainless steel products. The continuous annealing furnace includes a preheating section, a heating section, an air-cooling section, a mist-cooling section, and a water-cooling section. The furnace exhibits a three-dimensional radiation-convection coupled heat transfer process between furnace gas, furnace lining, and strip, resulting in strong nonlinearity, large hysteresis, time-varying parameters, and significant disturbances under various operating conditions.
[0003] Existing furnace temperature control mainly employs control schemes such as PID, fuzzy PID, and ordinary MPC, which have the following drawbacks: Traditional PID / fuzzy PID control can only achieve single-loop post-feedback, which cannot effectively overcome the large hysteresis characteristics of annealing furnaces. The furnace temperature fluctuates greatly and the overshoot is obvious. It has poor adaptability to conditions such as steel grade switching, strip steel specification changes, fuel calorific value fluctuations, and furnace gas emissivity drift. The control accuracy is difficult to meet the requirements of high-end products.
[0004] Single MPC control relies on an accurate mechanistic model, but the annealing furnace is subject to complex factors such as ternary radiative heat transfer, material property changes, and furnace lining aging. Fixed models are prone to serious mismatches. Furthermore, existing solutions do not effectively filter out high-temperature meter measurement noise and coupled heat transfer disturbances, resulting in insufficient prediction accuracy and poor robustness.
[0005] In addition, most existing control schemes do not perform zoned modeling for the differences in heat exchange mechanisms between the heating and cooling sections, nor do they include key time-varying parameters such as total heat absorption rate, total radiation exchange area, and convective heat transfer coefficient in the online identification scope, making it difficult to cover the high-precision temperature control requirements of the entire process. Summary of the Invention
[0006] The technical problem to be solved by this invention is to provide an online process control method and system for stainless steel annealing furnaces based on UKF-MPC, which can accurately adapt to the segmented heat exchange and multi-condition disturbance characteristics of continuous annealing furnaces, achieve low temperature fluctuation, no overshoot, and high stability control throughout the furnace, improve product qualification rate and reduce energy consumption.
[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: I. An Online Process Control Method for Stainless Steel Annealing Furnaces Based on UKF-MPC This invention provides an online process control method for a stainless steel annealing furnace based on UKF-MPC, which mainly includes the following steps: S1, Section division and variable definition: The stainless steel continuous annealing furnace is divided into multiple temperature control sections along the strip running direction, and then integrated into heating control area and cooling control area according to the heating and cooling process, and the control quantity and observation quantity of each temperature control section are defined respectively. S2, Establishment of segmented dynamic model: The total heat absorption rate method is used to construct a discrete dynamic model of strip temperature in each temperature control zone. The radiation-convection coupled heat transfer in the furnace is equivalent to the total heat absorption rate parameter, forming a nonlinear state-space model suitable for online calculation. S3, UKF Joint State Estimation: Using strip temperature and total heat absorption rate as joint state variables, the state variables are iteratively estimated based on unscented Kalman filtering, while filtering out observation noise and identifying the total heat absorption rate parameter online, and outputting the optimal state estimate. S4, MPC Rolling Optimization: Based on the optimal state estimate output by UKF, constrained model predictive controllers are constructed for the heating control zone and the cooling control zone respectively, and the optimal control sequence is solved with temperature tracking accuracy and control stability as objectives. S5, Partitioned Closed-Loop Execution: The control quantities solved by MPC are sent down to the underlying actuators, and the field measured data are collected according to the control cycle to update the UKF state and MPC prediction conditions, so as to realize the full-process online rolling closed-loop control.
[0008] Preferably, in step S1, the temperature control section includes a preheating section, a heating section, an air-cooling section, a mist-cooling section, and a water-cooling section; The heating control zone includes a preheating section and a heating section. Its controlled quantity is the furnace temperature of each temperature control section; its observed quantity is the strip temperature at the outlet of each section. The cooling control zone includes an air-cooled section, a mist-cooled section, and a water-cooled section. Its control quantity is the flow rate of the cooling medium in each temperature-controlled section; its observation quantity is the outlet strip temperature of each section.
[0009] Preferably, in step S2, the discrete dynamic model of the strip temperature is as follows:
[0010] In the formula, T s (k) represents the strip temperature of this section at the k-th sampling time; φ CF (k) represents the total heat absorption rate of this section at the k-th sampling time; u(k) represents the control input at the k-th sampling time (the furnace gas temperature setpoint for the heating control zone and the cooling medium flow rate setpoint for the cooling control zone); d(k) represents the measurable disturbance vector.
[0011] Preferably, the measurable disturbance vector includes: strip running speed, strip thickness, strip width, steel grade, inlet strip temperature, and air preheating temperature.
[0012] Preferably, in step S3, the state vector is defined:
[0013] in
[0014] Define the observation vector:
[0015] In the formula, g is the nonlinear state equation function; h is the nonlinear observation equation function; W and V are the current noise of the state equation and observation equation, respectively; Z is the measured value of the strip temperature at the outlet of each temperature control section, which is obtained by measuring through a preset radiation pyrometer.
[0016] Preferably, in step S3, the UKF state joint estimation specifically includes: S31: Set initial state values, process noise covariance, and measurement noise covariance; S32: Employ a symmetric sampling strategy to generate (2n+1) Sigma points (n=2) based on the current state mean and covariance. S33: Substitute all Sigma points into the state equation, calculate the Sigma point set after one prediction step, and sum them by weight to obtain the state prediction mean and prediction error covariance. S34: Substitute the predicted Sigma points into the observation equation to obtain the set of observed and predicted Sigma points, and calculate the weighted mean of the observed predictions, the observed and predicted covariance, and the cross-covariance between the state and the observations. S35: Calculate the Kalman gain based on the observed-predicted covariance and cross-covariance, and update the state estimate and error covariance matrix by combining the deviation between the measured temperature and the observed-predicted value at the current moment. S36: Output the optimal estimated value of the strip temperature and the total heat absorption rate identification value at the current moment, and enter the next sampling cycle.
[0017] Preferably, in step S4: Define the MPC objective function:
[0018] In the formula, T s,ref Np represents the target temperature value of the strip exit in each temperature control zone, Nc represents the prediction time domain, Nc represents the control time domain, and Δu represents the control increment coefficient used to maintain stability.
[0019] Preferably, in step S4, the constraints include: 1) Heating control zone: upper and lower limits of furnace temperature and upper limit of strip temperature rise rate in each temperature control zone; 2) Cooling control zone: upper and lower limits of cooling medium flow rate and upper limit of strip temperature drop rate in each temperature control zone.
[0020] Preferably, in step S5, within each control cycle, the current state estimated by UKF is used as the initial value, and the future N is predicted based on the piecewise dynamic model. p The temperature trajectory of the strip steel is determined step by step. Within the constraints, the control increment sequence that minimizes the objective function is solved by quadratic programming. Only the first control increment of the sequence is applied to the system to achieve rolling optimization. Meanwhile, measurable disturbances such as strip speed, specifications, and steel type are introduced into the prediction model in advance to achieve disturbance feedforward compensation and improve the response speed when switching operating conditions.
[0021] II. An Online Process Control System for a Stainless Steel Annealing Furnace Based on the same inventive concept, this invention also provides an online process control system for a stainless steel annealing furnace, employing the UKF-MPC-based online process control method for stainless steel annealing furnaces as described above, mainly including: 1) Zoned data acquisition module: used for real-time acquisition of furnace temperature, cooling medium flow rate, strip speed, and steel grade parameters; 2) Online identification module for total coefficient: used to calculate and update the total heat absorption rate φ of each temperature control zone in real time based on the heat balance formula. CF ; 3) Segmented UKF State Estimation Module: Used to output the estimated strip temperature and the identified total heat absorption rate for each temperature control zone; 4) Segmented MPC optimization module: used to output the optimal furnace temperature for each temperature control segment in the heating control zone and the optimal cooling medium flow rate for each temperature control segment in the cooling control zone; 5) Execution drive module: used to adjust the system heating valve group and cooling valve group according to the optimal furnace temperature and optimal cooling medium flow rate; 6) Setting storage module: used to store and call the target outlet temperature and process library parameters of each temperature control zone.
[0022] Compared with the prior art, the present invention has the following main advantages: 1. This invention constructs a piecewise dynamic model using the total heat absorption rate method and combines it with unscented Kalman filtering to achieve joint estimation of strip temperature and total heat absorption rate. On the one hand, it can effectively filter out the measurement noise of the field pyrometer and improve the reliability of the temperature feedback signal; on the other hand, it can identify time-varying thermal parameters online and correct model deviations in real time, significantly enhancing the adaptability of the control model to disturbances under multiple operating conditions of the annealing furnace.
[0023] 2. This invention adopts a partitioned model predictive control architecture, designing control objectives and constraints for the heating and cooling zones respectively. With the strip outlet temperature as the core objective, it automatically coordinates the control quantities of each segment through rolling optimization, which can effectively overcome the inherent characteristics of large lag and strong coupling of the annealing furnace, achieve low fluctuation and no overshoot control of the strip temperature throughout the process, greatly improve the uniformity of strip temperature, and significantly improve the product qualification rate.
[0024] 3. This invention can be directly deployed and implemented on the basis of existing PLC / DCS control systems without the need for adding a large number of new hardware detection and execution equipment. The transformation cost is low and the project implementation is strong. At the same time, by reducing ineffective heating and excessive cooling through precise temperature control, it effectively reduces fuel consumption and cooling medium usage, and has both good economic benefits and energy-saving value. Attached Figure Description
[0025] Figure 1 This is an overall flowchart of the online process control method for stainless steel annealing furnace in an embodiment of the present invention; Figure 2 This is a schematic diagram of the operation of the online process control system for the stainless steel annealing furnace in an embodiment of the present invention. Detailed Implementation
[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.
[0027] It should be noted that, depending on the implementation needs, the various steps / components described in this application can be broken down into more steps / components, or two or more steps / components or parts of the operation of steps / components can be combined into new steps / components to achieve the purpose of this invention.
[0028] In this invention, unless otherwise expressly specified and limited, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise expressly and specifically limited.
[0029] Example 1: This example provides an online process control method for a stainless steel annealing furnace based on UKF-MPC, such as... Figure 1 As shown, the main steps include the following: Step S1, Segmentation and Variable Definition The stainless steel continuous annealing furnace is divided into five independent temperature control zones along the strip running direction: preheating zone, heating zone, air cooling zone, mist cooling zone, and water cooling zone. These zones are then integrated into two collaborative control areas: a heating control area and a cooling control area. 1) Heating control zone, including: preheating section and heating section; its controlled quantity is: furnace temperature of each temperature control section; its observed quantity is: strip temperature at the outlet of each section; 2) Cooling control zone, including: air cooling section, mist cooling section, water cooling section; its control quantity is: the flow rate of cooling medium in each temperature control section; its observation quantity is: the outlet strip temperature of each section.
[0030] Meanwhile, measurable disturbance variables such as strip running speed, strip thickness, strip width, steel grade, inlet strip temperature, and air preheating temperature are considered as feedforward compensation inputs to improve the operating condition response speed.
[0031] Step S2, Establishment of segmented dynamic model The total heat absorption rate method is used to construct a discrete dynamic model of strip temperature in each section, and the coupling effect of three-dimensional radiation heat transfer and convection heat transfer between furnace gas, furnace lining and strip is equivalent to the total heat absorption rate φ. CF This greatly simplifies complex heat transfer calculations and enables online real-time calculation of strip temperature; The discrete dynamic model of strip temperature for each section is as follows:
[0032] In the formula, T s (k) represents the strip temperature of this section at the k-th sampling time; φ CF (k) represents the total heat absorption rate of this section at the k-th sampling time, which comprehensively reflects the time-varying characteristics of furnace gas emissivity, total radiation exchange area, and convective heat transfer coefficient; u(k) represents the control input at the k-th sampling time (furnace gas temperature setpoint in the heating control area and cooling medium flow rate setpoint in the cooling control area); d(k) represents the measurable disturbance vector, which includes parameters such as strip speed, thickness, and inlet temperature.
[0033] In this embodiment, the sampling period is matched with the scanning period of the on-site PLC and is set to be adjustable from 100ms to 500ms to ensure real-time control.
[0034] Step S3, UKF Joint State Estimation A nonlinear state-space model is constructed for the two core state variables of strip temperature and total heat absorption rate. Unscented Kalman filtering is used to achieve joint state estimation, and temperature measurement noise filtering and total heat absorption rate identification are completed simultaneously to correct model parameter mismatch in real time.
[0035] Define the state vector:
[0036] in
[0037] Define the observation vector:
[0038] In the formula, g is the nonlinear state equation function; h is the nonlinear observation equation function; W and V are the current noise of the state equation and observation equation, respectively; Z is the measured value of the strip temperature at the outlet of each temperature control section, which is obtained by measuring through a preset radiation pyrometer.
[0039] The UKF joint state estimation specifically includes the following steps: S31: Set initial state values, process noise covariance, and measurement noise covariance; S32: Employ a symmetric sampling strategy to generate (2n+1) Sigma points (n=2) based on the current state mean and covariance. S33: Substitute all Sigma points into the state equation, calculate the Sigma point set after one prediction step, and sum them by weight to obtain the state prediction mean and prediction error covariance. S34: Substitute the predicted Sigma points into the observation equation to obtain the set of observed and predicted Sigma points, and calculate the weighted mean of the observed predictions, the observed and predicted covariance, and the cross-covariance between the state and the observations. S35: Calculate the Kalman gain based on the observed-predicted covariance and cross-covariance, and update the state estimate and error covariance matrix by combining the deviation between the measured temperature and the observed-predicted value at the current moment. S36: Output the optimal estimated value of the strip temperature and the total heat absorption rate identification value at the current moment, and enter the next sampling cycle.
[0040] Step S4, MPC Scroll Optimization With the goal of achieving optimal temperature tracking accuracy at the strip outlet in each section and stable change in control quantity, a zoned model predictive controller is constructed, and rolling optimization is performed based on the optimal state estimate output by the UKF.
[0041] Define the MPC objective function:
[0042] In the formula, T s,ref Np represents the target temperature value of the strip exit in each temperature control zone, Nc represents the prediction time domain, Nc represents the control time domain, and Δu represents the control increment coefficient, which is used to maintain stability. Define the constraints: The processes are set up separately for the characteristics of the two zones, and online adjustments are supported: 1) Heating control zone: upper and lower limits of furnace temperature and upper limit of strip temperature rise rate in each temperature control zone; 2) Cooling control zone: upper and lower limits of cooling medium flow rate and upper limit of strip temperature drop rate in each temperature control zone.
[0043] Step S5, Partition control execution The execution of partition control specifically includes the following steps: S51: Heating control zone: MPC outputs the optimal furnace temperature for each temperature control zone; S52: Cooling Control Zone: MPC outputs the optimal cooling medium flow rate for each temperature control zone; S53: Within each control cycle, using the current state estimated by UKF as the initial value, predict the future N based on the piecewise dynamic model. p The temperature trajectory of the strip steel is determined step by step. Within the constraints, a control increment sequence that minimizes the objective function is obtained through quadratic programming. Only the first control increment of the sequence is applied to the system to achieve rolling optimization.
[0044] Meanwhile, measurable disturbances such as strip speed, specifications, and steel type are introduced into the prediction model in advance to achieve disturbance feedforward compensation and improve the response speed when switching operating conditions.
[0045] Example 2: This example uses a continuous annealing furnace as the application object. The implementation object is: a 150m stainless steel continuous annealing furnace, steel grade 304, strip thickness 2.5mm, strip speed 55m / min; process requirements: heating zone outlet 1050℃, air cooling section 450℃, mist cooling section 120℃, water cooling section ≤40℃.
[0046] The implementation steps are as follows: 1. Divide the control zones: The heating control zone includes the preheating section and the heating section; the cooling control zone includes the air-cooling section, the mist-cooling section, and the water-cooling section. 2. Establish a segmented model: Use the summation coefficient method to establish a dynamic temperature model for the strip steel; 3. UKF Estimation: Configure UKF parameters. The initial state values are taken as the inlet strip temperature and the initial total thermal absorptivity. The process noise covariance Q and the observation noise covariance R are statistically calibrated based on the field temperature measurement data. A state update is completed every 200ms, and the filtered strip temperature and the identified φ are output synchronously. CF ; 4. MPC Parameters: Configure MPC parameters, prediction time domain Np=20, control time domain Nc=5; set constraints according to the process specifications, upper limit of furnace temperature in heating section 1250℃, single cycle temperature rise not exceeding 15℃; solve the quadratic programming problem once every 200ms and output the optimal control increment.
[0047] 5. Closed-loop control: Control commands are sent to the field PLC's bottom-level loop, and measured data is collected and updated iteratively in a 1-second cycle to update estimates and optimize output.
[0048] Implementation results: Field operation verification showed that after adopting this method, the temperature error of strip steel exit in each zone was ≤±1.2℃, the fluctuation was significantly reduced, and the consistency of product mechanical properties was improved.
[0049] Example 3: Based on the same inventive concept, this example also provides an online process control system for a stainless steel annealing furnace, employing the UKF-MPC-based online process control method for stainless steel annealing furnaces as described above. Figure 2 As shown, it mainly includes: 1) Zoned data acquisition module: used for real-time acquisition of furnace temperature, cooling medium flow rate, strip speed, and steel grade parameters; 2) Online identification module for total coefficient: used to calculate and update the total heat absorption rate φ of each temperature control zone in real time based on the heat balance formula. CF ; 3) Segmented UKF State Estimation Module: Used to output the estimated strip temperature and the identified total heat absorption rate for each temperature control zone; 4) Segmented MPC optimization module: used to output the optimal furnace temperature for each temperature control segment in the heating control zone and the optimal cooling medium flow rate for each temperature control segment in the cooling control zone; 5) Execution drive module: used to adjust the system heating valve group and cooling valve group according to the optimal furnace temperature and optimal cooling medium flow rate; 6) Setting storage module: used to store and call the target outlet temperature and process library parameters of each temperature control zone.
[0050] Furthermore, all parts of this application that are not described in detail are the same as or implemented using existing technology.
[0051] In summary: 1. This invention constructs a piecewise dynamic model using the total heat absorption rate method and combines it with unscented Kalman filtering to achieve joint estimation of strip temperature and total heat absorption rate. On the one hand, it can effectively filter out the measurement noise of the field pyrometer and improve the reliability of the temperature feedback signal; on the other hand, it can identify time-varying thermal parameters online and correct model deviations in real time, significantly enhancing the adaptability of the control model to disturbances under multiple operating conditions of the annealing furnace.
[0052] 2. This invention adopts a partitioned model predictive control architecture, designing control objectives and constraints for the heating and cooling zones respectively. With the strip outlet temperature as the core objective, it automatically coordinates the control quantities of each segment through rolling optimization, which can effectively overcome the inherent characteristics of large lag and strong coupling of the annealing furnace, achieve low fluctuation and no overshoot control of the strip temperature throughout the process, greatly improve the uniformity of strip temperature, and significantly improve the product qualification rate.
[0053] 3. This invention can be directly deployed and implemented on the basis of existing PLC / DCS control systems without the need for adding a large number of new hardware detection and execution equipment. The transformation cost is low and the project implementation is strong. At the same time, by reducing ineffective heating and excessive cooling through precise temperature control, it effectively reduces fuel consumption and cooling medium usage, and has both good economic benefits and energy-saving value.
[0054] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0055] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0056] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for online process control of a stainless steel annealing furnace based on UKF-MPC, characterized in that, Includes the following steps: S1, Section division and variable definition: The stainless steel continuous annealing furnace is divided into multiple temperature control sections along the strip running direction, and then integrated into heating control area and cooling control area according to the heating and cooling process, and the control quantity and observation quantity of each temperature control section are defined respectively. S2, Establishment of segmented dynamic model: The total heat absorption rate method is used to construct a discrete dynamic model of strip temperature in each temperature control zone. The radiation-convection coupled heat transfer in the furnace is equivalent to the total heat absorption rate parameter, forming a nonlinear state-space model suitable for online calculation. S3, UKF Joint State Estimation: Using strip temperature and total heat absorption rate as joint state variables, the state variables are iteratively estimated based on unscented Kalman filtering, while filtering out observation noise and identifying the total heat absorption rate parameter online, and outputting the optimal state estimate. S4, MPC Rolling Optimization: Based on the optimal state estimate output by UKF, constrained model predictive controllers are constructed for the heating control zone and the cooling control zone respectively, and the optimal control sequence is solved with temperature tracking accuracy and control stability as objectives. S5, Partitioned Closed-Loop Execution: The control quantities solved by MPC are sent down to the underlying actuators, and the field measured data are collected according to the control cycle to update the UKF state and MPC prediction conditions, so as to realize the full-process online rolling closed-loop control.
2. The online process control method for stainless steel annealing furnace based on UKF-MPC according to claim 1, characterized in that... In step S1, the temperature control section includes a preheating section, a heating section, an air-cooling section, a mist-cooling section, and a water-cooling section; The heating control zone includes a preheating section and a heating section. Its controlled quantity is the furnace temperature of each temperature control section; its observed quantity is the strip temperature at the outlet of each section. The cooling control zone includes an air-cooled section, a mist-cooled section, and a water-cooled section. Its control quantity is the flow rate of the cooling medium in each temperature-controlled section; its observation quantity is the outlet strip temperature of each section.
3. The online process control method for stainless steel annealing furnace based on UKF-MPC according to claim 1, characterized in that... In step S2, the discrete dynamic model of the strip temperature is as follows: In the formula, T s (k) represents the strip temperature of this section at the k-th sampling time; φ CF (k) represents the total heat absorption rate of this section at the k-th sampling time; u(k) represents the control input at the k-th sampling time (the furnace gas temperature setpoint for the heating control zone and the cooling medium flow rate setpoint for the cooling control zone); d(k) represents the measurable disturbance vector.
4. The online process control method for stainless steel annealing furnace based on UKF-MPC according to claim 3, characterized in that, The measurable disturbance vectors include: strip running speed, strip thickness, strip width, steel grade, inlet strip temperature, and air preheating temperature.
5. The online process control method for stainless steel annealing furnace based on UKF-MPC according to claim 1, characterized in that... In step S3, the state vector is defined as follows: in Define the observation vector: In the formula, g is the nonlinear state equation function; h is the nonlinear observation equation function; W and V are the current noise of the state equation and observation equation, respectively; Z is the measured value of the strip temperature at the outlet of each temperature control section, which is obtained by measuring through a preset radiation pyrometer.
6. The online process control method for stainless steel annealing furnace based on UKF-MPC according to claim 1, characterized in that... In step S3, the joint estimation of UKF states specifically includes: S31: Set initial state values, process noise covariance, and measurement noise covariance; S32: Employ a symmetric sampling strategy to generate (2n+1) Sigma points (n=2) based on the current state mean and covariance. S33: Substitute all Sigma points into the state equation, calculate the Sigma point set after one prediction step, and sum them by weight to obtain the state prediction mean and prediction error covariance. S34: Substitute the predicted Sigma points into the observation equation to obtain the set of observed and predicted Sigma points, and calculate the weighted mean of the observed predictions, the observed and predicted covariance, and the cross-covariance between the state and the observations. S35: Calculate the Kalman gain based on the observed-predicted covariance and cross-covariance, and update the state estimate and error covariance matrix by combining the deviation between the measured temperature and the observed-predicted value at the current moment. S36: Output the optimal estimated value of the strip temperature and the total heat absorption rate identification value at the current moment, and enter the next sampling cycle.
7. The online process control method for stainless steel annealing furnace based on UKF-MPC according to claim 1, characterized in that... In step S4: Define the MPC objective function: In the formula, T s,ref Np represents the target temperature value of the strip exit in each temperature control zone, Nc represents the prediction time domain, Nc represents the control time domain, and Δu represents the control increment coefficient used to maintain stability.
8. The online process control method for stainless steel annealing furnace based on UKF-MPC according to claim 7, characterized in that... In step S4, the constraints include: 1) Heating control zone: upper and lower limits of furnace temperature and upper limit of strip temperature rise rate in each temperature control zone; 2) Cooling control zone: upper and lower limits of cooling medium flow rate and upper limit of strip temperature drop rate in each temperature control zone.
9. The online process control method for stainless steel annealing furnace based on UKF-MPC according to claim 1, characterized in that... In step S5, within each control cycle, using the current state estimated by UKF as the initial value, the future N are predicted based on the piecewise dynamic model. p The temperature trajectory of the strip steel is determined step by step. Within the constraints, a control increment sequence that minimizes the objective function is obtained through quadratic programming. Only the first control increment of the sequence is applied to the system to achieve rolling optimization.
10. An online process control system for a stainless steel annealing furnace, employing the UKF-MPC-based online process control method for stainless steel annealing furnaces as described in any one of claims 1 to 9, characterized in that, include: 1) Zoned data acquisition module: used for real-time acquisition of furnace temperature, cooling medium flow rate, strip speed, and steel grade parameters; 2) Online identification module for total coefficient: used to calculate and update the total heat absorption rate of each temperature control zone in real time according to the heat balance formula; 3) Segmented UKF State Estimation Module: Used to output the estimated strip temperature and the identified total heat absorption rate for each temperature control zone; 4) Segmented MPC optimization module: used to output the optimal furnace temperature for each temperature control segment in the heating control zone and the optimal cooling medium flow rate for each temperature control segment in the cooling control zone; 5) Execution drive module: used to adjust the system heating valve group and cooling valve group according to the optimal furnace temperature and optimal cooling medium flow rate; 6) Setting storage module: used to store and call the target outlet temperature and process library parameters of each temperature control zone.