A method for hot rolling strip shape and thickness coordinated control

Abstract

The application provides a hot-rolled coil shape and thickness cooperative control method, relates to the technical field of hot continuous rolling automation control, and aims at the problems of control fragmentation between finish rolling, laminar cooling and coiling, implicit plate shape defect lagging, mutual interference of thickness control and plate shape control, and insufficient terminal quality prediction, establishes a material tracking model, a wide mode reconstruction model, a unified state vector, a terminal defect risk function, a quality debt inversion distribution model, a thickness main control and zero space plate shape compensation model, a zero mean difference cooling model and an online identification updating mechanism; through full-process cooperative control on thickness, wedge shape, crown shape, high-order plate shape mode, residual stress, temperature difference and organization state, the hot-rolled coil specification hit rate, the plate shape qualified rate, the cooling stability and the coiling stability are improved, and the method is suitable for online deployment, parameter setting, coil level switching and continuous operation of a hot continuous rolling production line.

Description

Technical Field

[0001] This invention relates to the field of automated control technology for hot rolling mills, specifically a method for coordinated control of hot-rolled coil shape and thickness. Background Technology

[0002] Existing hot-rolled strip steel production lines typically arrange average thickness control, crown shape control, strip shape control, laminar flow cooling control, and coiling control as multiple independent control loops. Average thickness control revolves around stabilizing the flow rate per second and adjusting the roll reduction; strip shape control revolves around adjusting the bending rolls, shifting rolls, and a small amount of tension; and laminar flow cooling control revolves around the target final cooling temperature. This type of control structure can maintain basic production under steady-state conditions, but under conditions of thin gauge, wide strip steel, high-speed rolling, and steel grade switching, the coupling effect between thickness control and strip shape control, the transmission effect between the finishing rolling process and the cooling process, and the cumulative effect of latent defects cannot be simultaneously covered, making it difficult to improve the product specification hit rate and strip shape qualification rate at the same time.

[0003] The root cause of the above problems lies in the following: Roll reduction correction alters rolling force, roll elastic flattening, and lateral metal flow, thus changing the width-to-thickness mode; roll bending and shifting corrections change the lateral elongation distribution and stand rolling force distribution, thereby negatively impacting average thickness and wedge shape; the lateral temperature difference and phase transformation rate difference generated by laminar cooling further alter residual stress and microstructure, ultimately manifesting as edge waviness, center waviness, side waviness, and dimensional deviations after the coiling tension is released. Existing control structures lack a unified expression of the coupling mechanism of the entire process, making it difficult to achieve stable control of the final quality. Summary of the Invention

[0004] Technical problems to be solved

[0005] To address the shortcomings of existing technologies, this invention provides a method for coordinated control of hot-rolled coil shape and thickness, which solves the problems existing in the control of hot-rolled coils, such as the separation of thickness control and shape control, the disconnect between cooling process and final quality control, the inability to identify latent defects, the repeated action of the controller on the same actuator, and the insufficient depth of disclosure to support engineering reproduction.

[0006] Technical solution

[0007] To achieve the above objectives, the present invention provides a method for the coordinated control of shape and thickness of hot-rolled coils, comprising the following steps:

[0008] Establish a material tracking model for the hot continuous rolling production line, mapping the finishing mill inlet data, process data of each stand, finishing mill outlet profile and shape data, laminar cooling zone data, and pre-coiling data to the same longitudinal material unit;

[0009] The thickness field, temperature field, and longitudinal stress field of the longitudinal material unit are modally expanded to form a unified state vector;

[0010] A cross-section state propagation model is established based on the unified state vector between the finishing rolling section, the laminar cooling section, and the coiling section;

[0011] Construct a terminal defect risk function based on the pre-coupling state and calculate the mass debt of the current longitudinal material unit;

[0012] The mass debt is inverted and allocated based on the digestion capacity of the subsequent racks and subsequent cooling zones; based on the allocation results, the thickness master control solution and zero space plate shape compensation solution are performed on the current rack, and the zero mean difference cooling solution is performed on the current cooling zone.

[0013] Based on the online measurement results, the state estimation parameters, sensitivity parameters, and model parameters are updated, thereby achieving coordinated control of hot-rolled coil thickness, shape, residual stress, and final coiling quality.

[0014] Preferably, the modal expansion is constructed using broadband normalized coordinates and broadband orthogonal basis functions. The thickness field consists of average thickness, first-order wedge mode, second-order crown mode, and fourth-order higher-order thickness mode. The temperature field consists of average temperature, first-order temperature difference mode, and second-order temperature difference mode. The broadband longitudinal stress field consists of average longitudinal stress, first-order side wave stress mode, second-order mid-wave stress mode, and fourth-order edge wave stress mode.

[0015] Preferably, the unified state vector consists of average thickness, wedge mode coefficient, crown mode coefficient, higher-order thickness mode coefficient, side wave stress mode coefficient, middle wave stress mode coefficient, edge wave stress mode coefficient, first-order temperature difference mode coefficient, second-order temperature difference mode coefficient, roll system thermal crown state, roll system wear stiffness comprehensive state, plastic memory state, and microstructure phase variable. The plastic memory state is used to record the historical effects of wide-direction plastic inhomogeneity formed by the previous stand but not yet manifested.

[0016] Preferably, the plastic memory state is updated based on the weighted result of the previous frame plastic memory state and the current frame wide-axis plastic strain non-uniformity mode. The wide-axis plastic strain non-uniformity mode is obtained by projecting the result of the logarithmic compression strain field formed by the current frame inlet thickness field and outlet thickness field, minus the average compression strain, onto the wide-axis orthogonal basis function. At the same time, a modal history memory matrix is ​​established and the potential defect strength index is updated based on the current thickness mode, stress mode, and temperature difference mode.

[0017] Preferably, the terminal defect risk function is composed of the average thickness deviation before winding, wedge deviation, crown deviation, higher-order profile deviation, side wave stress mode deviation, middle wave stress mode deviation, edge wave stress mode deviation, temperature difference mode deviation, thermal convexity deviation, microstructure deviation, and the excess of residual strain relative to critical buckling strain. The quality debt is obtained by the difference between the predicted terminal defect risk and the preset terminal risk upper limit.

[0018] Preferably, the quality debt inversion allocation is performed using a constrained optimization method. The constraints include: the sum of the rack quality debt share allocated to each subsequent rack and the cooling quality debt share allocated to each subsequent cooling zone is the current quality debt; the quality debt share of each rack does not exceed the maximum capacity of the corresponding rack; the cooling quality debt share does not exceed the maximum capacity of the corresponding cooling zone; and the objective function is composed of the sum of the cost borne by each rack and the cost borne by each cooling zone.

[0019] Preferably, the current frame control solution includes: constructing a thickness main target sub-vector and a shape cooperative sub-vector based on the current frame output vector; solving for the thickness main control quantity based on the current frame local linearization sensitivity matrix; constructing a null space projection matrix based on the thickness main target sensitivity matrix; solving for the shape compensation control quantity in the control degrees of freedom defined by the null space projection matrix; and superimposing the thickness main control quantity and the shape compensation control quantity to form a pressing correction quantity, a speed correction quantity, a tension correction quantity, a bending roll correction quantity, and a shifting roll correction quantity.

[0020] Preferably, the cooling zone control adopts the zero-mean differential cooling principle, which means that the sum of the current cooling zone width spray flow correction and the upper and lower surface spray flow correction is zero; the cooling zone optimization target consists of terminal stress mode deviation and terminal tissue state deviation, and the constraints include the upper and lower limits of zone flow, zone flow change rate constraint, target winding temperature constraint, and target tissue state constraint.

[0021] Preferably, the online update is jointly performed by a state estimator and a parameter identifier. The state estimator predicts the unified state vector based on the cross-segment state propagation model and corrects it based on the thickness measurement, plate shape measurement, temperature measurement, cooling flow feedback value, and pre-winding measurement value. The parameter identifier updates the rack sensitivity parameter, cooling sensitivity parameter, and thermal convexity propagation parameter based on the correspondence between the current control variable change and the output variable change.

[0022] Preferably, the method divides the entire roll into a head section, a middle section, and a tail section based on its longitudinal position. The head section performs thickness locking and tension establishment control, the middle section performs full-volume collaborative control driven by terminal risk, and the tail section performs control quantity decay and stable winding control. When there is a measurement anomaly, actuator saturation, or model prediction error exceeding the limit, the system switches to a degradation control mode. The degradation control mode includes freezing zero-space plate shape compensation, freezing differential cooling optimization, and maintaining thickness master control.

[0023] Beneficial effects

[0024] This invention provides a method for the coordinated control of shape and thickness of hot-rolled coils. It has the following beneficial effects:

[0025] 1. This invention uses a combination of "material tracking module + contour reconstruction module + unified state vector" to map thickness, temperature, stress, thermal convexity, plastic history and microstructure to the same longitudinal material unit. Therefore, asynchronous data from the rack inlet, rack outlet, cooling section and before winding can be superimposed on the same control object, eliminating the control mismatch problem caused by different measuring points corresponding to different material units in the existing system.

[0026] 2. This invention uses a combination of "terminal defect risk function + quality debt inversion allocation model" to quantify the defect manifestation risk of the current state before future winding into a unified value, and allocates the risk to the rack or cooling zone with the highest bearing capacity. Therefore, the system no longer relies on the current rack for passive correction, but performs active coordination across racks and cooling sections, which significantly reduces thickness fluctuations and plate shape rebound caused by overcompensation of a single rack.

[0027] 3. This invention adopts a layered solution scheme of "thickness master control solution + zero space plate shape compensation solution". First, the average thickness and wedge shape are locked. Then, crown shape, higher order mode and stress mode compensation are performed in the degree of freedom without affecting the thickness master objective. Therefore, there is no mutually canceling path between thickness control and plate shape control. The specification hit rate and plate shape qualification rate can be improved simultaneously.

[0028] 4. This invention uses a cooling control scheme of "zero mean differential cooling + terminal stress and microstructure constraint" to change the distribution of wide-direction temperature difference and phase transformation rate while maintaining the average coiling temperature and average microstructure state to meet the process objectives. Therefore, the laminar cooling section is transformed from a simple temperature control section into a residual stress reshaping section. The risks of edge waves, middle waves and side waves that are not fully digested in the finishing rolling stage can continue to be reduced in the cooling section.

[0029] 5. This invention uses a combination of "online state estimation + parameter identification + abnormal degradation control" to automatically update model parameters, sensitivity parameters and noise covariance as steel grade changes, roll temperature changes, wear accumulation and cooling fluctuations occur. At the same time, it switches to degradation control when measurement is abnormal, actuator saturation and model mismatch occur. Therefore, this solution has the robustness and maintainability required for long-term field operation. Attached Figure Description

[0030] Figure 1 This is a diagram of the overall system architecture of the present invention;

[0031] Figure 2 This is a flowchart illustrating the data synchronization and state reconstruction process of the present invention.

[0032] Figure 3 This is a flowchart of the frame-side thickness main control and zero-space plate shape compensation of the present invention;

[0033] Figure 4 This is a flowchart of the cooling-side zero-mean differential control of the present invention;

[0034] Figure 5 This is a flowchart of the closed-loop state estimation and online identification process of the present invention;

[0035] Figure 6 This is a flowchart of the production line deployment and timing control of the present invention. Detailed Implementation

[0036] 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 some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Specific Implementation Example 1:

[0038] Reference Figures 1 to 6A method for coordinated control of hot-rolled coil shape and thickness is proposed. On the production line side, a thickness profile measurement device, a shape measurement device, a temperature measurement device, a rolling force measurement device, a speed measurement device, a tension measurement device, a bending roll and shifting roll feedback device, and a laminar cooling flow feedback device are arranged. On the control side, a material tracking module, a profile reconstruction module, a state estimation module, a terminal risk prediction module, a quality debt allocation module, a stand solving module, a cooling solving module, an execution management module, and a model update module are set up. A unified state vector describes the geometric state, stress state, thermal state, historical state, and microstructure of the current longitudinal material unit. A terminal defect risk function maps the current state to the pre-coiling terminal quality risk. A quality debt inversion allocation model allocates the terminal risk correction task to the most suitable stand and cooling zone. Thickness master control and zero-space shape compensation form the stand execution quantity, and zero-mean difference cooling forms the cooling execution quantity. Online state estimation and parameter identification ensure that the model is updated in real time according to changes in steel grade, specification, roll temperature, and cooling conditions.

[0039] The technical logic of this invention comprises five layers. The first layer is the data synchronization layer, responsible for mapping thickness, temperature, shape, tension, speed, rolling force, cooling flow rate, and historical process data to the same longitudinal material unit. The second layer is the state reconstruction layer, responsible for converting transverse thickness distribution, temperature distribution, and stress distribution into finite-dimensional modes. The third layer is the risk prediction layer, responsible for predicting terminal defect risks before coiling based on the states of the finishing and cooling ends. The fourth layer is the task allocation layer, responsible for decomposing tasks based on the quality debt and equipment capacity. The fifth layer is the execution layer, responsible for issuing deterministic setpoints to the stand actuators and cooling zone actuators.

[0040] The material tracking module receives inlet velocity, speed of each rack, rack geometric distance, and cooling section length information, and outputs unified longitudinal material coordinates; the contour reconstruction module receives transverse thickness contour, plate shape segment value, and width temperature value, and outputs thickness mode, stress mode, and temperature difference mode; the state estimation module receives unified state equation, measurement vector, and control input, and outputs unified state estimate; the terminal risk prediction module receives the final rack state, cooling strategy, and material parameters, and outputs the terminal risk before winding; the quality debt allocation module receives terminal risk, risk limit, and the capacity of each piece of equipment, and outputs the quality debt share of each rack and each cooling zone; the rack solving module outputs the increments of pressing down, speed, tension, bending roll, and skew roll; the cooling solving module outputs the increments of spray flow rate for each zone; the execution management module completes the limiting, speed limiting, and fault switching; and the model update module completes parameter identification, confidence update, and model write-back.

[0041] The core algorithms include an end-defect risk prediction algorithm, a quality debt inversion and allocation algorithm, a thickness master control and zero-space shape compensation algorithm, a zero-mean difference cooling algorithm, and an online state estimation algorithm. The end-defect risk prediction algorithm maps the end-of-rolling state and cooling strategy to explicit defect risks before coiling; the quality debt inversion and allocation algorithm assigns risk mitigation tasks to subsequent stands and cooling zones; the thickness master control and zero-space shape compensation algorithm completes shape compensation while ensuring the convergence of the main thickness target; the zero-mean difference cooling algorithm redistributes residual stress while maintaining the average coiling temperature and average microstructure without deviating from the target; and the online state estimation algorithm fuses measured data with model predictions to form a high-confidence state vector.

[0042] Data sources include both real-time production data and offline process data. Real-time production data comes from the L1 automation system, thickness profiler, shape gauge, infrared temperature measurement device, rolling force sensor, speed encoder, tension measurement device, and feedback from cooling valve group. Offline process data comes from the L2 process database, steel grade material parameter library, historical coil quality database, buckling threshold calibration library, and cooling phase transformation database. Offline data is used to initialize the state propagation matrix, sensitivity matrix, risk weight matrix, and phase transformation parameters; online data is used for real-time state estimation, online correction, and fault identification. The data usage process is as follows: First, the planned specifications and steel grade are read; then, yield strength, elastic modulus, Poisson's ratio, phase transformation characteristic parameters, and target temperature range are extracted from the material parameter library; then, the real-time sensor data is mapped to the measurement vector of the current longitudinal material element; next, the state estimation results are used to update the terminal risk prediction and quality debt allocation; finally, the optimization results are written back to the L1 execution layer.

[0043] Table 1 Key data items, data sources, sampling periods, and uses.

[0044] Data Items Data source Typical sampling period Usage process Inlet thickness, inlet width, inlet temperature Precision rolling mill inlet thickness gauge, width gauge, and pyrometer 30 ms Used to initialize the current volume's entry state and breadth boundaries. Rolling force, pressing position, speed, tension Hydraulic systems of each frame, encoders, and tension measurement circuits 10 ms Used for rack-side local linearization and material tracking Lateral thickness profile, plate shape measurement values Finishing mill exit profile gauge and plate shape gauge 30 ms Used for thickness mode and stress mode reconstruction Zoned temperature, flow rate, valve position feedback Laminar flow cooling thermometer, flow meter, valve assembly feedback 150 ms Used for cooling-side state updates and differential control Steel grade parameters, yield strength, phase transformation curve, historical roll data L2 process database and material parameter library Load once per volume Used for initial model invocation and online identification priors

[0045] The typical sampling periods in Table 1 are used to illustrate the implementation depth of this invention. The inlet thickness and inlet temperature are obtained from the finishing mill inlet measuring device, and the sampling period is [missing information]. This is used to initialize the current longitudinal material unit; rolling force, reduction position, and data on bent and shifted rolls are obtained from the process measurement devices of each stand, with a sampling period of [missing information]. This is used for constructing rack state propagation and sensitivity updates; rack speed and inter-rack tension are derived from a speed encoder and a tension measuring device, with a sampling period of [missing information]. Used for material tracking and control constraints; the transverse thickness profile is obtained from the finishing mill exit profiler, with a sampling period of [missing information]. This is used to reconstruct the average thickness, wedge shape, crown shape, and higher-order thickness modes; the plate shape segment values ​​are obtained from the plate shape analyzer at the finishing mill exit, with a sampling period of [missing information]. It is used to reconstruct the stress modes of side waves, mid-waves, and edge waves; the broadband temperature is obtained from an infrared thermometer with a sampling period of [missing information]. Used to reconstruct the temperature difference mode; the cooling zone flow rate comes from the valve group feedback device, with a sampling period of [missing information]. Used for cooling solution and online identification; pre-winding temperature and winding tension are obtained from the pre-winding measuring device, with a sampling period of [missing information]. The parameters are used for terminal risk verification and model correction; steel material parameters are from the material parameter library and updated by volume, used to calculate critical buckling strain and phase transformation propagation parameters; historical volume quality data are from the historical database and updated by volume, used to initialize the weight matrix and threshold.

[0046] When deploying the controller, the device-level closed-loop sampling period is set to... The rack-side collaborative optimization sampling period is set to The cooling side optimized sampling period is set to The model update cycle is set to This cycle configuration corresponds to the typical computing power of the field PLC, L1 controller, and edge computing server, and is matched with the dynamic time constants of thickness, plate shape, and cooling process.

[0047] It is based on a seven-stand finishing mill, twelve laminar flow cooling zones and a coiling system.

[0048] The material tracking module first establishes a mapping relationship from the time domain to the length domain. The vertical material coordinates are calculated using the following formula:

[0049]

[0050] Indicates time The material coordinates of the corresponding material unit along the length of the strip, in meters; Indicates the current time; This indicates the reference time when the coil head enters the finishing mill. This indicates a reference speed, measured in meters per second. Represents the integral variable; This represents the time derivative. The purpose of this formula is to map measurement data collected at different times to the same longitudinal material element, avoiding inconsistencies in the physical objects corresponding to different measurement points.

[0051] In the width direction, establish a width-normalized coordinate system:

[0052]

[0053] Represents normalized width coordinates; This represents the physical coordinates of the strip width, in meters. Indicates vertical position The actual width of the strip at that location is in meters. The purpose of this formula is to map strips of different widths to the same interval, establishing a common independent variable for the unified modal function.

[0054] The contour reconstruction module performs modal expansion on the thickness field, temperature field, and longitudinal stress field in the broad direction, with the following expression:

[0055]

[0056]

[0057]

[0058] Indicates the first rack exit at location Thickness distribution at a given location, in millimeters; Indicates the average thickness; Indicates the first Thickness modal coefficients; This indicates the temperature distribution, with the unit being degrees Celsius. Indicates average temperature; Indicates the first First-order temperature modal coefficients; This indicates the longitudinal stress distribution along the width, with units of megapascals (MPa). Indicates the average longitudinal stress; Indicates the first First-order stress modal coefficients; Represents wide-axis orthogonal basis functions; Indicates the rack number; This indicates the modal order. The three sets of expansions serve to extract the main controlling components of thickness profile, temperature difference distribution, and plate stress, respectively, thereby reducing the control dimensionality.

[0059] This implementation uses the following basis functions:

[0060]

[0061]

[0062]

[0063] This represents a first-order odd-mode function, corresponding to a wedge shape; This represents a second-order even-mode function, corresponding to a crown shape; This represents the fourth-order even-mode function, corresponding to the edge wave and the middle wave precursor. The reason for using this set of three basis functions is that the first-order mode directly affects the thickness difference between the left and right sides, the second-order mode directly affects the convexity of the middle section, and the fourth-order mode directly affects the instability tendency of the edge and the middle section. These three factors already cover the main quality characteristics of hot-rolled coils.

[0064] For discrete thickness sampling data, the average thickness and thickness modal coefficients are calculated using the following formula:

[0065]

[0066]

[0067] Indicates the total number of transverse thickness sampling points; Indicates the first The weight of each sampling point; Indicates the first Thickness measurement values ​​at each width sampling point; Indicates the first The normalized width coordinates corresponding to the sampling points. The first equation is used to calculate the average thickness, and the second equation is used to project the thickness deviation onto the first... On the basis functions of order 1. The same projection method is used for temperature mode and stress mode reconstruction.

[0068] The unified state vector is defined as:

[0069]

[0070] Indicates the first The unified state vector at the rack; Indicates the thermal crowning state of the roller system; This indicates the overall wear stiffness status of the roller system; Indicates a state of plastic memory; This represents the phase variable of the organization; the meanings of the other symbols are explained above. This state vector integrates geometric quantities, stress, temperature difference, thermal state quantities, historical quantities, and organizational quantities into the same state space, which is the basis for subsequent risk prediction and optimization solutions.

[0071] The state propagation relationship between racks is as follows:

[0072]

[0073] Indicates the first rack status; Represents the state propagation matrix; Represents the control input matrix; Represents the external perturbation input matrix; This represents a scheduling parameter vector, which includes steel type, specifications, temperature, speed, and friction state. Indicates the first rack control vector; Represents the external perturbation vector; This indicates an unmodeled disturbance. The formula describes how the current rack state propagates to the next rack after being affected by control and external disturbances.

[0074] The rack control vector and external disturbance vector are defined as follows:

[0075]

[0076]

[0077] Indicates the amount of correction applied; Indicates the speed correction amount; Indicates the tension correction amount; Indicates the amount of correction for the bent roller; Indicates the amount of correction for misaligned rollers; This indicates the thickness deviation of the incoming material; Indicates the deviation in incoming material temperature; Indicates fluctuation in yield strength; This represents the fluctuation of the friction coefficient. This definition provides the specific physical details of the control input and external disturbance input.

[0078] Plastic memory state is updated using the following formula:

[0079]

[0080]

[0081]

[0082] Indicates the forgetting coefficient of plastic memory; Indicates the first Weights of plastic inhomogeneous modes; Indicates the first The first rack generated Modes of inhomogeneous plastic strain in the step width direction; Indicates the first Logarithmic compression strain of the frame; This represents the average compression strain. The purpose of the above formula is to preserve the plastic non-uniformity formed in the preceding frame but not yet manifested in the state variables, so that subsequent control can address potential defects in advance.

[0083] The modal history memory matrix and the potential defect strength index are defined by the following formula:

[0084]

[0085]

[0086]

[0087] Represents the modal history memory matrix; Indicates the historical forgetting coefficient; Represents the current modality column vector; Represents the historical mode weight matrix; Represents the largest eigenvalue operator; This represents the potential defect strength index. The purpose of this set of formulas is to record the cumulative state of the current material element across the multi-stand path, used to identify latent risks not yet apparent in the current shape meter readings.

[0088] The pre-contraction terminal defect risk function is defined as follows:

[0089]

[0090]

[0091]

[0092] This indicates the risk of terminal defects before winding; This represents the terminal state vector before winding; Represents the terminal state weight matrix; Indicates the first Step buckling risk penalty coefficient; Indicates the terminal number First residual strain mode; Indicates the first Critical buckling strain; Indicates the terminal number Stress modes; Indicates the terminal equivalent elastic modulus; Indicates the average thickness before winding; Indicates Poisson's ratio; Indicates the first The first equation measures the degree to which the terminal state deviates from the target, and the second equation measures the buckling exceedance. The second equation is used to convert the stress mode into the residual strain mode. The third equation is used to calculate the critical threshold of the terminal explicit defect.

[0093] The current definition of mass debt in a material unit is:

[0094]

[0095] This indicates the mass debt of the material unit at the current rack location; This indicates the terminal defect risk predicted based on the current state; This represents the upper limit of permissible terminal risk. The function of this formula is to convert future risks exceeding the limit into a control task that must be addressed currently.

[0096] The quality debt inversion allocation model is solved using the following formula:

[0097]

[0098] satisfy

[0099]

[0100]

[0101]

[0102] Indicates assignment to the first The quality debt share of the rack; Indicates assignment to the first Quality debt share of the cooling zone; Indicates the first Rack load capacity coefficient; Indicates the first Cooling zone capacity coefficient; Indicates the first Maximum load capacity of the rack; Indicates the first Maximum load capacity of the cooling zone; Indicates the total number of racks; This represents the total number of cooling zones. The goal of this optimization problem is to allocate mass debt to device objects that have the lowest cost and sufficient capacity.

[0103] The rack load capacity coefficient and the cooling load capacity coefficient are defined as follows:

[0104]

[0105]

[0106] Indicates the first Sensitivity matrix of the rack to critical states; Indicates the first Sensitivity matrix of cooling zones to critical states; and This represents the weight matrix for ability evaluation; and Indicates the saturation penalty coefficient; Indicates the saturation level of the rack actuator; Indicates the saturation level of the cooling valve assembly; This represents the matrix trace. The purpose of this set of formulas is to dynamically characterize the actual load-bearing capacity of equipment under current operating conditions.

[0107] The current rack output vector is defined as:

[0108]

[0109]

[0110] Indicates the current rack output vector; Represents the thickness principal target sub-vector; This represents the plate shape cooperative sub-vector. The purpose of this decomposition is to separate the "average thickness and wedge shape targets that must be satisfied first" from the "plate shape targets that are compensated for without compromising the thickness target".

[0111] The locally linearized model is written as:

[0112]

[0113]

[0114] Indicates the sampling time The output increment; Represents the overall sensitivity matrix; This represents the sensitivity matrix of the main target based on thickness. Represents the sensitivity matrix of the plate-shaped cooperative target; This represents the control vector increment. This model is used to correlate actuator increments with changes in output quality.

[0115] The thickness control quantity is solved by the following formula:

[0116]

[0117] satisfy

[0118]

[0119]

[0120] Indicates the main control quantity for thickness; Indicates the deviation of the main target thickness; Represents the thickness weight matrix; Represents the action penalty matrix; and Indicates the upper and lower limits of the control vector; Indicates the upper limit of the rate of change of the control quantity; This indicates the rack-side collaborative optimization sampling period. The purpose of this optimization is to prioritize locking the average thickness and wedge shape.

[0121] Zero-space plate shape compensation is solved by the following formula:

[0122]

[0123]

[0124]

[0125]

[0126] The thickness represents the null projection matrix of the principal target. Represents the identity matrix; This represents the generalized inverse of the thickness-based principal target sensitivity matrix; Indicates the free variables for plate shape compensation; Represents the free variables for optimal plate shape compensation; Represents the plate weight matrix; Represents the penalty matrix for board-shaped actions; This represents the remaining plate shape deviation after thickness master control correction. The purpose of this set of formulas is to ensure that plate shape compensation is only performed in the degrees of freedom that do not affect the primary thickness objective.

[0127] Let the first The cooling zone control vector is defined as follows:

[0128]

[0129] , , These represent the spray flow rates on the left, middle, and right sides of the upper surface, respectively. , , These represent the spray flow rates on the left, middle, and right sides of the lower surface, respectively. This definition provides the control degrees of freedom for the cooling section.

[0130] The zero mean difference constraint is:

[0131]

[0132] Represents a column vector consisting entirely of 1s; This represents the cooling zone flow correction vector. The purpose of this constraint is to adjust the spray distribution across the width and on the upper and lower surfaces without changing the current average total flow rate of the cooling zone.

[0133] The cooling optimization problem can be written as:

[0134]

[0135] satisfy

[0136]

[0137]

[0138]

[0139] This indicates the optimal cooling flow rate correction amount; Represents the cooling sensitivity matrix; This indicates the deviation between stress and the target tissue structure. Represents the cooling target weight matrix; Represents the penalty matrix for cooling actions; and Indicates the upper and lower limits of the valve group flow rate; This indicates the predicted average winding temperature; Indicates the target winding temperature; Indicates the winding temperature tolerance; Indicates the predicted average tissue condition; Indicates the target organizational status; This indicates the tolerance for microstructure. The purpose of this optimization is to reshape the terminal stress and microstructure without compromising the average thermal regime.

[0140] The state estimation and prediction steps are performed as follows:

[0141]

[0142]

[0143] The calibration procedure is performed according to the following formula:

[0144]

[0145]

[0146]

[0147] Indicates the predicted state; Indicates the calibration status; This represents the prediction error covariance; Indicates the correction error covariance; Represents the Jacobian matrix of the state equation; Represents the process noise covariance; Indicates Kalman gain; The Jacobian matrix represents the measurement equation; Indicates the measurement noise covariance; Represents the measurement vector; This represents the measurement function. The purpose of this set of formulas is to integrate model predictions and field measurements into a high-confidence state estimate.

[0148] Parameter identification is performed using the following formula:

[0149]

[0150]

[0151]

[0152] Indicates the parameter estimate; Indicates the recognition gain; Represents the regression vector; Represents the parametric covariance matrix; This represents the forgetting factor. This set of formulas is used to update rack sensitivity, cooling sensitivity, and thermal convexity propagation parameters online.

[0153] The present invention is deployed in the following order:

[0154] The first step is to establish a steel grade material parameter library, a historical coil quality database, a cooling phase transformation database, and a buckling threshold calibration library in the L2 database;

[0155] The second step is to establish a material tracking module, a state estimation module, a terminal risk prediction module, a quality debt allocation module, a rack solving module, and a cooling solving module on the L1 side.

[0156] The third step is to define a unified message format for inlet thickness, inlet temperature, rolling force, speed, tension, profile, shape, and cooling flow rate.

[0157] The fourth step is to configure sampling tasks according to a schedule;

[0158] Step 5: Initialize using the most recent thirty volumes of historical data of the same specifications. , , , , , , and ;

[0159] Step 6: When the first roll is put online, freeze the zero-space plate shape compensation and differential cooling, and only perform the main thickness control;

[0160] Step 7: Once the first volume stabilizes, begin terminal risk prediction, quality bond allocation, and cooling-off solution.

[0161] Step 8: Enable online parameter identification and abnormal degradation control in subsequent volumes. Specific Implementation Example 2:

[0163] Based on the technical solution of Specific Embodiment 1, further practical application examples are given.

[0164] The industrial implementation parameters for this embodiment are: the number of finishing mill stands. The number of cooling zones is taken The rack-side sampling period is taken The sampling period on the cooling side is taken The model update cycle is taken The initial segment weight amplification factor is taken as follows: The plate shape weight reduction factor is taken as The starting point of the tail section attenuation is taken as The final attenuation coefficient of the tail section is taken as Under these parameters, the system first calculates the unified state vector and terminal risk using formulas, then allocates the workload to each rack and cooling zone through quality debt inversion, and subsequently generates execution increments through hierarchical optimization and sends them to the PLC and valve group controller, ultimately achieving coordinated control of thickness, plate shape, and terminal quality.

[0165] First, data alignment is achieved using material tracking and width normalization formulas. Then, thickness, temperature, and stress modes are calculated using modal expansion and modal projection formulas. Next, a unified state and implicit risk indicators are formed using state propagation, plastic memory update, and historical memory formulas. Then, terminal risk and current workload are calculated using terminal risk and quality debt formulas. Finally, the quality debt allocation model is used to break down tasks to racks and cooling zones. Then, rack-side hierarchical optimization and cooling-side optimization formulas are used to generate control commands. Finally, state estimation and parameter identification formulas are used to complete the model closed-loop update. By following the above sequence, those skilled in the art can directly implement this invention based on this specification.

[0166] In summary, this invention achieves coordinated control of the entire hot-rolled coil process through terminal risk-driven mechanisms, quality debt allocation, thickness master control and zero-space shape compensation, zero-mean differential cooling, and online state estimation and parameter identification. All equivalent substitutions made within the spirit and principles of this invention fall within its protection scope.

[0167] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a reference structure" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0168] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for coordinated control of shape and thickness of hot-rolled coil, characterized in that, Includes the following steps: Establish a material tracking model for the hot continuous rolling production line, mapping the finishing mill inlet data, process data of each stand, finishing mill outlet profile and shape data, laminar cooling zone data, and pre-coiling data to the same longitudinal material unit; The thickness field, temperature field, and longitudinal stress field of the longitudinal material unit are modally expanded to form a unified state vector; A cross-section state propagation model is established based on the unified state vector between the finishing rolling section, the laminar cooling section, and the coiling section; Construct a terminal defect risk function based on the pre-coupling state and calculate the mass debt of the current longitudinal material unit; The mass debt is inverted and allocated based on the digestion capacity of the subsequent racks and subsequent cooling zones; based on the allocation results, the thickness master control solution and zero space plate shape compensation solution are performed on the current rack, and the zero mean difference cooling solution is performed on the current cooling zone. Based on the online measurement results, the state estimation parameters, sensitivity parameters, and model parameters are updated, thereby achieving coordinated control of hot-rolled coil thickness, shape, residual stress, and final coiling quality.

2. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 1, characterized in that, The modal expansion is constructed using broadband normalized coordinates and broadband orthogonal basis functions. The thickness field consists of average thickness, first-order wedge mode, second-order crown mode, and fourth-order higher-order thickness mode. The temperature field consists of average temperature, first-order temperature difference mode, and second-order temperature difference mode. The broadband longitudinal stress field consists of average longitudinal stress, first-order side wave stress mode, second-order mid-wave stress mode, and fourth-order edge wave stress mode.

3. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 1, characterized in that, The unified state vector consists of average thickness, wedge mode coefficient, crown mode coefficient, higher-order thickness mode coefficient, side wave stress mode coefficient, middle wave stress mode coefficient, edge wave stress mode coefficient, first-order temperature difference mode coefficient, second-order temperature difference mode coefficient, roll system thermal crown state, roll system wear stiffness comprehensive state, plastic memory state, and microstructure phase variable. The plastic memory state is used to record the historical effects of wide-direction plastic inhomogeneity formed by the previous stand but not yet manifested.

4. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 3, characterized in that, The plastic memory state is updated based on the weighted result of the previous frame plastic memory state and the current frame wide-axis plastic strain non-uniformity mode. The wide-axis plastic strain non-uniformity mode is obtained by projecting the result of the logarithmic compression strain field formed by the current frame inlet thickness field and outlet thickness field, minus the average compression strain, onto the wide-axis orthogonal basis function. At the same time, a modal history memory matrix is ​​established and the potential defect strength index is updated based on the current thickness mode, stress mode, and temperature difference mode.

5. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 1, characterized in that, The terminal defect risk function is composed of the average thickness deviation before winding, wedge deviation, crown deviation, higher-order profile deviation, side wave stress mode deviation, middle wave stress mode deviation, edge wave stress mode deviation, temperature difference mode deviation, thermal convexity deviation, microstructure deviation, and the excess of residual strain relative to critical buckling strain. The quality debt is obtained by the difference between the predicted terminal defect risk and the preset terminal risk upper limit.

6. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 1, characterized in that, The quality debt inversion allocation is performed using a constrained optimization method. The constraints include: the sum of the rack quality debt share allocated to each subsequent rack and the cooling quality debt share allocated to each subsequent cooling zone is the current quality debt; the quality debt share of each rack does not exceed the maximum capacity of the corresponding rack; the cooling quality debt share does not exceed the maximum capacity of the corresponding cooling zone; and the objective function is composed of the sum of the cost borne by each rack and the cost borne by each cooling zone.

7. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 1, characterized in that, The current frame control solution includes: constructing a thickness main target sub-vector and a shape cooperative sub-vector based on the current frame output vector; solving for the thickness main control quantity based on the local linearization sensitivity matrix of the current frame; constructing a null space projection matrix based on the thickness main target sensitivity matrix; solving for the shape compensation control quantity in the control degrees of freedom defined by the null space projection matrix; and superimposing the thickness main control quantity and the shape compensation control quantity to form the pressing correction quantity, speed correction quantity, tension correction quantity, roll bending correction quantity, and roll shifting correction quantity.

8. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 1, characterized in that, The cooling zone control adopts the zero-mean differential cooling principle, which means that the sum of the current cooling zone width spray flow correction and the upper and lower surface spray flow correction is zero. The cooling zone optimization objective consists of terminal stress mode deviation and terminal tissue state deviation. The constraints include the upper and lower limits of zone flow, zone flow change rate constraint, target winding temperature constraint, and target tissue state constraint.

9. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 1, characterized in that, The online update is jointly performed by the state estimator and the parameter identifier. The state estimator predicts the unified state vector based on the cross-segment state propagation model and corrects it based on the thickness measurement, plate shape measurement, temperature measurement, cooling flow feedback value and pre-winding measurement value. The parameter identifier updates the rack sensitivity parameter, cooling sensitivity parameter and thermal convexity propagation parameter based on the correspondence between the current control variable change and the output variable change.

10. The method for coordinated control of shape and thickness of hot-rolled coil according to claim 1, characterized in that, The method divides the entire roll into a head section, a middle section, and a tail section based on its longitudinal position. The head section performs thickness locking and tension establishment control, the middle section performs full-volume collaborative control driven by terminal risk, and the tail section performs control quantity decay and stable winding control. When there is a measurement anomaly, actuator saturation, or model prediction error exceeding the limit, the system switches to a degradation control mode. The degradation control mode includes freezing zero-space plate shape compensation, freezing differential cooling optimization, and maintaining thickness master control.

Images