Computer intelligent control system of rolling mill

Abstract

The invention relates to the technical field of metal rolling automation, and discloses a computer intelligent control system of a rolling mill, which comprises a rolling mill production line, and the rolling mill production line further comprises a tensiometer, a pressure sensor and a displacement sensor; and the computer intelligent control system is electrically connected with an executing mechanism and a sensor of the rolling mill production line and comprises a tension control module, an automatic thickness control module and an online model self-adaptive module. The method comprises the following steps: controlling strip steel tension through a tension control module; the strip steel outlet thickness is adjusted through the thickness automatic control module; and model parameters such as rolling mill rigidity and friction compensation are identified in real time in the rolling process through an online model self-adaption module, and continuous optimization of the control model is achieved. By introducing the online model adaptive module, real-time online correction of the core physical model of the control system is realized, and the problem of model mismatch caused by equipment wear and working condition change is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of metal rolling automation, in particular to a computer intelligent control system of a rolling mill. BACKGROUND

[0002] A tandem cold rolling mill is a key equipment in the production of metal strips, and its core control objective is to ensure that the finished strip has accurate thickness and uniform shape. To achieve this objective, modern tandem cold rolling mills are generally equipped with an automatic gauge control (AGC) system and a tension control system.

[0003] Existing control systems usually operate based on pre-established mathematical models. For example, the tension control system relies on a torque model that includes components such as reference tension, friction compensation, and dynamic inertia compensation; the AGC system relies on a rolling physical model centered on the mill bounce equation. Key parameters in these models, such as mill stiffness and transmission system friction coefficient, are usually determined through offline calibration or empirical setting at the initial stage of system commissioning.

[0004] However, in actual production processes, these key parameters are not constant. The mechanical parts of the mill will wear out over time, the lubrication conditions will change with temperature and emulsion state, and different rolling materials and process specifications will also cause changes in rolling physical properties. These factors will cause a gradual deviation between the pre-set control model and the actual physical properties of the mill, i.e., model mismatch.

[0005] This model mismatch problem directly reduces the performance of the control system. In dynamic conditions (such as acceleration and deceleration processes), inaccurate dynamic inertia compensation and friction compensation can cause significant fluctuations in tension, which can lead to production accidents such as strip breakage or poor shape. At the same time, inaccurate model parameters such as mill stiffness can weaken the effect of feedforward compensation for thickness control, or even have a negative effect, leading to a decrease in the accuracy of the outlet thickness, increasing the dependence on the feedback control of the backend, and thus reducing the production efficiency and yield; in addition, there is a significant coupling interference effect between tension and thickness during acceleration and deceleration processes; although traditional control systems also attempt to use feedforward compensation, due to the fixed model parameters they rely on, they cannot adapt to changes in working conditions, resulting in poor decoupling effect and difficulty in fundamentally eliminating the impact of dynamic disturbances on product quality.

[0006] Therefore, the present application proposes a computer intelligent control system of a rolling mill to solve the deficiencies of the prior art. SUMMARY

[0007] In view of the deficiencies of the prior art, the present application provides a computer intelligent control system of a rolling mill, which solves the problem of the decline of dynamic control performance caused by the misalignment of control model parameters due to equipment wear, working condition changes and other factors in the control system of a cold continuous rolling mill.

[0008] To achieve the above object, the present application is implemented by the following technical solutions: a computer intelligent control system of a rolling mill, the system comprising: A rolling mill production line, which is provided with, in sequence along the strip rolling direction, an uncoiler, a straightening device, a horizontal loop clamping and feeding device, a horizontal loop, an inlet S roller, a first main frame, a second main frame, a third main frame, a fourth main frame, a fifth main frame, a thickness gauge, an outlet S roller, a pinch roller, a flying shear, a deflector roller, a first coiler and a second coiler; The rolling mill production line further comprises tension meters arranged between the main frames, pressure sensors and displacement sensors arranged on the main frames, and rotary encoders for detecting the rotational speeds of the motors; A computer intelligent control system electrically connected to the actuators and sensors in the rolling mill production line, the computer intelligent control system comprising: A tension control module for calculating and outputting motor torque instructions for driving the uncoiler and the first coiler or the second coiler based on a multi-component torque model comprising a reference tension torque, a friction compensation torque and a dynamic inertia compensation torque; A thickness automatic control module for adjusting the rolling parameters of the first main frame to the fifth main frame through a thickness automatic control algorithm based on the deviation between the set target thickness and the actual thickness detected by the thickness gauge; An online model self-adaptive module for collecting and analyzing the operation data of the rolling mill production line in real time during rolling, and performing online correction on the parameters in the multi-component torque model and the rolling physical model parameters relied on by the thickness automatic control algorithm based on a preset self-adaptive algorithm, and updating the corrected parameters to the tension control module and the thickness automatic control module for continuous optimization of the control model.

[0009] Preferably, the thickness automatic control module further comprises: A dynamic decoupling feedforward unit configured to monitor the inter-frame tension variation acting on the main frame, and predict the outlet thickness variation caused by the tension variation based on a preset rolling physical model; The dynamic decoupling feedforward unit is further configured to generate a feedforward roll gap correction amount according to the predicted outlet thickness variation, and superimpose it with a feedback roll gap correction amount generated by a thickness feedback control loop to form a total roll gap correction instruction; Thus, when the tension between the stands changes, the thickness is actively compensated, so as to eliminate the coupling interference of the tension change on the thickness control.

[0010] Preferably, in the online model adaptive module, the step of online correcting the parameters of the rolling physical model comprises: Real-time collection of the feedback roll gap correction amount output by the thickness automatic control module to eliminate thickness deviation, and the corresponding rolling force change amount detected by the pressure sensor; Taking the feedback roll gap correction amount and the rolling force change amount as inputs, the rolling mill stiffness coefficient in the rolling physical model is identified through an online parameter identification algorithm; And the identified rolling mill stiffness coefficient is updated to the thickness automatic control module.

[0011] Preferably, the identified rolling mill stiffness coefficient is also updated to the dynamic decoupling feedforward unit for use in the outlet thickness change amount prediction calculation of the dynamic decoupling feedforward unit.

[0012] Preferably, the step of online correcting the parameters in the multi-component torque model by the online model adaptive module comprises: Under the quasi-steady state working condition of the rolling process, according to the actual output torque of the motor, the reference tension torque, the taper tension correction torque, and the output value of the current two-dimensional friction compensation model, a friction residual error is calculated; and the two-dimensional friction compensation model is online corrected according to the friction residual error; the calculation formula of the friction residual error is: ; In the formula, is the friction residual error, is the actual output torque of the motor, is the reference tension torque, is the taper tension correction torque, is the friction compensation torque.

[0013] Preferably, the multi-component torque model on which the tension control module is based calculates the total output torque provided to the motor of the unwinder or the coiler, and the formula is: ; In the formula, is the reference tension torque, which is the core torque required to maintain constant tension; is the friction compensation torque, which is the torque required to compensate for the mechanical friction of the transmission system; is the dynamic inertia compensation torque, which is the torque required to compensate for the change in the rotational inertia of the steel coil when the mill train is accelerating or decelerating; is the taper tension correction torque.

[0014] Preferably, the operation data of the rolling mill production line comprises: The inter-stand tension detected by the tension meter, the exit thickness detected by the thickness gauge, the rolling force detected by the pressure sensor, the roll gap position detected by the displacement sensor, and the motor speed detected by the rotary encoder.

[0015] Preferably, the entry S roller and the exit S roller are used to establish stable tension zones before and after the main body of the rolling mill, and the tension control module controls the driving motors of the entry S roller and the exit S roller to maintain the tension in the tension zones constant.

[0016] Preferably, the computer intelligent control system is further used for: Controlling the flying shear to perform a shearing action after the first coiler is full and before switching to the second coiler, for uninterrupted rolling.

[0017] The application also provides a computer intelligent control method for a rolling mill, comprising the following steps: S1. Based on a multi-component torque model comprising a reference tension torque, a friction compensation torque, and a dynamic inertia compensation torque, calculating and outputting a motor torque instruction for driving the uncoiler and the coiler; S2. Based on the deviation between the set target thickness and the actual thickness detected by the thickness gauge, adjusting the rolling parameters of the main stand through a thickness automatic control algorithm; S3. In the rolling process, real-time acquisition and analysis of the operation data of the rolling mill production line, and based on a pre-set adaptive algorithm, online correction of the parameters in the multi-component torque model and the rolling physical model parameters relied on by the thickness automatic control algorithm; S4. Updating the parameters corrected in the online adaptive step to the models relied on by the tension control step and the thickness control step, for continuous optimization of the control model.

[0018] The application provides a computer intelligent control system for a rolling mill, which has the following beneficial effects: 1. The online model adaptive module of the application can continuously correct the parameters of the multi-component torque model relied on by the tension control and the rolling physical model relied on by the thickness control through real-time acquisition of the operation data; this design overcomes the problem of precision decline of traditional static models due to equipment wear, temperature changes, or material differences, ensures that the control model always matches the actual working condition throughout the production cycle, and thus improves the long-term control precision of the exit thickness and the strip tension.

[0019] 2、The thickness automatic control module of the application is provided with a dynamic decoupling feedforward unit. When the tension control module is adjusted, the unit can calculate the coupling influence on the thickness in advance based on the rolling physical model, and generate a compensation signal to actively inject the thickness control loop; compared with the traditional feedback control which depends on the thickness deviation generated and then adjusts, this feedforward mechanism can almost instantaneously offset the tension disturbance, greatly improving the response speed and stability of the system under dynamic working conditions such as acceleration and deceleration.

[0020] 3、The application constructs a global information interaction and optimization closed loop through an online model adaptive module. The rolling mill stiffness coefficient identified by analyzing the closed loop correction amount of the thickness automatic control module is not only used to update the thickness control model itself, but also updated to the dynamic decoupling feedforward unit synchronously; so that the learning achievements of the subsystem can be directly used to improve the performance of another associated subsystem, breaking the limitation of independent optimization of each subsystem, and realizing the collaborative gain at the system level.

[0021] 4、The adaptive learning mechanism of the application is based on the physical model and the actual operation residual, rather than pure data statistics. This online identification and correction method based on the physical process makes the control model have better interpretability and generalization ability to working condition changes; therefore, the system can better adapt to different specifications and different materials, reducing the need for a large amount of manual re-adjustment due to production batch replacement, and enhancing the overall robustness. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 It is a schematic diagram of the main equipment arrangement of the rolling mill production line of the application; Figure 2 It is a schematic diagram of the hardware composition and network configuration of the computer intelligent control system of the application; Figure 3 It is a software function module architecture diagram of the computer intelligent control system of the application; Figure 4 It is a schematic diagram of the multi-component torque model of the tension control module of the application; Figure 5 It is a schematic diagram of the AGC dynamic decoupling feedforward control principle of the application; Figure 6 It is a schematic diagram of the working principle of the online model adaptive module of the application; Figure 7 It is a schematic diagram of the main screen of the man-machine interface of the application; Figure 8 It is a schematic diagram of the thickness automatic control man-machine interface of the application; Figure 9 It is a schematic diagram of the process setting screen of the application; Figure 10The key parameter trend picture of the application.

[0023] Wherein, 1, uncoiler; 2, straightening device; 3, horizontal loop pinch roll; 4, horizontal loop; 5, entry S roller; 6, first main stand; 7, second main stand; 8, third main stand; 9, fourth main stand; 10, fifth main stand; 11, thickness gauge; 12, exit S roller; 13, pinch roll; 14, flying shear; 15, deflector roller; 16, first coiler; 17, second coiler. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the specification of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the protection scope of the application.

[0025] Referring to Figure 1 , Figure 1 is a schematic diagram of the main equipment arrangement of a rolling mill production line according to an embodiment of the application; the computer intelligent control system provided by the application can be applied to the production line; in the production line, after the strip steel is uncoiled from the uncoiler 1, the strip steel passes through the straightening device 2, the horizontal loop pinch roll 3, the horizontal loop 4, and the entry S roller 5 in sequence, and enters the rolling area composed of the first main stand 6 to the fifth main stand 10; the rolled strip steel is detected in thickness by the thickness gauge 11, and then passes through the exit S roller 12, the pinch roll 13, and the flying shear 14, and is finally coiled by the first coiler 16 or the second coiler 17; the production line further includes tension meters arranged between the main stands, pressure sensors and displacement sensors arranged on the main stands, and rotary encoders for detecting the rotating speeds of the motors.

[0026] Referring to Figure 2 , Figure 2 is a schematic diagram of the hardware composition and network configuration of the computer intelligent control system according to an embodiment of the application; the computer intelligent control system includes a programmable controller (PLC) as a core control unit, for example, a SIEMENS S7-400 series CPU; the PLC is electrically connected with a plurality of drive units and a human-machine interface (HMI) through an industrial Ethernet (such as Profinet) or a field bus; the drive unit, for example, a DCM series DC converter, is used to receive the control instructions output by the PLC, and drive the motors of the uncoiler, the coiler, and each main stand on the production line to operate; the human-machine interface, for example, an industrial computer running WinCC software, provides a graphical interface for the operator to set the process parameters, monitor the running state of the system, and display alarm information.

[0027] Referring to Figure 3, Figure 3 is a computer intelligent control system software function module architecture diagram according to an embodiment of the present application; the computer intelligent control system of the embodiment of the present application, control logic of which runs on a PLC, comprises: a tension control module, an automatic thickness control module, and an online model adaptive module.

[0028] The tension control module is configured to calculate and output motor torque instructions for driving the unwinding machine 1 and the first coiler 16 or the second coiler 17 based on a multi-component torque model, so as to control the strip tension; The automatic thickness control module is configured to adjust rolling parameters of the first to fifth main stands 6-10 through a thickness automatic control algorithm based on a deviation between a set target thickness and an actual thickness detected by the thickness gauge 11, so as to control the strip outlet thickness; The online model adaptive module is configured to collect and analyze running data of the production line in real time during rolling, and correct parameters in the multi-component torque model relied on by the tension control module and rolling physical model parameters relied on by the automatic thickness control module based on a preset adaptive algorithm; The online model adaptive module is electrically or logically connected with the tension control module and the automatic thickness control module, and is used for updating the corrected parameters to the control models relied on by the latter two, so as to realize continuous optimization of the control models.

[0029] Referring to Figures 7 to 10 , this part is an example of a human-machine interface (HMI); an operator interacts with the computer intelligent control system through the human-machine interface; referring to Figure 7 , the main screen is used for centralized monitoring of main technical parameters in the production process, such as motor speed, tension, and coil diameter; referring to Figure 8 , the automatic gauge control (AGC) screen provides a dedicated monitoring and parameter setting interface for the automatic thickness control module; referring to Figure 9 , the process setting screen is used for the operator to input rolling process data before production, including incoming material specifications, target thickness, target tension, etc.; referring to Figure 10 , the trend chart screen is used to display real-time values and historical values of key parameters (such as main stand speed, coiling tension, thickness deviation, etc.) in the form of curves, for analyzing system control performance.

[0030] The main equipment units in the rolling mill production line shown in Figure 1 will be described in detail below; The uncoiler 1 includes a transmission mechanism driven by a DC motor, a speed reducer, a rotary encoder for speed feedback, and a winding core for winding the strip steel; the function of the uncoiler 1 is to run in a tension control mode to provide the strip steel material with a set tension for the subsequent rolling area; in an embodiment, a laser range finder can be configured to detect the real-time outer diameter of the steel coil as an auxiliary input for the calculation of the coil diameter.

[0031] The entry S roller 5 and the exit S roller 12 are respectively arranged at the entry and exit of the rolling area composed of the main frame; the entry S roller 5 is used to establish a stable back tension zone for the strip steel before it enters the first main frame 6; the exit S roller 12 is used to establish a stable front tension zone for the strip steel after it leaves the last main frame before entering the coiling area; the establishment of the two tension zones helps to maintain the stable operation of the strip steel during rolling and provides isolation for the precise control of the inter-frame tension.

[0032] The first main frame 6 to the fifth main frame 10 are the core parts for performing thickness reduction of the strip steel; in an embodiment, each main frame is a four-roller rolling mill structure, including a roller system composed of work rolls and backup rolls; the key components of each main frame include a hydraulic screwdown cylinder, a high-precision displacement sensor, and a pressure sensor; the displacement sensor is used to detect the roll gap position of the work roll in real time; the pressure sensor is used to detect the rolling force generated during rolling in real time; the signals detected by the displacement sensor and the pressure sensor are transmitted to the thickness automatic control module as the key inputs for the thickness control algorithm; the hydraulic screwdown cylinder is the main actuator of the thickness automatic control module, which adjusts the roll gap according to the output instructions of the module.

[0033] The tension meter is arranged between two adjacent main frames, for example, between the first main frame 6 and the second main frame 7; its function is to detect the tension of the strip steel in the section in real time; its detection signal is transmitted to the computer intelligent control system for closed-loop control or monitoring of the tension during rolling.

[0034] The thickness gauge 11, such as a non-contact radio thickness gauge, is arranged on the exit side of the last main frame (the fifth main frame 10); its function is to continuously measure the actual thickness of the rolled strip steel; its measurement principle is based on the specific relationship between the transmission amount after the ray penetrates the strip steel and the thickness of the strip steel; the real-time thickness signal output by the thickness gauge 11 is the main basis for the closed-loop feedback control of the thickness automatic control module; in some embodiments, a thickness gauge can also be arranged before the entry S roller 5 to measure the incoming thickness for feedforward control.

[0035] The pinch flying shear 14 is arranged between the exit S roller 12 and the coiler; in one embodiment, it is a drum-type flying shear driven by an independent motor; its function is to perform precise shearing action on the high-speed running strip when the No. 1 coiler 16 needs to switch to the No. 2 coiler 17 for continuous production after the coiler 16 has coiled a full reel, so as to realize non-stop coil change.

[0036] Coiler No. 16 and coiler No. 2 17 are used to coil the rolled strip into coils; each coiler includes a reel, an expansion and contraction mechanism, and a transmission part driven by a DC motor; during the rolling process, the currently working coiler operates in tension control mode, and the tension it generates provides stable front tension for the entire rolling zone.

[0037] Reference Figure 4 , Figure 4 is a schematic diagram of a multi-component torque model of a tension control module 210 according to an embodiment of the present invention; the core function of the tension control module 210 is to calculate and output a total output torque instruction for the drive motor of the uncoiler 1 or the coiler (16, 17); the total output torque is calculated based on a multi-component torque model, which decouples multiple physical factors that affect tension; the model calculates the total output torque The formula is: ; Where, is the reference tension torque, is the friction compensation torque, is the dynamic inertia compensation torque, is the taper tension correction torque; the calculation method of each torque component will be described in detail below.

[0038] In order to accurately calculate the above torque components, it is necessary to first obtain the real-time outer diameter of the steel coil ( In one embodiment, the calculation of the real-time outer diameter is based on the physical principle that the linear speed between the main roller (e.g., the outlet S roller 12) and the coiler (16 or 17) at the exit of the rolling zone is kept constant; the system obtains the speed of the main roller motor through the motor encoder ( ) and the speed of the coiler motor ( ), combined with the known mechanical parameters, the real-time coil diameter is calculated using the following formula: ; Where, The diameter of the main roller; The real-time speed of the main roller motor; is the real-time speed of the coiler motor; The transmission reduction ratio of the main roller; is the transmission reduction ratio of the coiler; the calculated The value is processed by a filter (such as a low-pass filter or a moving average filter) to eliminate the calculated value jump caused by electrical signal fluctuations and ensure the stability of the coil diameter value.

[0039] Reference tension torque is the core torque required to establish the target tension on the strip; in one embodiment, the torque is calculated using the maximum torque calculation method; first, a maximum coil diameter ( ) when maintaining constant tension ( ); Then, the current reference tension torque is calculated based on the proportional relationship between the real-time coil diameter and the maximum coil diameter; the calculation formula is: ; Where, is the reference tension torque; The maximum torque required to maintain the target tension at the maximum coil diameter; The real-time outer diameter of the steel coil; is the preset maximum finished coil diameter; the reference torque calculated by this method can make the tension acting on the strip ( ) remains constant throughout the entire winding process.

[0040] Friction compensation torque This is the compensating torque applied to overcome the mechanical friction resistance in the transmission system. This compensating torque is related to both the motor speed and the coil load (i.e., coil weight, which is related to the coil diameter). In one embodiment, the initial value of this torque is obtained through an offline automatic measurement program. This program controls the coiler to operate at multiple discrete speed points, from zero speed to maximum speed and then from maximum speed to zero speed, when unloaded or loaded with coils of varying weights. During stable operation at each speed point, the actual torque feedback value of the motor is collected and averaged to obtain a series of data points (speed, coil diameter, friction torque). These data points constitute a two-dimensional friction compensation lookup table or function: ; During the actual rolling process, the tension control module 210 adjusts the tension according to the current rolling line speed. and real-time roll diameter , through table lookup and interpolation operation, the friction compensation torque required under the current working condition is obtained .

[0041] Dynamic inertia compensation torque It is the compensating torque required to overcome the dynamic resistance or assistance generated by the change in the steel coil's own rotational inertia during the acceleration and deceleration process of the machine train. The calculation of this torque is based on the geometric dimensions and material properties of the steel coil and the acceleration of the machine train. In one embodiment, its calculation formula is: ; Where, is the strip width; is the density of the strip material; is the coil tightness coefficient, and its value range is 0.85 to 0.9; is the transmission reduction ratio of the coiler; The real-time outer diameter of the steel coil; is the empty drum diameter of the coiler drum; is the current rolling line speed; is the rate of change of rolling line speed, that is, linear acceleration; when the machine train accelerates, For positive, is the positive compensation torque; when the train decelerates, is negative, It is the negative compensation torque.

[0042] Taper tension correction torque To prevent the steel coil from suffering from quality problems such as coil collapse and internal stress, a part of the torque is deliberately subtracted from the reference tension as the coil diameter increases. It is not an independent torque, but an adjustment to the target tension, which is ultimately reflected in the correction of the total torque. In actual control, the target tension with a taper effect is usually calculated, and then the corresponding torque is calculated. The form of the taper is consistent, and the taper effect can be equivalent to a correction term subtracted from the reference torque; an equivalent taper tension correction torque is calculated as follows: ; Where, The taper coefficient is set by the operator according to the process requirements, and its value range is 0 to 1; The real-time outer diameter of the steel coil; is the empty drum diameter of the coiler drum; is the preset maximum finished product roll diameter; this formula makes the real-time roll diameter From the diameter of the empty cylinder Grow to maximum roll diameter During the process, the torque is corrected It grows linearly from 0 to .

[0043] This means that when winding to the maximum coil diameter, the actual applied tension torque is reduced by 1% compared to the reference torque. proportion.

[0044] Reference Figure 8 , Figure 8Fig. 1 is a schematic diagram of an automatic gauge control (AGC) human-machine interface according to an embodiment of the present application; and Fig. 2 is a schematic diagram of an automatic gauge control module 220 according to an embodiment of the present application, the core function of which is to accurately adjust the exit thickness of the strip based on a rolling physical model through a combination of closed-loop feedback control and feedforward control.

[0045] The basic feedback control loop of the automatic gauge control module 220, also referred to as gauge monitoring AGC, has the following working process: first, the thickness gauge 11 located at the exit of the rolling zone continuously measures the actual thickness of the rolled strip ; at the same time, the system obtains the target thickness of the current pass from the process settings (see Figure 9 ) ; the automatic gauge control module 220 calculates the thickness deviation between the two : ; Subsequently, the thickness deviation is input to a PID (proportional-integral-derivative) controller; the controller calculates a roll gap correction amount for eliminating the thickness deviation according to the proportional (P), integral (I), and derivative (D) parameters set in it ; the correction amount is sent to the hydraulic press-down system of the main stand through a control command to drive the hydraulic press-down cylinder to act, thereby adjusting the actual roll gap and making the exit thickness tend to the target thickness; in an embodiment, the output of the PID controller is limited within a pre-set range to prevent excessive adjustment from impacting the stability of the system.

[0046] Since the gauge monitoring AGC is a feedback control, the adjustment effect occurs necessarily later than the appearance of the thickness deviation, and there is a time delay determined by the installation distance of the thickness gauge and the rolling line speed; in order to overcome this limitation, the automatic gauge control module 220 of the present application further integrates a dynamic decoupling feedforward control unit.

[0047] Referring to Figure 5 , Figure 5 Fig. 3 is a schematic diagram of the AGC dynamic decoupling feedforward control principle according to an embodiment of the present application; the function of the dynamic decoupling feedforward control unit is to predict and compensate the thickness fluctuation caused by the inter-stand tension.

[0048] The process works as follows: The dynamic decoupling feedforward control unit monitors in real time the tension changes that directly affect the rolling process of the main frames. Specifically, for any main frame, the dynamic decoupling feedforward control unit monitors the set value changes of the inlet tension (post-tension) and outlet tension (pre-tension) of the main frame, or the actual tension changes directly detected by the tension meter installed between the main frames. It is important to emphasize that the inter-stand tension used for thickness feedforward calculation here is a pure rolling process parameter. It does not include and does not need to consider the taper tension correction or dynamic inertia compensation calculated specifically for the coiler or uncoiler. The theoretical basis for thickness calculation and control by the thickness automatic control module 220 is the Gage-Meter-Equation, which describes the initial roll gap setting value ( )、Rolling force( ) caused by the elastic deformation of the frame, the outlet thickness ( ); its basic form is: ; Where, is the outlet thickness after rolling; It is the roll gap setting value when there is no rolling force; is the actual rolling force; is the comprehensive stiffness coefficient of the rolling mill; this equation shows that the exit thickness depends not only on the roll gap setting, but is also affected by the rolling force and the rolling mill stiffness; feedforward control is based on this principle, actively compensating before the rolling force changes due to tension.

[0049] When the tension between the racks changes ( ), the unit predicts the effect of the tension change on the rolling force based on the physical model, and finally calculates the resulting change in outlet thickness ( ); Its prediction formula can be simplified and linearized as: ; Where, is the predicted feedforward thickness compensation caused by tension change; is the detected inter-rack tension variation at the inlet or outlet side of a specific main rack; is the tension influence coefficient, which combines the rolling mill stiffness, plastic deformation resistance and elastic deformation characteristics of the material, and characterizes the change in outlet thickness caused by unit tension change.

[0050] Calculated feedforward thickness compensation Converted into an equivalent roll gap feedforward correction In one embodiment, the conversion is based on the mill bounce equation, which can be expressed as: ; In the formula, is the plastic deformation resistance coefficient of the material; is the rigidity coefficient of the rolling mill.

[0051] Finally, the roll gap feedforward correction amount is superimposed with the feedback correction amount calculated by the PID controller to form the total roll gap correction instruction : ; The total correction instruction is sent to the hydraulic screwdown system for execution; in this way, the thickness disturbance caused by the inter-stand tension fluctuation is offset by the feedforward signal as soon as it is generated, greatly improving the thickness control accuracy under dynamic working conditions.

[0052] Referring to Figure 6 , Figure 6 is a schematic diagram of the working principle of the online model adaptive module according to an embodiment of the application; the online model adaptive module 230 is the core of realizing the continuous optimization of the control system performance of the application; the module performs online identification and correction on the key parameters in the control model by analyzing specific data in the production process in real time, so as to adapt to the equipment state changes and working condition disturbances.

[0053] One function of the online model adaptive module 230 is to perform online identification on the rolling physical model relied on by the thickness automatic control module 220, specifically, to perform online identification on the rolling mill rigidity coefficient (K) ); the rolling mill rigidity is a physical quantity describing the degree of elastic deformation of the rolling mill stand when it bears the rolling force, and is a core parameter in the rolling mill bounce equation; the identification principle is based on the following relationship: when the feedback control loop (thickness monitoring AGC) of the thickness automatic control module 220 produces a roll gap correction amount (Δh) ), there will be a corresponding change in the rolling force (ΔF) ); the ratio of the two reflects the current rolling mill rigidity; if the feedback control loop continuously outputs a non-zero correction amount, it indicates that there is a deviation between the rolling mill rigidity coefficient used in the current control model and the actual value.

[0054] The implementation steps of online identification of the rolling mill rigidity are as follows: in a preset time window with relatively stable rolling working conditions, the online model adaptive module 230 synchronously collects multiple groups of feedback roll gap correction amounts (Δh ) output by the thickness automatic control module 220 and the corresponding rolling force change amounts (ΔF ); after collecting enough data, an online parameter identification algorithm is used to calculate; in one embodiment, the algorithm is recursive least square (RLS); the algorithm iteratively calculates to find an optimal value, so that the sum of squared errors between the predicted rolling force change and the actual observed rolling force change is minimized; the simplified core idea is to fit the following linear relationship: ; Through this method, an updated rolling mill stiffness coefficient value closer to the current actual working condition can be obtained ; To ensure the accuracy and stability of the identification result, the online model adaptive module 230 will perform an effectiveness judgment before starting identification; the judgment includes: confirming whether the current rolling mill is in a stable rolling speed interval, and whether the signal-to-noise ratio of the collected rolling force change (F) ) and the roll gap correction amount (h) ) is higher than the preset threshold to exclude noise interference; after identifying the new stiffness coefficient , the module will also perform a convergence check, for example, whether the coefficient values calculated by continuous iterations are within a small error range; only when the new coefficient value passes the convergence check, it is confirmed as a valid value and used for subsequent global update.

[0055] In an alternative embodiment, the online parameter identification algorithm for identifying the rolling mill stiffness can also use Kalman filter (Kalman-Filter) algorithm or gradient descent (Gradient-Descent) algorithm, which can also iteratively optimize the model parameters according to the input and output data.

[0056] Another function of the online model adaptive module 230 is to perform online correction on the friction compensation model relied on by the tension control module 210; the correction principle is based on the following idea: in the quasi-steady state working condition during rolling (i.e. the rolling line speed and the strip tension are basically constant), the total output torque (T ) of the coiler motor is mainly composed of the reference tension torque (T ) and the friction compensation torque (T ); at this time, the dynamic inertia compensation torque (T ) tends to zero; therefore, by subtracting the theoretically calculated reference tension torque and taper tension correction torque from the actual output total torque of the motor, a torque value reflecting the current real friction can be obtained.

[0057] The online correction implementation steps of the friction compensation model are as follows: the system first judges whether the current is in the quasi-steady state working condition; if yes, a friction residual (R ) ; its calculation formula is: ; In the formula, is the actual total output torque obtained from the motor driver; and is the theoretical value calculated by the tension control module 210 according to the current setting; is the compensation value looked up from the existing friction compensation two-dimensional lookup table according to the current speed and roll diameter.

[0058] The calculated friction residual error , that is, the deviation between the current model value and the actual value; then, the system uses a recursive average filtering algorithm with a forgetting factor to smoothly update the friction residual error to the friction compensation two-dimensional lookup table; its update formula is: ; In the formula, is the updated friction compensation value corresponding to the current speed and roll diameter ; is the value before updating; is the forgetting factor, whose value range is 0 to 1, used to adjust the weight of new and old data; in this way, the friction compensation model can be gradually learned and corrected online during the production process.

[0059] In an alternative embodiment, the algorithm for smoothing the update of the friction residual error can also use the least mean square (LMS) adaptive filtering algorithm, which can also adjust the model parameters (i.e., the values in the lookup table) step by step according to the error signal (i.e., the friction residual error).

[0060] In order to realize the collaborative optimization of the overall performance of the system, the online model adaptive module 230 also includes a global parameter collaborative update mechanism; when the module identifies a new parameter that is stable and convergent through the above method (for example, the value fluctuates within a small tolerance range for consecutive multiple calculation periods), the system will perform a global update of the parameter.

[0061] Specifically, when the new rolling mill stiffness coefficient is confirmed to be valid, the value will be transmitted to two different units of the thickness automatic control module 220 at the same time: First, its basic feedback control loop, used to update the parameter in the rolling mill bounce equation; Second, in its dynamic decoupling feedforward control unit, used to update the parameter relied on when calculating the feedforward roll gap correction amount The parameters are consistent and accurate, and control conflicts or performance degradation caused by inconsistent model parameters are avoided.

[0062] The computer intelligent control system 200 and the cooperation of the modules will be described in detail below.

[0063] Rolling preparation stage: the operator first enters the process setting screen (see Figure 9 ) through the human-machine interface (HMI) and inputs the process parameters of the batch of strip steel; these parameters include the width and thickness of the incoming material, the target outlet thickness and target tension of each pass, the rolling speed, the taper coefficient, etc.; the computer intelligent control system 200 receives and stores these set values as the target of subsequent automatic control.

[0064] Stripping and tension building stage: the operator starts the stripping program, the system drives the motors of the production line at a low speed, sends the head of the strip from the uncoiler 1 to the entrance S roller 5, and then to each main stand in turn, and finally clamps it with the pinch roll of the coiler; after the stripping is completed, the system enters the tension building stage, which is carried out in zones: The precise back tension required for rolling is established between the entrance S roller 5 and the first main stand 6; at the same time, the uncoiler 1 operates in tension control mode, which mainly provides stable incoming tension for the entrance S roller 5 and ensures the storage amount of the entrance loop (if any).

[0065] The inter-stand tension is established by the speed difference between each adjacent main stand.

[0066] The main front tension required for rolling is established between the coiler and the exit S roller 12 (or the last main stand); through this way of building tension in zones, the tension in the core rolling zone is stable and is not directly affected by the dynamic process on the uncoiler side.

[0067] Acceleration and dynamic control stage: once the tension is stable, the operator issues the rolling start instruction, and the system controls the production line to accelerate from the stripping speed to the set rolling speed; during the entire acceleration process, the modules in the system work in high coordination: The tension control module 210 calculates and superimposes a dynamic inertia compensation torque , to overcome the inertia torque generated by the acceleration of the coil, ensuring the stability of the tension; at the same time, the dynamic decoupling feedforward control unit in the thickness automatic control module 220 is activated; it monitors the changes in the set value of the tension control module 210 in real time (for example, changes in torque caused by maintaining constant tension), and according to the tension influence model, it calculates in advance the thickness fluctuation caused by the tension change, and generates a feedforward roll gap correction , actively compensating for the roll gap of the main frame.

[0068] Through the synergistic effect of the above two modules, the system can maintain the stability of the strip tension and the accuracy of the outlet thickness at the same time under the condition of accelerating such severe dynamics.

[0069] Stable rolling and adaptive learning phase: after the production line reaches the set rolling speed, it enters the stable rolling phase; in this phase, the system continues to carry out high-precision closed-loop control, and the online model adaptive module 230 starts to work: The feedback control loop of the thickness automatic control module 220 continues to work, and according to the thickness deviation fed back by the thickness gauge 11, it calculates the feedback correction amount through the PID controller , and the feedback correction amount is superimposed with the possible feedforward correction amount (for compensating for incoming thickness fluctuations, etc.) to jointly adjust the roll gap; the online model adaptive module 230 is started in the background; it continuously monitors the feedback correction amount of the thickness automatic control module 220 and the corresponding change in rolling force , and when the preset identification conditions are met, it starts the online identification algorithm of the mill stiffness to calculate the updated stiffness coefficient value.

[0070] At the same time, this module also makes online corrections to the friction compensation model in the tension control module 210 by analyzing the residual error between the actual output torque of the motor and the theoretically calculated torque under quasi-steady state conditions; When the new parameters identified (such as ) are confirmed to be stable and reliable, the online model adaptive module 230 updates the new parameters to the feedback control loop and the feedforward control unit of the thickness automatic control module 220 at the same time through the global parameter collaborative update mechanism, realizing self-optimization of the control model; Parking and coil changing phase: when a coil is about to be completed, the system automatically starts the deceleration program according to the preset coil length or number of turns; similar to the acceleration phase, the tension control module 210 applies a negative dynamic inertia compensation torque, and the feedforward unit of the thickness automatic control module 220 also works accordingly, ensuring the smoothness of the deceleration process; the subsequent coil changing process varies depending on the equipment configuration: For the production line equipped with double coilers, the system smoothly reduces the line speed to a preset coiling speed without stopping; at this speed, the flying shear 14 is activated and precisely actuated to cut the strip; then, the strip head is automatically fed into the other empty coiler through the guide device and clamped; once the new coil is established, the production line is immediately accelerated to the rolling speed, thus realizing continuous coiling without stopping and greatly improving the production efficiency; For the production line equipped with single coiler, the system must reduce the line speed to zero and completely stop; after the production line stops, the operator performs the full coil uncoiling and the empty coil preparation; after the coiling is completed, the system is restarted to start the next rolling cycle.

[0071] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A computer intelligent control system for a rolling mill, characterized in that: The system comprises: A rolling mill production line, wherein the rolling mill production line is provided with: an uncoiler (1), a straightening device (2), a horizontal looper pinching (3), a horizontal looper (4), an entrance S roller (5), a No. 1 main frame (6), a No. 2 main frame (7), a No. 3 main frame (8), a No. 4 main frame (9), a No. 5 main frame (10), a thickness gauge (11), an exit S roller (12), a pinching roller (13), a flying shear (14), a deflector roller (15), a No. 1 coiler (16), and a No. 2 coiler (17); The rolling mill production line further includes a tension meter disposed between the main frames, a pressure sensor and a displacement sensor disposed on the main frames, and a rotary encoder for detecting the rotational speed of each motor; A computer intelligent control system electrically connected to the actuators and sensors in the rolling mill production line, the computer intelligent control system comprising: A tension control module is used to calculate and output a motor torque instruction for driving the uncoiler (1) and the first coiler (16) or the second coiler (17) based on a multi-component torque model including a reference tension torque, a friction compensation torque, and a dynamic inertia compensation torque; A thickness automatic control module for adjusting the rolling parameters of the first to fifth main frames (6) to (10) by a thickness automatic control algorithm based on a deviation between a set target thickness and an actual thickness detected by a thickness gauge (11); The online model adaptive module is used to collect and analyze the operating data of the rolling mill production line in real time during the rolling process, and based on a preset adaptive algorithm, online correct the parameters in the multi-component torque model and the rolling physical model parameters on which the thickness automatic control algorithm depends, and update the corrected parameters to the tension control module and the thickness automatic control module to continuously optimize the control model.

2. The computer intelligent control system for a rolling mill according to claim 1, characterized in that: The thickness automatic control module also includes: a dynamic decoupling feedforward unit configured to monitor a change in inter-stand tension acting on the main stand and, based on a preset rolling physics model, predict a change in outlet thickness that will be caused by the tension change; The dynamic decoupling feedforward unit is further configured to generate a feedforward roll gap correction value according to the predicted outlet thickness change, and superimpose the feedforward roll gap correction value with the feedback roll gap correction value generated by the thickness feedback control loop to form a total roll gap correction instruction; Therefore, when the tension between the frames changes, the thickness is actively compensated to eliminate the coupling interference of tension change on thickness control.

3. The computer intelligent control system for a rolling mill according to claim 1, characterized in that: In the online model adaptation module, the step of online correction of rolling physical model parameters includes: Real-time collection of the feedback roll gap correction value output by the thickness automatic control module to eliminate thickness deviation and the corresponding rolling force change detected by the pressure sensor; Taking the feedback roll gap correction amount and rolling force variation as input, identifying the rolling mill stiffness coefficient in the rolling physical model through an online parameter identification algorithm; The identified rolling mill stiffness coefficient is updated to the thickness automatic control module.

4. The computer intelligent control system for a rolling mill according to claim 3, characterized in that: The identified rolling mill stiffness coefficient is also updated to the dynamic decoupling feedforward unit for use in the exit thickness variation prediction calculation of the dynamic decoupling feedforward unit.

5. The computer intelligent control system for a rolling mill according to claim 1, characterized in that: The step of the online model adaptive module performing online correction on the parameters in the multi-component torque model comprises: Under the quasi-steady-state condition of the rolling process, the friction residual is calculated based on the actual output torque of the motor, the reference tension torque, the taper tension correction torque, and the output value of the current two-dimensional friction compensation model; and the two-dimensional friction compensation model is corrected online based on the friction residual. The calculation formula of the friction residual is: ； Where, is the friction residual, is the actual output torque of the motor, is the reference tension torque, is the taper tension correction torque, is the friction compensation torque.

6. The computer intelligent control system for a rolling mill according to claim 1, characterized in that: The multi-component torque model on which the tension control module is based calculates the total output torque provided to the uncoiler (1) or coiler motor as follows: ； Where, is the base tension torque, which is the core torque required to maintain constant tension; Friction compensation torque is the torque required to overcome the mechanical friction of the transmission system; Dynamic inertia compensation torque is the torque required to compensate for the change in the coil's rotational inertia when the train accelerates or decelerates. Correct torque for taper tension.

7. The computer intelligent control system for a rolling mill according to claim 1, characterized in that: The operating data of the rolling mill production line include: The inter-stand tension detected by the tension meter, the outlet thickness detected by the thickness gauge (11), the rolling force detected by the pressure sensor, the roll gap position detected by the displacement sensor, and the motor speed detected by the rotary encoder.

8. The computer intelligent control system for a rolling mill according to claim 1, characterized in that: The inlet S roller (5) and the outlet S roller (12) are used to establish a stable tension zone before and after the main part of the rolling mill, and the tension control module controls the driving motors of the inlet S roller (5) and the outlet S roller (12) to maintain constant tension in the tension zone.

9. The computer intelligent control system for a rolling mill according to claim 1, characterized in that: The computer intelligent control system is also used for: The flying shear (14) is controlled to perform a shearing action after the No. 1 coiler (16) is fully wound and before switching to the No. 2 coiler (17), for uninterrupted rolling.

10. A computer intelligent control method for a rolling mill, applied to the system according to any one of claims 1 to 9, characterized in that: The method comprises the following steps: S1. Calculate and output motor torque instructions for driving the uncoiler and coiler based on a multi-component torque model including a reference tension torque, a friction compensation torque, and a dynamic inertia compensation torque; S2, adjusting the rolling parameters of the main frame by an automatic thickness control algorithm based on the deviation between the set target thickness and the actual thickness detected by the thickness gauge (11); S3. During the rolling process, real-time data of the rolling mill production line is collected and analyzed, and based on a preset adaptive algorithm, parameters in the multi-component torque model and parameters of the rolling physical model on which the thickness automatic control algorithm depends are corrected online; S4. Updating the parameters corrected in the online adaptive step to the model on which the tension control step and the thickness control step depend, for continuous optimization of the control model.

Images