A first-steel hitting rate control method based on threading self-adaptation

Abstract

The present application relates to a kind of first steel hit rate control method of hot strip mill equipment, the hot strip mill equipment includes N frame that can be supplied with strip and first steel hit rate control system;Its characterized in that, the first steel hit rate control method of the hot strip mill equipment includes the following steps that the first steel hit rate control system executes: with zero value self-learning module receives the first steel big data module sent historical first steel data, obtains zero value self-learning value ΔL i ;Estimate the deformation resistance deviation average value Δki of each frame rolling force;Estimate the thickness deviation Δhi of each frame;The present application is based on the model of strip threading self-adaption, in combination with multiple algorithms, more accurate calculation rolling force, bending force, the deviation correction coefficient of roll gap, collects and comprehensively considers multiple first steel hit deviation, obtains zero value self-learning coefficient;Can obtain more than the set number of three previous frames, improve the hit rate of first steel.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of hot rolling production, and particularly relates to a first-steel hitting rate control method for a hot rolling strip steel equipment. BACKGROUND

[0002] The traditional strip passing adaptive dynamic model works on the first three stands, i.e. F1-F3, considers the deviation of the measured rolling force from the set rolling force and the deviation of the measured roll gap from the set roll gap, and uses smoothing processing to correct these deviations. With the improvement of the technology and level in the field of hot rolling strip steel, under the combined action of the model calculation and self-learning, the width, thickness and other indicators of each strip can accurately and stably meet the protocol requirements. In this way, although the thickness hitting rate of the strip head can be improved to a certain extent, the first steel is limited by many uncertain factors such as incoming material thickness, temperature and equipment grinding quality, which limits the accuracy of model calculation and cannot meet the current thickness requirements. SUMMARY

[0003] In view of the above technical problems, the present application relates to a first-steel hitting rate control method for a hot rolling strip steel equipment, which comprises N stands for strip passing and a first-steel hitting rate control system. The first-steel hitting rate control method comprises the following steps performed by the first-steel hitting rate control system:

[0004] The big data module receives, calculates and stores the data of the previous first steels;

[0005] The zero-value self-learning module receives the data of the previous first steels sent by the first-steel big data module, and finally obtains a zero-value self-learning value ΔL i The zero-value self-learning value ΔL i is sent to a presetting module and a dynamic setting module;

[0006] The correction calculation module receives the deviation of the measured rolling force from the preset rolling force when the stand is passing, estimates the average value Δki of the deformation resistance deviation of the stand rolling force, and corrects the stand rolling force according to the average value Δki of the deformation resistance deviation of the stand rolling force;

[0007] The correction calculation module receives the deviation of the measured rolling force from the preset rolling force when the stand is passing, estimates the thickness deviation Δhi of the stand roll gap, and corrects the stand roll gap according to the thickness deviation Δhi;

[0008] The correction calculation module calculates the fitting correction coefficient and outputs it to the dynamic setting module.

[0009] The second aspect of the present invention relates to a first strip hit rate control system for a hot-rolled strip steel equipment, comprising at least one processor; and a memory storing instructions which are executed by the at least one processor.

[0010] The beneficial effects of this invention are that, based on the adaptive belt-threading model, combined with artificial intelligence algorithms such as big data and machine learning, it can more accurately calculate the deviation correction coefficients of rolling force, bending roll force, and roll gap, collect and comprehensively consider the first steel hit deviation multiple times, obtain a zero-value self-learning coefficient, and obtain more than the set number of times for the first three stands, which greatly improves the hit rate of the first steel. Attached Figure Description

[0011] Figure 1 : Control model flow diagram;

[0012] Figure 2 : Control flow diagram;

[0013] Figure 3 Comparison chart of the first steel data of the control method of this invention and the original method;

[0014] (a) Deviation between measured thickness and target thickness;

[0015] (b) Yield rate;

[0016] A represents the original method;

[0017] B represents the control method of the present invention;

[0018] Figure 4 Diagram showing the relationship between strip steel (first strip), hardware equipment, and control system. Detailed Implementation

[0019] The following embodiments further illustrate the content of the present invention, but should not be construed as limiting the present invention. Any modifications or substitutions made to the methods, steps, or conditions of the present invention without departing from the spirit and essence of the invention are within the scope of the present invention.

[0020] Terminology Explanation

[0021] Front racks: rack 1, rack 2, and rack 3, i.e., F1 to F3;

[0022] Rear racks: Rack 5, Rack 6 and Rack 7, i.e. F5 to F7; or F1 to F7 can be used to represent racks 1 to 7.

[0023] Traditional adaptive dynamic strip rolling models operate on the first three stands (F1-F3), considering the deviations between measured and set rolling forces, and between measured and set roll gaps, using smoothing techniques to correct these deviations. With advancements in technology and expertise in hot-rolled strip steel, a combination of model calculation and self-learning methods allows for accurate and stable fulfillment of protocol requirements for width, thickness, and other parameters of each strip. While this improves the thickness accuracy of the strip head to some extent, the accuracy of model calculations for the first strip is limited by numerous uncertainties related to incoming material thickness, temperature, and grinding quality, ultimately failing to meet current thickness requirements.

[0024] Furthermore, traditional adaptive dynamic models for weaving still have the following problems:

[0025] 1. Traditional adaptive dynamic models for weaving only work in F1 to F3, and the adjustment range is limited.

[0026] 2. Only the deviation between the measured rolling force and the set rolling force, and the deviation between the measured roll gap and the set roll gap were considered, resulting in insufficient reference data.

[0027] 3. Only simple smoothing was used to correct the deviation, resulting in insufficient accuracy.

[0028] The specific embodiments provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0029] To address the technical problems in the prior art, some embodiments, such as Figure 1 As shown, a method for controlling the first strip hit rate of a hot-rolled strip steel equipment is disclosed. The hot-rolled strip steel equipment includes N stands for strip threading and a first strip hit rate control system. The method for controlling the first strip hit rate of a hot-rolled strip steel equipment includes the following steps performed by the first strip hit rate control system:

[0030] The zero-value self-learning module receives the data from each first steel bar sent by the first steel bar big data module, and finally obtains the zero-value self-learning value ΔL. i Send the zero-value self-learning value ΔL i To the preset module and the dynamic setting module;

[0031] The deviation between the measured rolling force and the preset rolling force during the tape threading process of the receiving frame is corrected by the calculation module. The average value of the deformation resistance deviation Δki of the frame rolling force is estimated, and the frame rolling force is corrected based on the average value of the deformation resistance deviation Δki of the frame rolling force.

[0032] The correction calculation module receives the deviation between the measured rolling force and the preset rolling force when the strip is threaded through the stand, estimates the thickness deviation Δhi of the roll gap between the stand, and corrects the roll gap between the stand based on the thickness deviation Δhi.

[0033] The correction calculation module calculates the fitting correction coefficients and outputs them to the dynamic setting module.

[0034] In some embodiments, the zero-value self-learning value ΔLi in the zero-value self-learning module is determined according to equation (I):

[0035] ΔL i =m i-1 *f(i-1)+m i-2 *f(i-2)+m i-3 *f(i-3)+……(I)

[0036] In the formula, ΔLi represents the self-learning value set to zero for the i-th time; f(i-1) represents the value after big data processing of the first steel support for the (i-1)-th time; m i-1 It represents the weighting coefficient of the first steel data processing value in the (i-1)th iteration relative to the zero self-learning value in the i-th iteration.

[0037] In some embodiments, such as Figure 1 and Figure 2 As shown, the deviation between the measured rolling force and the preset rolling force when the receiving frame is threaded is corrected by the calculation module, and the average value of the deformation resistance deviation Δki of the frame rolling force is estimated. The frame rolling force is then corrected based on the average value of the deformation resistance deviation Δki of the frame rolling force.

[0038] In some embodiments, such as Figure 1 He Ru Figure 2 As shown, the correction calculation module receives the deviation between the measured rolling force and the preset rolling force when the strip is threaded through the stand, estimates the thickness deviation Δhi of the stand roll gap, and corrects the stand roll gap based on the thickness deviation Δhi.

[0039] The fitting correction coefficients are calculated by the correction calculation module and output to the dynamic setting module.

[0040] In other embodiments, the average deformation resistance deviation Δki is determined according to Equation II:

[0041] Δki=ΔFi / Fi(II)

[0042] In the formula, ΔFi represents the deviation between the measured rolling force and the preset rolling force of the stand; Fi represents the preset rolling force of the stand; i represents the stand number, i=1,2,…,n.

[0043] The thickness deviation Δhi of the frame is determined according to Equation III:

[0044] Δhi=ΔSi+ΔFi / Mi(III)

[0045] In the formula, ΔSi is the deviation between the measured roll gap and the predicted roll gap of the mill stand; ΔFi is the deviation between the measured rolling force and the preset rolling force of the mill stand; and Mi is the mill stand stiffness.

[0046] In some embodiments, a method for controlling the first strip hit rate of a hot-rolled strip steel equipment further includes the following steps performed by the first strip hit rate control system:

[0047] The correction calculation module recalculates the stand speed Vpi based on the changed strip exit thickness hi and forward slip coefficient fi, and compares it with the set speed Vmi. The difference is used for stand speed correction. The roll gap variation is matched with the stand rolling speed to ensure equal flow rates between stands.

[0048] In some embodiments, the first steel data module receives the first steel data, performs data collection, storage and analysis, extracts key indicators and uses machine learning and deep learning technologies to establish a thickness hit rate control model, and stores and sends the processed data in a fixed format.

[0049] By collecting, cleaning, storing, and deeply analyzing the production data of each first steel bar, key indicators are extracted, and a thickness hit rate control model is established using technologies such as machine learning and deep learning. The processed data is then presented in a fixed format as the basis for subsequent use.

[0050] In some embodiments, such as Figure 1 As shown, a method for controlling the first strip hit rate of a hot-rolled strip steel equipment also includes an algorithm library interface, wherein the algorithm library is configured as follows:

[0051] The correction coefficients for the fitted rolling force deviation, roll gap deviation, and speed deviation in the dynamic setting module are calculated using a combination of linear regression, nonlinear regression, neural network, machine learning, and deep learning algorithms.

[0052] The following sections introduce the application of each algorithm in improving the hit rate of the first steel section thickness.

[0053] (1) Linear regression can be used to predict the linear relationship between the first steel thickness hit rate and other factors. For example, historical data can be used to predict the relationship between the first steel thickness hit rate and factors such as rolling force, temperature, and speed. Its effectiveness is affected by factors such as feature selection and data preprocessing. Commonly used evaluation indicators include mean square error and R². Its main parameters include the intercept term and characteristic coefficients.

[0054] (2) Nonlinear regression can be used to predict the relationship between the first steel thickness hit rate and nonlinear factors. For example, historical data can be used to predict the relationship between the first steel thickness hit rate and factors such as the higher power of rolling force and the exponential term of temperature. Its effectiveness depends on factors such as model selection, feature engineering, and model parameter tuning. Commonly used evaluation indicators include mean square error and R². Its main parameters include intercept term, characteristic coefficient, and regularization parameter.

[0055] (3) Neural networks can be used to predict the nonlinear relationship between the first steel thickness hit rate and multiple factors. For example, historical data can be used to predict the complex relationship between the first steel thickness hit rate and factors such as rolling force, temperature, and speed. Neural networks are usually trained using forward propagation or backpropagation algorithms, and commonly used activation functions include the sigmoid function, ReLU function, and tanh function. Their performance is related to factors such as network structure, learning rate, and regularization parameters. Commonly used evaluation metrics include accuracy, precision, and recall. Key parameters include the number of nodes in the input layer, hidden layer, and output layer, the learning rate, and the regularization parameters.

[0056] (4) Machine learning algorithms can be used to predict the relationship between the hit rate of the first steel section thickness and multiple factors. For example, decision trees, support vector machines, and other algorithms can be used for modeling to achieve multi-dimensional, non-linear prediction. In terms of parameter selection and model optimization, adjustments can be made through methods such as cross-validation and grid search. Commonly used evaluation metrics include accuracy, precision, and recall. Its main parameters include feature selection, model selection, and regularization parameters.

[0057] (5) Deep learning algorithms can be used to predict the nonlinear relationship between the hit rate of the first steel section thickness and multiple factors. For example, deep learning algorithms such as convolutional neural networks and recurrent neural networks can be used for modeling to achieve predictions of high-dimensional and complex relationships. In terms of parameter selection and model optimization, adjustments can be made through methods such as backpropagation and gradient descent. Commonly used evaluation metrics include accuracy, precision, and recall. Its main parameters include the number of network layers, the number of nodes, the activation function, and the learning rate.

[0058] Multiple algorithms are combined to calculate the correction coefficients of the fitted rolling force deviation, roll gap deviation, and speed deviation in the dynamic setting model.

[0059] In some embodiments, such as Figure 1 As shown, a method for controlling the first strip hit rate of a hot-rolled strip steel equipment also includes the following steps:

[0060] The first steel bar is prepared for threading, and the N sets of rolling mill mechanisms are started; the data of each first steel bar is input into the zero-value self-learning module for zero-value self-learning; dynamic settings and preset settings are performed; data is collected; the data is input into the correction calculation module for correction coefficient calculation, rolling force recalculation, roll gap recalculation, and speed recalculation; the correction coefficient is input into the dynamic setting module.

[0061] In some embodiments, a method for controlling the first strip hit rate of a hot-rolled strip steel equipment dynamically sets the first three stands and / or other stands out of N stands.

[0062] In some embodiments, based on the adaptive threading model, and combined with artificial intelligence algorithms such as big data and machine learning, the deviation correction coefficients of rolling force, bending roll force, and roll gap are calculated more accurately. Multiple first-strand hit deviations are collected and comprehensively considered to obtain a zero-value self-learning coefficient, which can achieve more than the set number of times F1 to F3. Figure 3 As shown, this greatly improves the hit rate of the first steel bar.

[0063] The adaptive strip threading model for finishing mills is built upon the finishing mill setting model, sharing the original data and setting parameters of the rolled piece. Specifically, during strip threading on the same stand, the finishing mill's secondary system, based on the stand's bite signal, completes the acquisition of measured values ​​required for adaptive strip threading calculations within a specified start and end time. The adaptive strip threading model compares the calculation results from the finishing mill setting model with the measured parameters, and completes the calculation of rolling parameters through validity verification. Then, it issues a procedure to the primary system to correct the roll gap and speed of subsequent stands. Due to the influence of inter-stand distance and rolling speed, the adaptive strip threading function cannot be deployed on every stand. Starting from the first finishing mill stand, after bite, the relevant parameters required for adaptive strip threading are prepared, and parameter corrections and dynamic settings are performed on subsequent stands.

[0064] The adaptive strip threading model for finishing mills refers to comparing the measured rolling force, roll gap, and speed parameters of the threading stand with the predicted parameters of the pre-set model during the finishing mill rolling process. After verifying the validity of the data, the rolling force, speed, and roll gap of the downstream stand are compensated. This eliminates the adverse effects of inaccurate finishing mill rolling parameter prediction caused by distortion of intermediate billet parameters or large deviations in the predicted parameters of the pre-set model. It ensures the thickness accuracy of the strip head at the exit of the final finishing mill stand, improves the prediction accuracy of the finishing mill model for rolling parameters, and ensures that the flow rate per second of each stand is matched, thereby stabilizing the rolling state and improving the actual product quality.

[0065] like Figure 3As shown, a portion of the thickness dimensions of the first strip produced in previous batches were selected. After data processing, the deviation between the measured thickness and the target thickness of the first strip, as well as the yield rate of the first strip, were obtained. Charts were created to compare the first strip data produced using the control method of this invention with those produced using the original method. After adopting the control method of this invention, the thickness deviation fluctuation is smaller than that produced using the original method, and the yield rate is significantly improved. Therefore, the control method of this invention can effectively improve the thickness accuracy of the strip head of the first strip, while also contributing to increased production output.

[0066] like Figure 4 As shown, the implementation and functional operation of the subject matter described in this specification can be carried out as follows: After the strip steel (i.e., the first strip) is heated to a certain temperature in a heating furnace, it is rolled to a certain thickness by roughing and finishing mills in sequence, and then bent into a steel coil by a coiling unit. During this process, an automated control system, including sensors, computers, controllers, communication equipment, etc., is used to monitor and control the entire production process. The control system can adjust the thickness and quality of the steel coil according to production data and instructions to ensure that the product meets requirements.

[0067] The control system mentioned in this specification includes, but is not limited to, digital electronic circuits, tangibly implemented computer software or firmware, computer hardware, including the structures disclosed in this specification and their equivalents, or combinations thereof. Embodiments of the subject matter described in this specification can be implemented as one or more computer programs, i.e., one or more modules of computer program instructions encoded on one or more tangible, non-transitory program carriers, for execution by a data processing unit or to control the operation of a data processing device.

[0068] Alternatively or additionally, program instructions may be encoded on artificially generated propagation signals, such as machine-generated electrical, optical, or electromagnetic signals, which are then generated as coded information to be transmitted to an appropriate receiver device executed by data processing equipment. The computer storage medium may be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or one or more combinations of the above.

[0069] This includes all kinds of devices, apparatuses, and machines for processing data, including, for example, programmable processors, computers, or multiprocessors or multiple computers. Devices may include special-purpose logic circuitry, such as FPGAs (Field-Programmable Gate Arrays) or ASICs (Application-Specific Integrated Circuits). In addition to hardware, devices may also include code that creates the execution environment for the associated computer programs, such as code constituting processor firmware, protocol stacks, database management systems, operating systems, or combinations thereof.

Claims

1. A method for controlling the first strip hit rate of a hot-rolled strip steel equipment, wherein the hot-rolled strip steel equipment includes N stands for strip threading and a first strip hit rate control system; characterized in that, The method for controlling the first strip hit rate of a hot-rolled strip steel equipment includes the following steps performed by the first strip hit rate control system: The zero-value self-learning module receives the data from each first steel bar sent by the first steel bar big data module, and finally obtains the zero-value self-learning value ΔL. i Send the zero-value self-learning value ΔL i To the preset module and the dynamic setting module; The deviation between the measured rolling force and the preset rolling force during the tape threading process of the receiving frame is corrected by the calculation module. The average value of the deformation resistance deviation Δki of the frame rolling force is estimated, and the frame rolling force is corrected based on the average value of the deformation resistance deviation Δki of the frame rolling force. The correction calculation module receives the deviation between the measured rolling force and the preset rolling force when the strip is threaded through the stand, estimates the thickness deviation Δhi of the roll gap between the stand, and corrects the roll gap between the stand based on the thickness deviation Δhi. The correction calculation module calculates the fitting correction coefficients and outputs them to the dynamic setting module; The zero-value self-learning value ΔLi is determined according to equation (I): ΔL i =m i-1 *f(i-1)+m i-2 *f(i-2)+m i-3 *f(i-3)+……m i-n In the formula *f(in)(I), ΔLi represents the self-learning value set to zero for the i-th time; f(i-1) represents the value after big data processing of the first steel support for the (i-1)-th time; m i-1 It represents the weighting coefficient of the first steel data processing value in the (i-1)th iteration relative to the zero self-learning value in the i-th iteration.

2. The method for controlling the first strip hit rate of a hot-rolled strip steel equipment according to claim 1, characterized in that, The average value of the deformation resistance deviation Δki is calculated according to Equation II: Δki = ΔFi / Fi (II) In the formula, ΔFi represents the deviation between the measured rolling force and the preset rolling force of the stand; Fi represents the preset rolling force of the stand; i represents the stand number, i = 1, 2, …, n.

3. The method for controlling the first strip hit rate of a hot-rolled strip steel equipment according to claim 1, characterized in that, The thickness deviation Δhi of the frame is determined according to Equation III: Δhi = ΔSi + ΔFi / Mi (III) In the formula, ΔSi is the deviation between the measured roll gap and the predicted roll gap of the mill stand; ΔFi is the deviation between the measured rolling force and the preset rolling force of the mill stand; and Mi is the mill stand stiffness.

4. The method for controlling the first strip hit rate of a hot-rolled strip steel equipment according to claim 1, characterized in that, The method for controlling the first strip hit rate of a hot-rolled strip steel equipment further includes the following steps performed by the first strip hit rate control system: The correction calculation module recalculates the rack speed Vpi based on the changed rack strip exit thickness h^i and forward slip coefficient fi, and compares it with the set speed Vmi. The difference is used for rack speed correction.

5. The method for controlling the first strip hit rate of a hot-rolled strip steel equipment according to claim 1, characterized in that, The first steel data module receives the data from the first steel section, collects, stores, and analyzes the data, extracts key indicators, and uses machine learning and deep learning technologies to establish a thickness hit rate control model. The processed data is then stored and sent in a fixed format.

6. The method for controlling the first strip hit rate of a hot-rolled strip steel equipment according to claim 1, characterized in that, It also includes an algorithm library interface, which is configured as follows: The correction coefficients for the fitted rolling force deviation, roll gap deviation, and speed deviation in the dynamic setting module are calculated using a combination of linear regression, nonlinear regression, neural network, machine learning, and deep learning algorithms.

7. The method for controlling the first strip hit rate of a hot-rolled strip steel equipment according to claim 1, characterized in that, The method for controlling the first strip hit rate of a hot-rolled strip steel equipment includes the following steps: The first steel bar is prepared for threading, and the N sets of rolling mill mechanisms are started; the data of each first steel bar is input into the zero-value self-learning module for zero-value self-learning; dynamic settings and preset settings are performed; data is collected; the data is input into the correction calculation module for correction coefficient calculation, rolling force recalculation, roll gap recalculation, and speed recalculation; the correction coefficient is input into the dynamic setting module.

8. The method for controlling the first strip hit rate of a hot-rolled strip steel equipment according to claim 1, characterized in that, The dynamic settings include the first three racks out of N racks and / or other racks.

9. A first strip hit rate control system for a hot-rolled strip steel equipment, the system comprising at least one processor; and a memory storing instructions that, when executed by the at least one processor, implement the steps of the method according to any one of claims 1-8.

Images