A Smart Control Method for Surface PC Value of Cold-Rolled Strip Steel Based on Optimization Algorithm

Abstract

This invention discloses an intelligent control method for the PC value of cold-rolled strip steel surface based on an optimization algorithm. By analyzing the mechanism of PC value transfer during cold rolling, and through model optimization algorithm calculation, the rolling process parameters are intelligently adjusted to achieve effective control of the PC value of the strip steel surface. The method described in this invention can be directly imported into the secondary control model of rolling for online parameter optimization, and can be directly applied to cold continuous rolling mill production, promoting the intelligent transformation of enterprises and improving control accuracy and stability.

Description

Technical Field

[0001] This invention belongs to the field of strip steel surface technology, and relates to an intelligent control method for the PC value of cold-rolled strip steel surface based on optimization algorithm. Background Technology

[0002] In the steel industry, cold-rolled strip steel is an important material, and its surface quality is crucial to the final performance of the product. The surface PC value, as a key parameter for measuring the surface morphology of strip steel, directly affects its performance in subsequent processing and use, such as coating adhesion, surface smoothness, corrosion resistance, and aesthetics. Therefore, controlling the surface PC value is of great significance for improving product quality and meeting customer needs. However, controlling the surface PC value presents considerable technical challenges. First, the cold rolling process involves multiple variables, such as rolling speed, rolling force, lubrication conditions, and cooling rate. These variables interact, complicating the control of the surface PC value. Furthermore, unavoidable process fluctuations and equipment wear during production also affect the surface PC value, making stable and precise control even more difficult.

[0003] Prior to this invention, the patents related to this invention mainly included the following:

[0004] (1) Patent CN113943899A discloses a method for controlling the surface morphology of cold-rolled deep-drawing steel, including smelting, continuous casting, hot continuous rolling, pickling continuous rolling, and continuous annealing processes. The chemical composition and mass percentage of the steel are: C: ≤0.0030%, Si: ≤0.030%, Mn: ≤0.30%, P: ≤0.015%, S: ≤0.015%, Als: 0.020~0.060%, Ti: ≤0.080%, Nb: 0.003~0.008%, N: ≤0.0040%, with the remainder being Fe and unavoidable impurities. This invention obtains a cold-rolled deep-drawing steel with stable surface morphology by controlling key process parameters such as smelting chemical composition, hot rolling lubrication state, surface morphology of cold-rolled leveling rolls, and elongation. In addition to meeting the forming performance of deep drawing, it can achieve the high surface quality requirements of automobiles without intermediate coating. In the acid continuous rolling process, the surface roughness Ra of the work rolls on the cold rolling stands F1 and F2 is 0.5~1.0 μm, the surface roughness Ra of the work rolls on the F3 and F4 stands is 0.3~0.6 μm, and the surface roughness Ra of the work rolls on the F5 stand is 3.0~3.5 μm. The surface roughness Ra of the steel strip after cold continuous rolling is 0.6~1.2 μm, and the peak value Rpc is 90~100 pieces / cm. This method does not control the PC value of the work rolls and cannot optimize the surface parameters of the work rolls for the target PC setting.

[0005] (2) Patent CN114032467A discloses a cold-rolled high-strength steel plate and its preparation method. The chemical composition of the cold-rolled high-strength steel plate billet, by mass percentage, is: C: 0.1~0.2%, Mn: 1.5~2.5%, S: ≤0.005%, P≤0.015%, Si: 0.5~2.0%, Als: 0.015~0.060%, N≤0.005%, with the balance being Fe and unavoidable impurities. The process flow of the preparation method is as follows: smelting and casting, hot rolling and coiling, pickling and cold rolling, continuous annealing, and leveling. The phosphate film layer on the surface of cold-rolled automotive steel sheet prepared by this invention has a dense structure, uniform crystallization, granular appearance, and low porosity, exhibiting high paintability. It effectively solves the problem of poor phosphate performance in existing high-formability cold-rolled high-strength steel sheets. After leveling, the surface roughness Ra value of the strip is controlled to be 1.0~1.3 μm, the Rpc value to be 80~130, and the surface energy to be 1.2~1.8 J / m. 2 This method mainly controls surface parameters through the leveling process, but does not impose control requirements on the process parameters of the cold rolling process. Furthermore, the method focuses on designing target values ​​for the surface parameters of the strip without providing detailed instructions on how to achieve these target values.

[0006] (3) Patent CN114231822A discloses a method for improving the paintability of cold-rolled automotive steel sheets. The process flow of the method is as follows: smelting and casting, hot rolling and coiling, pickling and cold rolling, continuous annealing, leveling and oiling, degreasing and pickling, surface conditioning and phosphating. The phosphating film layer on the surface of the cold-rolled automotive steel sheets prepared by this method has a dense structure, uniform crystallization, and granular shape with an average grain size of 2~3 μm and low porosity, resulting in high paintability. This effectively solves the problem of poor paintability of cold-rolled automotive steel sheets in existing methods. In the leveling and oiling process, the surface roughness Ra value of the strip after leveling is controlled to be 0.7~1.3 μm and the Rpc value to be 80~130. This method focuses on controlling the surface parameters of the strip in the leveling process, but only proposes target values ​​for the surface parameters without proposing specific control measures. Summary of the Invention

[0007] The purpose of this invention is to provide an intelligent control method for the surface PC value of cold-rolled strip steel based on an optimization algorithm. By performing mechanism analysis and mathematical modeling on the cold-rolled roll transfer, which has a significant impact on the surface PC value, the rolling process parameters are intelligently adjusted through the model optimization algorithm to achieve precise control of the surface PC value of the strip steel.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] (a) Set the target PC value RPcTi , i Indicates the number of rolling mill stands;

[0010] (b) Collect initial parameters: roll PC value RPc wi ;

[0011] (c) Initialize the learning rate α and tolerance error β ;

[0012] (d) Define the loss function ;

[0013] (e) Initialize the iteration counter ( t =0);

[0014] (f) Calculate the initial PC value of the strip at the outlet of each stand. RPc i ;

[0015] (g) Calculate the total loss through step (d). L ;

[0016] (h) For each RPc wi Calculate the partial derivatives ( ), and update the PC values ​​of each stand's rolls;

[0017] (i) Update the PC value of the strip at the exit of each rack RPc i ;

[0018] (j) Repeat steps (g)~(i);

[0019] (k) Output the optimized roll PC value RPc wi .

[0020] Furthermore, in step (c), the learning rate needs to be balanced. If the learning rate is too small, convergence will be slow; if the learning rate is too large, the loss will oscillate or even increase. After comprehensively analyzing multiple initializations of the learning rate and tolerance error, the final initial learning rate is determined. α and tolerance error β， Make it satisfied β / α =100 is a functional relationship.

[0021] Furthermore, in step (d), RPc i From the formula The calculation shows that, in the formula a 0 = 168.85 ± 4.93 a 1 = 50.25 ± 6.12, a2 = -18.79 ± 7.48 a 3 = 25.26 ± 1.28.

[0022] Furthermore, in step (j), until () is satisfied L < β The calculation stops when the maximum number of iterations is reached.

[0023] This invention provides an intelligent control method for the surface PC value of cold-rolled strip steel based on an optimization algorithm. By analyzing the mechanism of PC value transfer during cold rolling, the method optimizes the parameters of the PC value of each stand's rolls using an optimization algorithm, achieving effective control of the surface PC value without the need for other auxiliary equipment or extensive modifications. The method described in this invention can be directly imported into the secondary control model of rolling for online parameter optimization, and can be directly applied to cold continuous rolling mill production, promoting the intelligent transformation of enterprises and improving control accuracy and stability. Attached Figure Description

[0024] Figure 1 Flowchart of intelligent control method for PC value on surface of cold-rolled strip steel. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The following description of exemplary embodiments of the experimental method is merely illustrative and is in no way intended to limit the present invention or its application or use. 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.

[0026] Example 1:

[0027] (a) Set the target PC value RPc Ti ={140,215,220,122,190} i ={1,2,3,4,5};

[0028] (b) Collect initial parameters: roll PC value RPc wi ={125,153,143,168,107} i ={1,2,3,4,5};

[0029] (c) Initialize the learning rate α =0.1 and tolerance error β =10;

[0030] (d) Define the loss function ;

[0031] (e) Initialize the iteration counter ( t =0);

[0032] (f) Calculate the initial PC value of the strip at the outlet of each stand. RPc i ={133,202,213,115,173} i ={1,2,3,4,5};

[0033] (g) Calculate the total loss L =605;

[0034] (h) For each RPc wi Calculate the partial derivatives = {3.5, -5.02, 1.89, -2.09, -6.23}. i ={1,2,3,4,5}; and update each rack. RPc wi Values ​​= {124.64, 153.50, 142.81, 168.21, 107.62} i ={1,2,3,4,5};

[0035] (i) Update the PC value of the strip at the exit of each rack RPc i ={126.05,196.13,216.35,121.09,168.54} i ={1,2,3,4,5};

[0036] (j) Repeat steps (g) to (i) until the following condition is met: L < β );

[0037] (k) Output the optimized roll PC value RPc wi ={178,192,196,173,136} i ={1,2,3,4,5}.

[0038] Example 2:

[0039] (a) Set the target PC value RPc Ti ={130,135,150,140,160} i ={1,2,3,4,5};

[0040] (b) Collect initial parameters: roll PC value RPcwi ={120,130,150,160,155} i ={1,2,3,4,5};

[0041] (c) Initialize the learning rate α =1 and tolerance error β =100;

[0042] (d) Define the loss function ;

[0043] (e) Initialize the iteration counter ( t =0);

[0044] (f) Calculate the initial PC value of the strip at the outlet of each stand. RPc i ={118,152,210,155,149} i ={1,2,3,4,5};

[0045] (g) Calculate the total loss L =4379;

[0046] (h) For each RPc wi Calculate the partial derivatives = {-0.16, 5.86, -3.32, -6.07, -5.78}. i ={1,2,3,4,5}; and update each rack. RPc wi Values ​​= {118.60, 124.48, 153.32, 166.07, 160.78} i ={1,2,3,4,5};

[0047] (i) Update the PC value of the strip at the exit of each rack RPc i ={118.60,124.48,197.05,126.86,152.52} i ={1,2,3,4,5};

[0048] (j) Repeat steps (g) to (i) until the following condition is met: L < β );

[0049] (k) Output the optimized roll PC value RPc wi ={125,164,161,163,159} i ={1,2,3,4,5}.

Claims

1. A method for intelligent control of PC value on the surface of cold-rolled strip steel based on optimization algorithms, characterized in that, Specifically include, (a) Set the target PC value RPc Ti , i Indicates the number of rolling mill stands; (b) Collect initial parameters: roll PC value RPc wi ; (c) Initialize the learning rate α and tolerance error β ; (d) Define the loss function , RPc i From the formula The calculation shows that, in the formula a 0 = 168.85 ± 4.93 a 1 = 50.25 ± 6.12, a 2 = -18.79 ± 7.48 a 3 = 25.26 ± 1.28; (e) Initialize the iteration counter. t =0; (f) Calculate the initial PC value of the strip at the exit of each stand. RPc i ; (g) Calculate the total loss through step (d). L ; (h) for each RPc wi Calculate the partial derivatives And update the PC values ​​of each stand's rolls. RPc wi The calculation formula is ; (i) Update the PC value of strip steel at the exit of each rack RPc i ; (j) Repeat steps (g) to (i) until the condition is met. L < β， The calculation may stop when the maximum number of iterations is reached; (k) Output the optimized roll PC value RPc wi .

2. The intelligent control method for PC value of cold-rolled strip steel based on optimization algorithm according to claim 1, characterized in that, In step (c), the initial learning rate is determined. α and tolerance error β satisfy β / α =100 is a functional relationship.

Images