Full-automatic closed-loop control method for high-uniformity poor-thickness zinc layer

Abstract

The invention relates to the technical field of steel smelting system control, and particularly discloses a full-automatic closed-loop control method for a high-uniformity poor-thickness zinc layer, which comprises the following steps of: collecting production data of a poor-thickness zinc layer product on a production line; establishing a multi-task learning AI model with a multi-gate hybrid expert network structure, and inputting parameters into the AI model; carrying out system training on the AI model, wherein the system training comprises cross validation training, discarding learning training and hyper-parameter training; when the strip steel head welding seam position passes through the zinc pot, the AI model is used for conducting technological parameter technology control, and an air knife control system is issued for execution; the difference between the zinc layer and the order target zinc layer is analyzed online and measured in real time; analyzing the transverse uniformity of the zinc layer on line; the steps are continued until the production of the coil of strip steel is finished; according to the invention, feedforward control and feedback control of the zinc layer of the differential-thickness zinc-layer galvanized plate are realized, closed-loop automatic control of the through plate of the differential-thickness zinc-layer galvanized plate is realized, the production efficiency is obvious, and the quality is obviously improved.

Description

Technical Field

[0001] This invention relates to the field of steel smelting system control technology, specifically to a fully automatic closed-loop control method for a thick zinc layer with high uniformity. Background Technology

[0002] Currently, with the rapid development of industries such as automobiles and home appliances, market capacity is gradually becoming excessive, and competition is becoming increasingly fierce. Major manufacturers are trying various cost-reduction and efficiency-enhancing strategies. As a key raw material for these industries, galvanized steel sheets have naturally become a focus. Customers have demanded both reduced raw material costs and guaranteed corrosion resistance. Therefore, differential thickness galvanized steel sheets are gaining popularity among manufacturers due to their excellent corrosion resistance, weldability, and lower cost. Furthermore, with the rapid development of hot-dip galvanized steel sheets, surface requirements are becoming increasingly stringent, especially regarding the uniformity of the zinc layer. Improving the uniformity of the zinc layer is also a cost-reduction measure for steel mills.

[0003] Currently, the production of galvanized sheets with varying thicknesses primarily relies on manual adjustment of the distance and pressure of the air knife to achieve the target zinc coating. This traditional method is prone to errors due to manual adjustment, leading to inaccurate zinc coating control. Furthermore, the thickness in the strip width direction is easily affected by the sheet shape, causing thickness fluctuations and impacting zinc coating uniformity. While closed-loop zinc coating control technology is widely promoted both domestically and internationally for equal-thickness zinc coatings, the unique characteristics of varying-thickness zinc coatings, involving the adjustment of multiple variables and logical calculations, present significant challenges. If the technology is not mature enough, defects such as zinc spots and discrepancies in zinc coating weight can easily occur. Moreover, existing technologies, whether for equal-thickness or varying-thickness zinc coatings, lack quantitative analysis and real-time compensation adjustment mechanisms for the lateral uniformity of the zinc coating.

[0004] Invention patent CN 105316570A, "A Hot-Dip Galvanized Steel Sheet with Differential Thickness Zinc Coating and its Production Method," mainly uses conventional methods to obtain the target differential thickness zinc coating through the galvanizing process and adjustment of the air knife distance. This method relies on manual experience for adjustment, making it difficult to achieve closed-loop control of the zinc coating and easily leading to inaccurate zinc coating control. Furthermore, this method relies solely on adjusting the zinc coating thickness through air knife distance and pressure, making it difficult to achieve high uniformity control of the zinc coating, especially regarding zinc coating thickness fluctuations in the width direction of the strip.

[0005] The invention patent CN 114959536B, "A Method and Device for Controlling the Thickness of a Galvanized Coating," primarily solves the problem of inconsistent and uneven zinc coating weight control caused by human error by acquiring zinc coating control parameters from a database and applying them to the current steel coil. However, this method relies on database maintenance and can only achieve feedforward control of the zinc coating under the premise that historical process conditions, order specifications, and zinc coating are consistent. It cannot handle zinc coating control problems under new working conditions or abnormal situations. Furthermore, this method cannot perform closed-loop feedback control based on the measurement results of a zinc coating thickness gauge. In addition, this method cannot address the problem of poor transverse uniformity of the strip steel.

[0006] The invention patent CN 110565039B, "A Method for Controlling Zinc Layer Thickness in Hot-Dip Galvanizing Units," primarily achieves automatic control of the zinc layer by collecting parameters of the zinc layer control unit, judging the unit's operating conditions, finding the closest set of process parameters in the database, and then adjusting them. However, this method can only achieve feedforward control of the zinc layer under the premise that historical process conditions, order specifications, and zinc layer consistency are met. It cannot handle zinc layer control problems under new operating conditions or abnormal situations. Furthermore, this method cannot perform closed-loop feedback control based on the measurement results of a zinc layer thickness gauge, and it cannot address the problem of poor transverse uniformity of the strip steel.

[0007] Therefore, it is necessary to design a fully automatic closed-loop control method for high uniformity of zinc layer thickness to solve the problem that existing galvanizing units cannot achieve high uniformity control of zinc layer thickness when rolling galvanized sheets with zinc layer thickness, resulting in poor product quality and low production efficiency. Summary of the Invention

[0008] To address the problems existing in the prior art, the purpose of this invention is to provide a fully automatic closed-loop control method for high uniformity differential zinc layer thickness. Differential zinc layer galvanized sheet refers to galvanized sheet with inconsistent target zinc layer control on the upper and lower surfaces of the strip, while equal thickness galvanized sheet refers to galvanized sheet with consistent target zinc layer control on the upper and lower surfaces of the strip.

[0009] The technical solution adopted by this invention to solve its technical problem is: a fully automatic closed-loop control method for a thick zinc layer with high uniformity, comprising the following steps:

[0010] S1. Collect production data for products with differentially thick zinc coatings on the production line;

[0011] S2. Establish a multi-task learning AI model with a multi-subject hybrid expert network structure, and input parameters into the AI ​​model;

[0012] S3. Systematically train the AI ​​model, including cross-validation training, dropout learning training, and hyperparameter training.

[0013] S4. Connect the trained AI model to the on-site PLC network. When the strip passes through the zinc pot at the weld seam of the strip head, use the AI ​​model to control the process parameters and send the air knife control system to execute.

[0014] S5. Online analysis and real-time measurement of the difference between the zinc layer and the target zinc layer in the order. During the adjustment process, the thickness of the zinc layer on the upper surface and the zinc layer on the lower surface are monitored and the Y value is calculated respectively. When Y≥2%, the adjustment begins.

[0015] S6. Analyze the transverse uniformity of the zinc layer online and make adjustments based on the analysis results;

[0016] S7. The above steps continue until the production of this coil of strip steel is completed.

[0017] Specifically, the production data for the differential thickness zinc layer in step S1 includes, but is not limited to, the current unit production speed, steel type, specifications, and parameters of the target zinc layer, as well as the process parameters for predicting the pressure, height, distance, tilt value, and displacement of the air knife.

[0018] Specifically, the input parameters of the AI ​​model in step S2 include strip thickness, strip width, target zinc layer thickness on the upper and lower surfaces, and production line speed, while the output parameters are upper and lower surface pressure, upper and lower surface distance, air knife height, and tilt displacement.

[0019] Specifically, the difference between the measured zinc layer and the target zinc layer in step S5 is adjusted according to Formula 1:

[0020] In the formula: Y is the average total zinc layer deviation, in %; X is the current average zinc layer thickness; X0 is the target value of zinc layer thickness; and Δp is the pressure adjustment amount.

[0021] Specifically, the lateral uniformity of the zinc layer in step S6 is analyzed and controlled based on the uniformity of the zinc layer on the lower surface.

[0022] Specifically, the lateral uniformity of the zinc layer is defined as the ratio of the lateral deviation of the zinc layer in the strip to... At that time, adjust the position of the air knife frame according to Formula 2. Formula 2:

[0023] In the formula: d is the displacement value of the air knife frame, in mm; d0 is the initial displacement value of the air knife frame, in mm; v is the strip speed, in m / min; σ is the standard deviation of the transverse zinc coating, in g / m. 2 X0 represents the target weight of the zinc coating, in g / m³. 2 lateral deviation of zinc layer unit%.

[0024] The present invention has the following beneficial effects:

[0025] The fully automatic closed-loop control method for high uniformity zinc layer thickness variation designed in this invention realizes feedforward control and feedback control of zinc layer in galvanized sheets with varying zinc layer thickness, and achieves closed-loop automatic control of the through-plate of galvanized sheets with varying zinc layer thickness. No manual intervention is required, resulting in significant production efficiency and quality improvement.

[0026] The fully automatic closed-loop control method for high uniformity differential zinc layer designed in this invention significantly improves the uniformity of longitudinal control of zinc layer in galvanized sheet, greatly reduces zinc layer consumption, and has significant economic benefits.

[0027] The fully automatic closed-loop control method for high uniformity differential zinc layer designed in this invention introduces a zinc layer transverse uniformity evaluation standard for the first time, and uses the air knife frame displacement as an independent control variable to realize closed-loop control of the transverse zinc layer of the strip steel. This significantly improves the uniformity of the zinc layer of the galvanized sheet, resulting in a significant improvement in product quality, a significant reduction in zinc consumption, and significant economic benefits.

[0028] The fully automatic closed-loop control method for high uniformity differential zinc layer designed in this invention has been applied to the air knife model control of galvanized sheets with equal zinc layer thickness. Attached Figure Description

[0029] Figure 1 This is a flowchart of a fully automatic closed-loop control method for a thick zinc layer with high uniformity. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0031] like Figure 1 As shown, a fully automatic closed-loop control method for a thick zinc layer with high uniformity includes the following steps:

[0032] 1. Collect production data for products with differential thickness zinc coatings on the production line; the production data for products with differential thickness zinc coatings includes, but is not limited to, the current unit production speed, steel type, specifications and parameters of the target zinc coating, as well as the process parameters for predicting the pressure, height, distance, tilt value and displacement of the air knife.

[0033] 2. Establish a multi-task learning AI model with a multi-gate hybrid expert (MMoE) network structure, and input parameters into the AI ​​model. The input parameters of the AI ​​model include strip thickness, strip width, target zinc layer thickness of upper and lower surfaces, and production line speed. The output parameters are upper and lower surface pressure, upper and lower surface distance, air knife height, and tilt displacement.

[0034] 3. Conduct systematic training for the AI ​​model, including cross-validation training, dropout learning training, and hyperparameter training.

[0035] 4. Connect the trained AI model to the on-site PLC network. When the strip passes through the zinc pot at the weld seam of the strip head, use the AI ​​model to control the process parameters and send the air knife control system to execute.

[0036] 5. Online analysis of the difference between the measured zinc layer and the target zinc layer in the order. The difference between the measured zinc layer and the target zinc layer in the order is adjusted according to Formula 1. Formula 1:

[0037] In the formula: Y is the average total zinc layer deviation, in %; X is the current average zinc layer thickness; X0 is the target value of zinc layer thickness; and Δp is the pressure adjustment amount.

[0038] The adjustment process simultaneously monitors the thickness of the zinc layer on the upper and lower surfaces and calculates the Y value for each. Adjustment begins when Y ≥ 2%.

[0039] 6. Online analysis of zinc layer lateral uniformity; the analysis and control logic for zinc layer lateral uniformity is based on the uniformity of the zinc layer on the lower surface.

[0040] When the transverse deviation of the zinc coating on the strip is compared to At that time, adjust the position of the air knife frame according to Formula 2. Formula 2:

[0041] In the formula: d is the displacement value of the air knife frame, in mm; d0 is the initial displacement value of the air knife frame, in mm; v is the strip speed, in m / min; σ is the standard deviation of the transverse zinc coating, in g / m. 2 X0 represents the target weight of the zinc coating, in g / m³. 2 lateral deviation of zinc layer unit%.

[0042] 7. The above steps continue until the production of this coil of strip steel is completed.

[0043] This invention accurately identifies the basic information of the steel coil before production, including data such as the target zinc layer, unit speed, strip thickness, and width. A multi-disciplinary hybrid expert AI model is established to accurately predict parameters such as the initial height, pressure, spacing, and frame tilt displacement of the air knife, achieving feedforward control. During production, an online zinc layer thickness gauge monitors the weight and distribution of the zinc layer in real time, and closed-loop feedback control is implemented based on the results. By comparing the actual measured zinc layer of the strip with the target zinc layer, the average total zinc layer deviation is calculated. Based on the results, the air knife pressure and distance are further precisely adjusted to ensure the accuracy of zinc layer control. Furthermore, the transverse zinc layer deviation ratio of the strip is calculated, and the air knife frame displacement is dynamically adjusted based on the calculation results to effectively ensure the uniformity of the transverse zinc layer of the strip. This ultimately achieves fully automatic closed-loop control of the air knife.

[0044] Multi-mode hybrid expert AI model refers to the establishment of a multi-task learning model with MMoE network structure, accompanied by system training functions such as cross-validation, dropout learning, and hyperparameters, to identify the current unit production speed, steel type, specifications, and target zinc layer, and to judge and predict process parameters such as pressure, height, distance, tilt value, and displacement of the air knife, thereby realizing automatic pre-control of air knife parameters.

[0045] By integrating multiple hybrid expert AI models into the production line, the data from the online zinc layer thickness gauge on the production line is processed in depth, and the air knife parameters are automatically adjusted in a closed-loop feedback manner based on the processing results.

[0046] Example 1:

[0047] In one production run, the strip steel specifications were 0.3*1250mm, the machine speed was 120m / min, and the ordered zinc coating was 40g / m² on the top surface. 2 The lower surface has a density of 60 g / m². 2 The AI ​​model outputs the following initial air knife parameters: air knife height H is 200mm, the distance between the upper and lower air knives S is 7mm, the air knife frame displacement d is 1mm, the upper air knife pressure is 300mbar, and the lower air knife pressure is 220mbar.

[0048] The online zinc coating thickness gauge measured an average zinc coating thickness of 40.5 g / m² on the upper surface. 2 The average zinc layer on the lower surface is 60.6 g / m². 2 The zinc coating weight on the lower operating side is 61.2 g / m². 2 The zinc coating weight on the lower drive side is 60.2 g / m². 2 Online analysis yielded σ = (61.2 - 60.2) = 1 g / m³ 2 Z was 1.67%, and the trigger d was adjusted to 1.43 mm. The final average zinc coating on the upper surface of the strip was 40.5 g / m. 2 The average zinc layer on the lower surface is 60.6 g / m².2 The zinc coating weight on the lower operating side is 60.8 g / m². 2 The zinc coating weight on the lower drive side is 60.4 g / m². 2 σ=(60.8-60.4)=0.4g / m 2 Z is 0.67%, and the air knife parameters are kept stable until the production of the roll is completed.

[0049] Example 2:

[0050] In one production run, the strip steel specifications were 1.0*1800mm, the machine speed was 140m / min, and the ordered zinc coating was 50g / m² on the top surface. 2 The lower surface has a density of 250 g / m². 2 The AI ​​model outputs the following initial air knife parameters: air knife height H = 180mm, upper air knife distance S = 7mm, lower air knife distance S = 12mm, air knife frame displacement d = 0.8mm, upper air knife pressure = 350mbar, and lower air knife pressure = 90mbar.

[0051] The online zinc coating thickness gauge measured an average zinc coating thickness of 50.8 g / m² on the upper surface. 2 The average zinc layer on the lower surface is 251.9 g / m². 2 The zinc coating weight on the lower operating side is 253.6 g / m². 2 The zinc coating weight on the lower drive side is 250.2 g / m². 2 Online analysis yielded σ = (253.6 - 250.2) = 3.4 g / m³ 2 Z was 1.36%, and the trigger d was adjusted to 2.18 mm. The final average zinc coating on the upper surface of the strip was 50.8 g / m². 2 The average zinc layer on the lower surface is 251.9 g / m². 2 The zinc coating weight on the lower operating side is 252.4 g / m². 2 The zinc coating weight on the lower transmission side is 251.4 g / m². 2 σ=(252.4-251.4)=1g / m 2 Z is 0.4%, and the air knife parameters are kept stable until the production of the roll is completed.

[0052] Example 3:

[0053] In one production run, the strip steel specifications were 2.5*1500mm, the machine speed was 60m / min, and the ordered zinc coating was 40g / m² on the top surface. 2 The lower surface has a density of 80 g / m². 2The AI ​​model outputs the following initial parameters for the air knife: air knife height H = 160mm, upper air knife distance S = 7mm, lower air knife distance S = 10mm, air knife frame displacement d = 0.5mm, upper air knife pressure = 240mbar, and lower air knife pressure = 140mbar.

[0054] The online zinc coating thickness gauge measured an average zinc coating thickness of 41.6 g / m² on the upper surface. 2 The average zinc layer on the lower surface is 82.4 g / m². 2 The zinc coating weight on the lower operating side is 84.2 g / m². 2 The zinc coating weight on the lower drive side is 80.6 g / m². 2 Online analysis yielded σ = (84.2 - 80.6) = 3.6 g / m³ 2 Z was set to 4.5%, and the trigger diameter (d) was adjusted to 3.05 mm. The final average zinc coating on the upper surface of the strip was 41.6 g / m². 2 The average zinc layer on the lower surface is 82.4 g / m². 2 The zinc coating weight on the lower operating side is 83.4 g / m². 2 The zinc coating weight on the lower transmission side is 81.4 g / m². 2 σ=(82.6-82.2)=0.4g / m 2 Z is 0.5%, and the air knife parameters are kept stable until the production of the roll is completed.

[0055] This invention is not limited to the above-described embodiments. Anyone should know that any structural changes made under the guidance of this invention, and any technical solutions that are the same as or similar to this invention, fall within the protection scope of this invention.

[0056] The technologies, shapes, and structures not described in detail in this invention are all known technologies.

Claims

1. A fully automatic closed-loop control method for a thick zinc layer with high uniformity, characterized in that, Includes the following steps: S1. Collect production data for products with differentially thick zinc coatings on the production line; S2. Establish a multi-task learning AI model with a multi-subject hybrid expert network structure, and input parameters into the AI ​​model; S3. Systematically train the AI ​​model, including cross-validation training, dropout learning training, and hyperparameter training. S4. Connect the trained AI model to the on-site PLC network. When the strip passes through the zinc pot at the weld seam of the strip head, use the AI ​​model to control the process parameters and send the air knife control system to execute. S5. Online analysis and real-time measurement of the difference between the zinc layer and the target zinc layer in the order. During the adjustment process, the thickness of the zinc layer on the upper surface and the zinc layer on the lower surface are monitored and the Y value is calculated respectively. When Y≥2%, the adjustment begins. S6. Analyze the transverse uniformity of the zinc layer online and make adjustments based on the analysis results; S7. The above steps continue until the production of this coil of strip steel is completed.

2. The fully automatic closed-loop control method for high-uniformity, poor-thickness zinc layers according to claim 1, characterized in that, The production data for the differential thickness zinc layer in step S1 includes, but is not limited to, the current unit production speed, steel type, specifications, and parameters of the target zinc layer, as well as the process parameters for predicting the pressure, height, distance, tilt value, and displacement of the air knife.

3. The fully automatic closed-loop control method for a high-uniformity, poor-thickness zinc layer according to claim 1, characterized in that, The input parameters of the AI ​​model in step S2 include strip thickness, strip width, target zinc layer thickness on the upper and lower surfaces, and production line speed. The output parameters are pressure on the upper and lower surfaces, distance between the upper and lower surfaces, air knife height, and tilt displacement.

4. The fully automatic closed-loop control method for a high-uniformity, poor-thickness zinc layer according to claim 1, characterized in that, The difference between the measured zinc layer and the target zinc layer in step S5 is adjusted according to Formula 1, which is: In the formula: Y is the average total zinc layer deviation, in %; X is the current average zinc layer thickness; X0 is the target value of zinc layer thickness; and Δp is the pressure adjustment amount.

5. The fully automatic closed-loop control method for a high-uniformity, poor-thickness zinc layer according to claim 1, characterized in that, The lateral uniformity of the zinc layer in step S6 is analyzed and controlled based on the uniformity of the zinc layer on the lower surface.

6. The fully automatic closed-loop control method for a thick zinc layer with high uniformity as described in claim 5, characterized in that, The lateral uniformity of the zinc layer is defined as the ratio of the lateral deviation of the zinc layer in the strip steel to... At that time, adjust the position of the air knife frame according to Formula 2. Formula 2: In the formula: d is the displacement value of the air knife frame, in mm; d0 is the initial displacement value of the air knife frame, in mm; v is the strip speed, in m / min; σ is the standard deviation of the transverse zinc coating, in g / m. 2 X0 represents the target weight of the zinc coating, in g / m³. 2 lateral deviation of zinc layer unit%.

Images