Automatic control method for camber of rough rolling intermediate bloom

Abstract

The application discloses a kind of rough rolling intermediate blank camber automatic control method, the method is established rough rolling process parameter table, genetic learning table, rough rolling R1, R2 rolling mill leveling value calculation model, in rough rolling process, the center line deviation point data of intermediate blank rolling stock is collected and handled, and the leveling value is calculated in combination with plan data, pre-computed data, process parameters, genetic learning correction value etc., and the calculation result is issued to the executing machine to carry out rolling mill control, to realize the automatic accurate adjustment of rolling mill roll gap deviation, improve the problem of poor shape caused by camber.It is innovative in core to construct process parameter table according to process experience and requirement to provide different parameters for different specifications of rolling stock for model calculation, and to compensate and correct the leveling setting by using genetic learning table, significantly reduce the standard deviation of center line deviation, improve the automatic leveling rate to more than 90%, effectively reduce production failure and improve productivity.

Description

Technical Field

[0001] This invention relates to the field of metal rolling technology, specifically to an automatic control method for the sickle bend of intermediate slabs in roughing mills. More particularly, it relates to an automatic control technology for the sickle bend of intermediate slabs in roughing mills. Background Technology

[0002] When a rolled piece bends to the left or right after exiting the rolls, it is called a sickle bend. A sickle bend can seriously affect the centering of the rolled piece in downstream passes or stands, leading to asymmetrical shape defects such as wedges or single-sided waves, which significantly disrupts the stability of the rolling process.

[0003] Currently, many steel production enterprises face the following technical problem: when severe camber occurs in rolled steel, there is no effective automatic camber control technology that can accurately calculate and correct it. Therefore, it is necessary to rely on operators to continuously observe the shape changes of the rolled piece during the rolling process and adjust the rolling process based on experience. This manual adjustment method is not only time-consuming and labor-intensive, but also prone to operational errors and inconsistencies, which in turn affect the rolling quality and production efficiency.

[0004] Patent document CN111842507B discloses a method for controlling the centerline deviation of a slab. This method utilizes the centerline deviation information of the slab detected by the width gauge at the exit of the roughing mill, and automatically adjusts the oil column deviation value of the R2 mill to achieve automatic control of the centerline deviation of the slab at the exit of the last pass of the R2 mill. However, this method has the following drawbacks: it cannot calculate different parameters for rolled materials with different specifications, thus limiting its application range; furthermore, due to the lack of a self-learning correction mechanism, it cannot achieve self-learning correction processing of the control process, making it difficult to adapt to dynamic changes in rolling conditions, affecting control accuracy and stability. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide an automatic control method for the sickle bend of intermediate billets in roughing mills.

[0006] An automatic control method for the sickle bend of a rough-rolled intermediate billet provided by the present invention includes the following steps: S1: Construct a process parameter table for the sickle bend control model to store the exclusive process parameters corresponding to different rolled material properties; S2: Construct a genetic learning table for the sickle-shaped bend control model, and reserve a field for leveling correction values ​​to provide compensation correction data for leveling value calculation; S3: Construct calculation formulas for R1 leveling value and R2 leveling value. The input items of the calculation formula include the exclusive process parameters, the leveling correction value and the sickle bend degree data. S4: Collect production data; S5: Calculate the sickle bend degree data based on the production data collected in S4; S6: Match the process parameter table of S1 according to the rolled material properties to obtain the exclusive process parameters, query the genetic learning table of S2 to obtain the leveling correction value, combine the camber degree data of S5, and substitute it into the calculation formula of S3 to calculate the R1 leveling value and R2 leveling value. S7: Adjust the mill roll gap according to the R1 leveling value and R2 leveling value of S6 to control the sickle bend; S8: After rolling is completed, obtain the actual values ​​of R1 leveling and R2 leveling, calculate the deviation from the R1 leveling value and R2 leveling value in S6, and update the R1 leveling correction value and R2 leveling correction value in the genetic learning table in S2.

[0007] Preferably, the rolled material properties include: steel grade classification code, final rolling thickness level, roughing width level, material hardness level, and final rolling temperature level.

[0008] Preferably, step S1 further includes constructing a classification index based on the properties of the rolled material, supplementing the index with corresponding production target values ​​and process parameters, and designing fuzzy matching rules so that rolled materials with different properties can be matched with exclusive process parameters.

[0009] Preferably, the genetic learning table and the process parameter table use the same rolling material attribute classification index; S2 further includes: establishing a query and update interface for the genetic learning table, wherein the query interface is used to query the leveling correction value in S6, and the update interface is used to update the leveling correction value in S8.

[0010] Preferably, in step S3, the formula for calculating the R1 leveling value is:

[0011] In the formula: The calculated R1 leveling value; R1 is the basic leveling value; Signed functions; The R1 centerline deviation value of the previous rolled material; R1 Sickle Curve Target Value; R1 width influences constants; R1 width influence coefficient; R1 width; R1 is a constant that affects hardness. R1 hardness influence coefficient; Planned hardness levels; E1 is the constant influenced by lateral pressure. E1 lateral pressure influence coefficient; E1 lateral pressure value; R1 leveling correction value.

[0012] Preferably, in step S3, the formula for calculating the R2 leveling value is:

[0013] In the formula: The calculated R2 leveling value; R2 baseline leveling value; R2 Sickle Curve Target Value; The R2 centerline deviation value of the previous rolled material; R2 width influences constants; R2 width influence coefficient; R2 width; R2 hardness influence constant; R2 hardness influence coefficient; Planned hardness levels; R2 thickness influence constant; R2 thickness influence coefficient; R2 thickness; R1 feedforward factor; R1 centerline deviation value; The R1 centerline deviation value of the previous rolled material; R2 leveling correction value.

[0014] Preferably, in step S4, the production data includes rolled material attribute data, R1 centerline deviation point data, and R2 centerline deviation point data; In step S5, the centerline deviation point data of R1 and the centerline deviation point data of R2 are denoised, sampled and averaged to obtain the centerline deviation values ​​of R1 and R2.

[0015] Preferably, step S6 further includes optimization processing of the leveling values ​​of R1 and R2: If the deviation between the R1 leveling value and the R1 leveling value of the previous rolled material is greater than the leveling change limit, then the R1 leveling value is limited to the range of the leveling change limit of the R1 leveling value of the previous rolled material plus / minus. If the absolute value of the R1 leveling value is less than the R1 sickle bend dead zone, then the R1 leveling value is set to zero, that is, the R1 roll gap deviation is not adjusted. If the deviation between the R2 leveling value and the R2 leveling value of the previous rolled material is greater than the leveling change limit, then the R2 leveling value is limited to the range of the leveling change limit of the R2 leveling value of the previous rolled material plus / minus. If the absolute value of the R2 leveling value is less than the R2 sickle bend dead zone, then the R2 leveling value is set to zero, that is, the roll gap deviation of R2 is not adjusted.

[0016] Preferably, in step S8, updating the leveling correction value in the genetic learning table includes: The difference between the actual R1 leveling value and the R1 leveling value calculated in S6 is calculated, and after being constrained by a preset ratio and a preset upper limit, the R1 leveling correction value is obtained. The difference between the actual R2 leveling value and the R2 leveling value calculated in S6 is calculated, and after being constrained by a preset ratio and a preset upper limit, the R2 leveling correction value is obtained. Update the R1 leveling correction value and R2 leveling correction value to the leveling correction value field under the corresponding rolled material attribute index in the genetic learning table.

[0017] Preferably, in step S7, adjusting the mill roll gap according to the leveling values ​​R1 and R2 includes: If the R1 leveling value is negative, perform a roll gap lifting operation on the R1 mill to shift the rolled material toward the operating side; If the R1 leveling value is positive, the R1 mill is subjected to a roll gap pressing operation, causing the rolled material to shift towards the drive side; If the R1 leveling value is zero, roll gap adjustment will not be performed on the R1 mill. If the R2 leveling value is negative, perform a roll gap lifting operation on the R2 mill to shift the rolled material toward the operating side; If the R2 leveling value is positive, the R2 mill is subjected to a roll gap pressing operation, causing the rolled material to shift towards the drive side; If the R2 leveling value is zero, roll gap adjustment will not be performed on the R2 mill.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention calculates the leveling value by combining the rolling plan, process parameters and the degree of camber, and adjusts the roll gap deviation of the rolling mill based on the leveling value, which effectively improves the problems of poor plate shape and production failure caused by camber, and improves the rolling quality and production process stability. 2. This invention effectively filters out noise interference in the data by removing and partially filtering the collected centerline deviation point data, thus solving the problem of large calculation errors in centerline deviation values ​​caused by noise. 3. By setting up a process parameter table categorized according to the properties of rolled materials, this invention enables rolled materials with different properties to be matched with calculation parameters that are suitable for their characteristics, avoiding the calculation deviation caused by parameter generalization in the prior art, and significantly improving the accuracy of leveling value calculation. 4. By employing an adaptive genetic learning table, this invention can automatically acquire the leveling value deviation of historically rolled materials with the same properties during the rolling process of the current block of rolled material, and correct the leveling value of the current block of rolled material based on the deviation, effectively offsetting the impact of fluctuations in actual production conditions on the leveling value calculation, and further improving the accuracy of the leveling value calculation. Attached Figure Description

[0019] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart illustrating the main model construction process of this invention; Figure 2 This is a schematic diagram illustrating the deviation of the center line of the sickle-shaped rolled material, which is the main feature of this invention. Figure 3 This is a schematic diagram illustrating the data acquisition rules for the centerline deviation points of R1 and R2 in this invention. Figure 4 This is a side view illustrating the adjustment process performed during the rolling process, which is the main feature of this invention. Figure 5 This is a schematic diagram of the cross-section of a rolling production line that mainly embodies the principle of roll gap deviation adjustment in this invention; Figure 6 This is a flowchart illustrating the main production process of this invention.

[0020] As shown in the figure: Detailed Implementation

[0021] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0022] This invention provides an automatic control method for camber in roughing mill intermediate slabs, applicable to roughing mill production lines in steel enterprises, addressing the camber problem that easily occurs during intermediate slab rolling. This embodiment achieves automatic and precise control of camber through a closed-loop process of model building, data acquisition, computational control, and self-learning updates.

[0023] Specifically, this method establishes a roughing rolling process parameter table, a genetic learning table, and a calculation model for the leveling values ​​of the R1 and R2 mills in the roughing rolling process. During the roughing rolling process, it collects and processes the centerline deviation data of the intermediate slab, and calculates the leveling value by combining the planned data, pre-calculated data, process parameters, and genetic learning correction values. The calculation results are then sent to the actuator for mill control, thereby achieving automatic and precise adjustment of the mill roll gap deviation and improving the problem of poor plate shape caused by camber.

[0024] The core innovation of this method lies in constructing a process parameter table based on process experience and requirements to provide different parameters for model calculations for rolled materials with different specifications and properties. The genetic learning table is used to compensate and correct the leveling settings, significantly reducing the standard deviation of the centerline deviation, increasing the automatic leveling rate to over 90%, effectively reducing production failures and increasing production line capacity.

[0025] This method mainly includes an initial model building phase and a later production application phase. The initial model building provides the computational foundation for automatic control and regulation, while the later production application enables the automatic control process to be implemented.

[0026] Figure 1 This is a flowchart illustrating the model building process of this invention. The model building process mainly includes: S1: Construct the process parameter table for the sickle bend control model; S2: Construct the genetic learning table for the sickle bend control model; and S3: Construct the calculation formulas for R1 and R2 leveling values. The implementation methods for each step are described below.

[0027] S1: Construct the process parameter table for the sickle-shaped bend control model. The specific implementation is as follows: Considering that different specifications of rolled materials require different process parameters for calculation, based on process experience, some representative attribute tuples are designed to be used as the basis for classifying different specifications. These tuples describe the type, specifications, and material of the rolled materials. This ensures that the rolled materials can be effectively classified without making parameter maintenance too difficult due to too many tuple items. For example, attribute tuples are used to classify the rolled materials and establish an index to form a process parameter table index structure.

[0028] Considering the requirements of different specifications and materials for process parameters, the core properties of the rolled material were initially determined to be steel grade, final rolled thickness, rough rolled width, material hardness, and final rolled temperature. However, it was found that there was a defect in the attribute tuple: the original measurement data of attributes such as final rolled thickness and rough rolled width were floating-point numbers. Even if the overall properties of the rolled material were similar, a small difference in the value of a single floating-point attribute, such as two rolled materials with a final rolled thickness difference of 0.2 mm, would cause them to be classified into different categories. This would increase the difficulty of maintaining the parameter table and would not allow for the reuse of parameters for similar rolled materials, which would not meet production expectations.

[0029] To solve this problem, the numerical ranges of each attribute are stratified. That is, the continuous values ​​of each attribute are divided into several ranges, and values ​​within the same range are grouped into one level. The original floating-point numbers are replaced with integer codes starting from 1. The specific stratification logic is as follows: Steel grades can be categorized into several numerical ranges based on the carbon content of the rolled material. Each range is represented by a code starting with 1. Special steel grades such as stainless steel and heat-resistant steel can be assigned unique codes, resulting in more than 20 steel grade classification codes.

[0030] The final rolling thickness can be stratified according to a continuous numerical range of the final rolling thickness, such as 10-12mm, 12-14mm, etc., numbered starting from 1, forming more than 20 levels. The roughing width, material hardness, and final rolling temperature can also be divided into several levels starting from number 1, similar to the above method. The index structure of the process parameter table after stratification is shown in Table 1.

[0031] Table 1. Index Structure of Process Parameters

[0032] Based on production experience, after establishing the above index, production target values ​​and process parameter tuples are added, specifically including: R1 sickle bend target value, R1 sickle bend dead zone, R1 basic leveling value, R2 sickle bend target value, R2 sickle bend dead zone, R2 basic leveling value, R1 feedforward leveling factor, and leveling variation limit. This forms the production target value and process parameter structure table, as shown in Table 2.

[0033] Table 2. Production Target Values ​​and Process Parameter Structure

[0034] Since it is impractical to cover all index combinations of rolled material attributes, an 8-level fuzzy matching rule is designed to avoid omissions and enhance the maintainability of the process parameter table, as shown in Table 3. The core logic of fuzzy matching is that it is not required that all rolled material attributes and index tuples in the process parameter table match exactly. When no matching is found, some index items of the rolled material attributes can be treated as 0 to relax the matching conditions, and multiple matching attempts can be made until a corresponding record is found. The specific rules and process are as follows: The 8-level matching rule is represented by eight 5-bit 0-1 arrays. The array elements correspond to five index items: steel grade classification code, final rolling thickness level, roughing width level, material hardness level, and final rolling temperature level. A value of 1 indicates that the index item needs to be accurately matched with the parameter table, while 0 indicates that an accurate match is not required. The matching attempt starts from [1,1,1,1,1], meaning all index items need to be accurately matched. If no match is found in the process parameter table, the matching process continues in the following order: [1,1,1,1,0], [1,1,1,0,1], [1,1,1,0,0], [0,1,1,1,0], [0,1,1,0,0], [0,1,0,0,0], [0,0,0,0,0], with the matching requirements gradually becoming more lenient. If there are no matching results in the first 7 levels, the last level of all 0 items is triggered, and the general parameters corresponding to the all 0 index are preset in the process parameter table to ensure that all rolled materials can be matched with parameters.

[0035] Table 3 Fuzzy Matching Rules for Process Parameters

[0036] It can also build a process parameter table access interface, which can be used to calculate the process program by using the rolled material attributes for fuzzy matching indexing to obtain the corresponding production target value and process parameters for use in leveling value calculation.

[0037] S2: Construct the genetic learning table for the sickle-shaped bend control model. This step provides compensation and correction data for the leveling value calculation. The specific implementation is as follows: Considering the difficult-to-quantify influencing factors in the production process, such as environmental and equipment factors, which cannot or are extremely difficult to directly participate in the calculation as parameters, a genetic learning table for different rolled materials with different properties is designed to improve the adaptability and accuracy of the model calculation and correct the leveling value calculation results. This step can use rolled material attribute classification tuples that are completely consistent with the S1 process parameter table, namely, steel grade classification code, final rolling thickness level, roughing width level, material hardness level, and final rolling temperature level. Rolled materials with the same attribute tuples are grouped into one category, and this tuple is used as the unique index to locate a certain type of rolled material, ensuring that rolled materials with the same attributes can be associated with and matched process parameters and leveling correction values, as shown in Table 4.

[0038] Table 4. Genetic Learning Table Index Structure

[0039] Following the above-mentioned rolling material attribute classification index, a tuple for leveling correction values ​​is reserved, including fields for R1 leveling correction value and R2 leveling correction value. Initially, if there is no historical rolling data for a certain type of rolling material, the correction value can be set to 0 by default, as shown in Table 5.

[0040] Table 5. Structure of Leveling Correction Values ​​in Genetic Learning Tables

[0041] Two core interfaces of the genetic learning table can be constructed to implement query and update functions respectively: (1) Query interface: When S6 calculates the leveling, it can obtain the corresponding R1 leveling correction value and R2 leveling correction value according to the attribute index of the current rolled material, which are used for the compensation calculation of the leveling value; (2) Update interface: After S8 finishes rolling, it can write the new R1 leveling correction value and R2 leveling correction value. The updated correction value will be used for the leveling value calculation of other rolled materials with the same attribute in the subsequent production, so as to realize the self-learning iteration of the model.

[0042] S3: Construct the calculation formulas for the R1 and R2 leveling values, as follows: The calculation of the R1 leveling value uses the process parameters of S1, the leveling correction value of S2, and the camber degree data of S5 as core inputs. It needs to be based on process parameters such as the target value of R1 camber of S1 and the basic leveling value of R1, while also considering the R1 centerline deviation value of the previous rolled material (the difference with the target value is used to determine the adjustment direction of the centerline deviation). The rolling material width, hardness, and E1 vertical roll side pressure are used as proportional coefficients for the adjustment amount. The coefficients of each influencing factor are determined through linear regression analysis. Finally, the leveling correction value of S2 is added to obtain the R1 leveling value.

[0043]

[0044] In the formula: The initial R1 leveling value to be determined; The basic leveling value of R1 in the process parameter table comes from S1; The defined symbolic function returns 1 if its input parameter is greater than 0, 0 if it is equal to 0, and -1 if it is less than 0. The previous R1 centerline deviation value was collected from S4; The target value for the R1 camber curve in the process parameter table comes from S1; The influence constant of the R1 width given a value is obtained through linear regression; The R1 width influence coefficient, given a value, is obtained through linear regression. The pre-calculated width R1 comes from S4; The R1 hardness influence constant, given a value, is obtained through linear regression. The R1 hardness influence coefficient, given a value, is obtained through linear regression. The planned hardness level is derived from S4; The E1 lateral pressure influence constant, given a value, is obtained through linear regression. The E1 lateral pressure influence coefficient, given a value, is obtained through linear regression. The pre-calculated E1 lateral pressure value comes from S4; The R1 leveling correction value for the genetic coefficient table is derived from S2.

[0045] The R2 leveling value, based on the calculation logic of R1, adds the feedforward effect of R1 on R2. It needs to be based on process parameters such as the target value of R2 camber in S1 and the basic leveling value of R2, considering the R2 centerline deviation value of the previous rolled piece (differentiating the adjustment direction from the target value), using the width, hardness, and thickness of the rolled piece as proportional coefficients for the adjustment amount. Simultaneously, the R1 centerline deviation values ​​of the previous and current rolled pieces are included, and the R1 feedforward leveling factor in S1 is used as a proportional coefficient to achieve feedforward compensation of R1 on R2. After determining each coefficient through linear regression, the leveling correction value of S2 is added to obtain the R2 leveling value. The formula is as follows:

[0046] In the formula: The initial R² leveling value to be determined; The basic leveling value of R2 in the process parameter table comes from S1; The target value for R2 sickle bend in the process parameter table comes from S1; The previous R2 centerline deviation value was collected from S4; The influence constant of the R² width given a value is obtained through linear regression; The R² width influence coefficient, given a value, is obtained through linear regression. The planned R2 width comes from S4; The constant affecting R2 hardness, given a value, is obtained through linear regression. The R2 hardness influence coefficient, given a value, is obtained through linear regression. The planned hardness level is derived from S4; The R² thickness influence constant, given a value, is obtained through linear regression. The R² thickness influence coefficient, given a value, is obtained through linear regression. The pre-calculated R2 thickness comes from S4; The R1 feedforward factor in the process parameter table comes from S1; The collected R1 centerline deviation value comes from S5; The previous R1 centerline deviation value was collected from S4; The R² leveling correction value for the heritability coefficient table is derived from S2.

[0047] The production application phase of this method includes: S4: collecting production data; S5: calculating the degree of camber based on the production data; S6: calculating the R1 leveling value and R2 leveling value; S7: adjusting the mill roll gap according to the leveling value to control the camber; and S8: updating the leveling correction value in the genetic learning table.

[0048] Figure 2 This is a schematic diagram illustrating the deviation of the center line of the sickle-shaped rolled material, which is the main feature of this invention. Figure 3 This is a schematic diagram illustrating the data acquisition rules for the centerline deviation points of R1 and R2 in this invention. Figure 4 This is a side view illustrating the adjustment process during the rolling process, which is the main feature of this invention. Below, in conjunction with... Figures 2 to 4 This section describes the specific implementation method of this technique.

[0049] S4: Collect production data. This step is used to obtain the raw data required for subsequent calculations. The specific implementation is as follows: Before the rolling mill begins production, rolling mill planning data is received and stored from the upper-level planning and scheduling system, including steel grade classification code, final rolling thickness level, roughing width level, material hardness level, final rolling temperature level, and R2 width.

[0050] When a piece of rolled material passes through the width measuring instruments at the exits of the roughing mills R1 and R2, data on the centerline deviation point from the last exit width measuring instrument of the roughing mills R1 and R2 is collected. For example... Figure 4 As shown, the side press 51, R1 mill 52, and R2 mill 53 are arranged along the rolling direction. The second rolled material 32 is the "preceding rolled material" relative to the third rolled material 33, and the first rolled material 31 is the "preceding rolled material" relative to the second rolled material 32. Before the third rolled material 33 reaches the R1 mill 52, the second rolled material 32 has already reached the R1 mill 52. It is necessary to collect the centerline deviation of the second rolled material 32, calculate the R1 leveling value for the third rolled material 33, and send it to the R1 actuator. Figure 4 The execution process shown in the lower part is later than the production line status of the calculation process. During the execution process, the third rolled material 33 has arrived at the R1 mill 52. At this time, the R1 executor has adjusted the roll gap deviation according to the R1 leveling value, thereby reducing the degree of camber of the third rolled material 33.

[0051] Similarly, during the "calculation process," the second rolled material 32 is located at the exit of the R1 mill 52, preparing to enter the R2 mill 53. At this time, the first rolled material 31 has already passed through the R2 mill 53. Therefore, it is necessary to collect the R1 centerline deviation of the second rolled material 32, as well as the R1 and R2 center block deviations of the first rolled material 31, so as to calculate the R2 leveling value and send it to the R2 actuator. When the second rolled material 32 reaches the R2 mill 53, the R2 actuator has already adjusted the roll gap deviation according to the R2 leveling value, thereby reducing the degree of camber of the second rolled material 32 at the R2 mill 53.

[0052] like Figure 3 As shown, data acquisition begins when the head of the rolled material just reaches the R1 or R2 width measuring instrument and ends when the tail of the rolled material just leaves the R1 or R2 width measuring instrument. An optional acquisition method is as follows: the width measuring instrument located at the exit of R1 mill 52 collects a centerline deviation point at 5mm intervals within a 1000mm range at both the head and tail of the rolled material; the width measuring instrument located at the exit of R2 mill 53 collects a centerline deviation point at 10mm intervals within a 1000mm range at both the head and tail of the rolled material. The width measuring instruments at the exits of R1 mill 52 and R2 mill 53 collect a centerline deviation point at 100mm intervals in the middle of the rolled material, and the collected deviation points can reflect the horizontal offset of the rolled material's centerline.

[0053] When the rolling mill begins to draw steel, the R1 width, R2 thickness, and E1 lateral pressure value are collected in advance by the model process machine.

[0054] S5: Calculate the sickle bend degree data based on production data. This step converts the deviation point data into a usable sickle bend degree index. The specific implementation is as follows: First, data denoising is performed. To obtain the degree of camber at the head of the rolled material and reduce the impact of noise on subsequent calculations, after collecting the R1 centerline deviation point data, points with an absolute value greater than 100 are removed. At a given position at the beginning of the rolled material, several points are taken at regular intervals, and the average value of these points is calculated as the R1 centerline deviation value. After collecting the R2 centerline deviation point data, points with an absolute value greater than 100 are removed. At a given position at the beginning of the rolled material, several points are taken at regular intervals, and the average value of these points is calculated as the R2 centerline deviation value.

[0055] S6: Calculate the leveling values ​​of R1 and R2, as follows: Calculate the R1 leveling value: After the rolled material passes through the side press, i.e. before rolling R1, the R1 leveling value is calculated. Based on the index tuple (steel grade classification code, final rolling thickness level, rough rolling width level, material hardness level, final rolling temperature level), the process parameter table is accessed to obtain the corresponding R1 camber target value, R1 camber dead zone, R1 basic leveling value, and leveling variation limit.

[0056] Access the genetic learning table based on the index tuple (steel grade classification code, final rolling thickness level, rough rolling width level, material hardness level, final rolling temperature level) to obtain the corresponding R1 leveling correction value. If the original index of this type of attribute tuple does not exist, the R1 leveling correction value is 0.

[0057] Substitute the above parameters into the S3 construction The formula calculates the initial R1 leveling value.

[0058] The initial leveling values ​​are constrained and optimized to ensure they meet the requirements for equipment operation safety and control accuracy, including: Calculate the difference between the current R1 leveling value and the previous R1 leveling value. If the absolute value exceeds the leveling change limit obtained by S1, then limit the current R1 leveling value to the range of the previous leveling value plus / minus the leveling change limit. If the absolute value of the R1 leveling value is less than the R1 sickle bend dead zone value obtained by S1, then the R1 leveling value is set to zero, that is, there is no need to adjust the roll gap deviation. If the R1 leveling value exceeds the allowable physical adjustment range of the rolling mill equipment (which can be the upper / lower limit of a given value), it will be forcibly set to the corresponding upper / lower limit.

[0059] Calculate the R2 leveling value: After the rolled material passes through the R1 mill and before entering the R2 mill, the R2 leveling value is calculated. Based on the index tuples (steel grade classification code, final rolling thickness level, roughing width level, material hardness level, final rolling temperature level), the process parameter table is accessed to obtain the corresponding R2 camber target value, R2 camber dead zone, R2 basic leveling value, R1 feedforward leveling factor, and leveling variation limit.

[0060] Access the genetic learning table based on the index tuple (steel grade classification code, final rolling thickness level, rough rolling width level, material hardness level, final rolling temperature level) to obtain the corresponding R2 leveling correction value. If the original index of this type of attribute tuple does not exist, the R2 leveling correction value is 0.

[0061] Substitute the above parameters into the S3 construction The formula calculates the initial R² leveling value.

[0062] Constraint optimization is performed on the initial R2 leveling value, including: Calculate the difference between the current R2 leveling value and the previous R2 leveling value. If the absolute value exceeds the leveling change limit obtained by S2, then limit the current R2 leveling value to the range of the previous leveling value plus / minus the leveling change limit. If the absolute value of the R2 leveling value is less than the R2 sickle bend dead zone value obtained by S1, then the R2 leveling value is set to zero, that is, there is no need to adjust the roll gap deviation. If the R2 leveling value exceeds the allowable physical adjustment range of the rolling mill equipment (which can be the upper / lower limit of another given value), it will be forcibly set to the corresponding upper / lower limit.

[0063] S7: Adjust the mill roll gap according to the leveling value to control the camber, as follows: like Figure 5 As shown in the diagram above, when the rolled material is about to enter R1, the actuator controls the R1 equipment based on the calculated R1 leveling value. Specifically, if the R1 leveling value is negative, the R1 mill needs to be raised to increase the roll gap, making the pressure on the drive side of the rolled material in the direction perpendicular to the production line greater than the pressure on the operating side. Due to the material flow characteristics, the rolled material shifts towards the operating side, reducing centerline deviation. If the R1 leveling value is positive, the R1 mill needs to be pressed to increase the roll gap, making the pressure on the drive side of the rolled material in the direction perpendicular to the production line greater than the pressure on the operating side. Due to the material flow characteristics, the rolled material shifts towards the drive side, increasing centerline deviation. If the R1 leveling value is zero, no roll gap deviation adjustment is performed.

[0064] like Figure 5As shown in the diagram below, when the rolled material is about to enter the R2 mill, the actuator in the R2 mill calculates the final R2 leveling value and controls the R2 equipment. The specific execution process is as follows: If the R2 leveling value is negative, the R2 mill needs to perform a roll gap raising operation, so that the pressure on the transmission side of the rolled material in the direction perpendicular to the production line is greater than the pressure on the operating side. Due to the material flow characteristics, the rolled material shifts towards the operating side, reducing the centerline deviation. If the R2 leveling value is positive, the R2 mill needs to perform a roll gap pressing operation, so that the pressure on the transmission side of the rolled material in the direction perpendicular to the production line is greater than the pressure on the operating side. Due to the material flow characteristics, the rolled material shifts towards the transmission side, increasing the centerline deviation. If the R2 leveling value is zero, no roll gap deviation adjustment is performed.

[0065] S8: Update the leveling correction values ​​in the genetic learning table. This step is used to achieve self-learning iteration, and the specific implementation is as follows: After R1 and R2 rolling are completed, the actual leveling values ​​of R1 and R2 are collected from the actuator.

[0066] The difference between the actual R1 leveling value and the R1 leveling value is used as the R1 leveling correction value, constrained by a given ratio and upper limit. The difference between the actual R2 leveling value and the R2 leveling value is used as the R2 leveling correction value, constrained by a given ratio and upper limit.

[0067] Update the R1 and R2 leveling correction values ​​to the tuples belonging to the rolled material in the genetic learning table. If the original index of this type of attribute tuple does not exist, insert a new item for use in calculating the leveling value of the rolled material with the same attribute next time.

[0068] In the description of this application, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0069] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. An automatic control method for the sickle bend of a rough-rolled intermediate billet, characterized in that, Includes the following steps: S1: Construct a process parameter table for the sickle bend control model to store the exclusive process parameters corresponding to different rolled material properties; S2: Construct a genetic learning table for the sickle-shaped bend control model, and reserve a field for leveling correction values ​​to provide compensation correction data for leveling value calculation; S3: Construct calculation formulas for R1 leveling value and R2 leveling value. The input items of the calculation formula include the specific process parameters, the leveling correction value and the sickle bend degree data. S4: Collect production data; S5: Calculate the sickle bend degree data based on the production data collected in S4; S6: Match the process parameter table of S1 according to the rolled material properties to obtain the exclusive process parameters, query the genetic learning table of S2 to obtain the leveling correction value, combine the camber degree data of S5, and substitute it into the calculation formula of S3 to calculate the R1 leveling value and R2 leveling value. S7: Adjust the mill roll gap according to the R1 leveling value and R2 leveling value of S6 to control the sickle bend; S8: After rolling is completed, obtain the actual value of R1 leveling and the actual value of R2 leveling, calculate the deviation from the R1 leveling value and R2 leveling value in S6, and update the R1 leveling correction value and R2 leveling correction value in the genetic learning table of S2. The rolled material attributes include: steel grade classification code, final rolled thickness level, rough rolled width level, material hardness level, and final rolled temperature level; S1 further includes constructing a classification index based on the properties of rolled materials, supplementing the index with corresponding production target values ​​and process parameters, and designing fuzzy matching rules so that rolled materials with different properties can be matched with exclusive process parameters. The genetic learning table and the process parameter table use the same rolling material attribute classification index; S2 further includes: establishing a query and update interface for the genetic learning table, wherein the query interface is used to query the leveling correction value in S6, and the update interface is used to update the leveling correction value in S8. In S3, the formula for calculating the R1 leveling value is: In the formula: The calculated R1 leveling value; R1 is the basic leveling value; Signed functions; The R1 centerline deviation value of the previous rolled material; R1 Sickle Curve Target Value; R1 width influences constants; R1 width influence coefficient; R1 width; R1 is a constant that affects hardness. R1 hardness influence coefficient; Planned hardness levels; E1 is the constant influenced by lateral pressure. E1 lateral pressure influence coefficient; E1 lateral pressure value; R1 leveling correction value; In step S3, the formula for calculating the R2 leveling value is: In the formula: The calculated R2 leveling value; R2 baseline leveling value; R2 Sickle Curve Target Value; The R2 centerline deviation value of the previous rolled material; R2 width influences constants; R2 width influence coefficient; R2 width; R2 hardness influence constant; R2 hardness influence coefficient; Planned hardness levels; R2 thickness influence constant; R2 thickness influence coefficient; R2 thickness; R1 feedforward factor; R1 centerline deviation value; The R1 centerline deviation value of the previous rolled material; R2 leveling correction value.

2. The automatic control method for the sickle bend of the intermediate billet in roughing mill as described in claim 1, characterized in that, In step S4, the production data includes rolled material attribute data, R1 centerline deviation point data, and R2 centerline deviation point data. In step S5, the centerline deviation point data of R1 and the centerline deviation point data of R2 are denoised, sampled and averaged to obtain the centerline deviation values ​​of R1 and R2.

3. The automatic control method for the sickle bend of the intermediate billet in roughing mill as described in claim 1, characterized in that, S6 also includes optimization processing of the R1 and R2 leveling values: If the deviation between the R1 leveling value and the R1 leveling value of the previous rolled material is greater than the leveling change limit, then the R1 leveling value is limited to the range of the leveling change limit of the R1 leveling value of the previous rolled material plus / minus. If the absolute value of the R1 leveling value is less than the R1 sickle bend dead zone, then the R1 leveling value is set to zero, that is, the R1 roll gap deviation is not adjusted. If the deviation between the R2 leveling value and the R2 leveling value of the previous rolled material is greater than the leveling change limit, then the R2 leveling value is limited to the range of the leveling change limit of the R2 leveling value of the previous rolled material plus / minus. If the absolute value of the R2 leveling value is less than the R2 sickle bend dead zone, then the R2 leveling value is set to zero, that is, the roll gap deviation of R2 is not adjusted.

4. The automatic control method for the sickle bend of the intermediate billet in roughing mill as described in claim 1, characterized in that, In step S8, updating the leveling correction value in the genetic learning table includes: The difference between the actual R1 leveling value and the R1 leveling value calculated in S6 is calculated, and after being constrained by a preset ratio and a preset upper limit, the R1 leveling correction value is obtained. The difference between the actual R2 leveling value and the R2 leveling value calculated in S6 is calculated, and after being constrained by a preset ratio and a preset upper limit, the R2 leveling correction value is obtained. Update the R1 leveling correction value and R2 leveling correction value to the leveling correction value field under the corresponding rolled material attribute index in the genetic learning table.

5. The automatic control method for the sickle bend of the intermediate billet in roughing mill as described in claim 1, characterized in that, In step S7, adjusting the mill roll gap according to the leveling values ​​R1 and R2 includes: If the R1 leveling value is negative, perform a roll gap lifting operation on the R1 mill to shift the rolled material toward the operating side; If the R1 leveling value is positive, the R1 mill is subjected to a roll gap pressing operation, causing the rolled material to shift towards the drive side; If the R1 leveling value is zero, roll gap adjustment will not be performed on the R1 mill. If the R2 leveling value is negative, perform a roll gap lifting operation on the R2 mill to shift the rolled material toward the operating side; If the R2 leveling value is positive, the R2 mill is subjected to a roll gap pressing operation, causing the rolled material to shift towards the drive side; If the R2 leveling value is zero, roll gap adjustment will not be performed on the R2 mill.

Images