Short stroke control based stand roll mill gap control method
By constructing a fifth-order polynomial head and tail profile curve and a fourth-order speed model, combined with a deviation regression prediction model, the problem of poor opening control effect of vertical roll mill was solved, and more precise workpiece width control and yield improvement were achieved.
Patent Information
- Application Number
- CN202511608045.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-05
- Publication Date
- 2026-02-13
- Estimated Expiration
- 2045-11-05
AI Technical Summary
The existing method for controlling the opening of vertical roll mills, which uses a quadratic polynomial head and tail profile curve, is insufficient, resulting in poor control of the opening of vertical roll mills and an inability to effectively prevent workpieces from losing width at the head and tail, thus affecting the yield.
By combining a quintic polynomial head-and-tail profile curve with a quartic velocity model and a deviation regression prediction model, the segmented rolling information of the rolled portion of the target workpiece in the vertical roll mill is obtained. A quintic polynomial head-and-tail profile curve is constructed, the first derivative of time is calculated to obtain a quartic velocity model, and a deviation regression prediction model is constructed to accurately describe the changes in workpiece profile and width, providing a precise data foundation for the opening degree control of the vertical roll mill.
It improves the accuracy of vertical roll mill opening control, reduces cutting losses caused by traditional PID control, and increases yield.
Smart Images
Figure CN121060967B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of vertical roll rolling, and particularly relates to a vertical roll rolling mill opening degree control method based on short stroke control. BACKGROUND
[0002] The short stroke control technology is to dynamically control the opening degree of the vertical roll rolling to prevent the head and tail width loss of the workpiece caused by large lateral pressure, so as to control the shape of the head and tail end of the workpiece in the rolling process, reduce material cutting loss, and improve the yield. The basic content of the short stroke control technology is: according to the head and tail profile curve of the workpiece caused by large lateral pressure, the roll gap (opening degree) of the vertical roll rolling mill is continuously controlled in the vertical roll rolling production process, so that the change amount of the vertical roll rolling mill roll gap just meets the compensation of the width loss of the workpiece head and tail, so as to reduce the difference between the width of the head and tail end of the workpiece after horizontal roll rolling and the target width.
[0003] Based on the above, the approximation degree of the head and tail profile curve of the workpiece and the expression become the key of the short stroke control of the head and tail shape. At present, the opening degree control method of the vertical roll rolling mill is to measure the width of the workpiece in the rolling process, then to construct the quadratic polynomial head and tail profile curve of the workpiece based on the measured width, and to control the opening degree of the vertical roll rolling mill combined with the PID control algorithm. However, the quadratic polynomial head and tail profile curve is insufficient to describe the width increase / loss of the workpiece, resulting in poor control effect of the opening degree of the vertical roll rolling mill. SUMMARY
[0004] To solve the above technical problems, the present application provides a vertical roll rolling mill opening degree control method based on short stroke control. The technical scheme of the present application is as follows:
[0005] The vertical roll rolling mill opening degree control method based on short stroke control comprises:
[0006] S1, obtaining the rolling information of a plurality of segments of the already rolled part of the target workpiece in the vertical roll rolling mill;
[0007] S2, constructing a quintic polynomial head and tail profile curve according to the preset vertical roll reduction, the preset horizontal roll reduction and the rolling information of each segment of the target workpiece;
[0008] S3, obtaining a quartic speed model by taking the first order derivative of time of the quintic polynomial head and tail profile curve, and extracting a plurality of feature points of the quartic speed model;
[0009] S4, constructing a deviation degree regression prediction model between the plurality of feature points of the quartic speed model and the preset vertical roll reduction, the preset horizontal roll reduction and the rolling information of each segment;
[0010] S5, controlling the opening degree of the vertical roll rolling mill in the vertical roll rolling according to the deviation degree regression prediction model.
[0011] Preferably, the S1 comprises:
[0012] S11, according to the acquisition frequency of the monitoring device arranged in advance on the edger mill and the preset moving speed of the target workpiece, obtaining the original plate width data and the original plate thickness data corresponding to each time node of the rolled part of the target workpiece;
[0013] S12, segmenting the original plate width data and the original plate thickness data corresponding to each time node to obtain the rolling information of each segment.
[0014] Preferably, the S12 comprises:
[0015] S121, segmenting the original plate width data and the original plate thickness data corresponding to all time nodes according to a preset segment number to obtain a plurality of segments;
[0016] S122, taking the first time node sorted in time sequence in each segment as the origin of the workpiece standard coordinate system of the segment, and constructing the workpiece coordinate point corresponding to each time node after the origin in each segment, wherein each workpiece coordinate point contains four-dimensional information, which are: original plate width data, original plate thickness data, time node and segment; all workpiece coordinate points in each segment constitute the rolling information of each segment.
[0017] Preferably, the S2 comprises:
[0018] S21, calculating the plate width standard deviation of each segment according to the rolling information of each segment of the target workpiece, and taking the segment with the smallest plate width standard deviation as the stable rolling section of the target workpiece;
[0019] S22, calculating the plate width mean of the stable rolling section according to the rolling information of the stable rolling section, and taking the plate width mean as the standard plate width of the target workpiece;
[0020] S23, calculating the difference between each workpiece coordinate point in each segment of the target workpiece and the standard plate width according to the standard plate width of the target workpiece;
[0021] S24, determining the compensation amount of each workpiece coordinate point according to the difference between each workpiece coordinate point and the standard plate width, and fitting the compensation amounts of all workpiece coordinate points to obtain a quintic polynomial head-tail contour curve.
[0022] Preferably, the S24 comprises:
[0023] S241, taking the compensation amounts of the two workpiece coordinate points at the head and tail of each segment as the opposite numbers, and keeping the compensation amounts of the middle workpiece coordinate points in each segment unchanged;
[0024] S242, calculating an optimal coefficient set of a preset quintic polynomial function according to the compensation amounts of all workpiece coordinate points of all segments, and determining a quintic polynomial head-tail contour curve according to the optimal coefficient set and the preset quintic polynomial function.
[0025] Preferably, the S242 calculates the optimal coefficient set {ai} of the preset quintic polynomial function according to the compensation amounts of all workpiece coordinate points of all segments by formula (1):
[0026] (1);
[0027] wherein ai represents the i-th coefficient in the optimal coefficient set, and i∈[0, 5]; N represents the number of workpiece coordinate points; p( ) represents the preset quintic polynomial function; Xk represents the k-th workpiece coordinate point in time node order; C(Xk) represents the compensation amount of the k-th workpiece coordinate point; and min ai represents the minimized optimal coefficient ai.
[0028] Preferably, the S3 comprises:
[0029] S31, performing first-order derivative operation on the quintic polynomial head-tail contour curve with respect to time, and obtaining a quartic velocity model by combining the polynomial derivation rule;
[0030] S32, segmenting the quartic velocity model according to the segmentation results of all time nodes to obtain segmented velocity models of a preset segmentation number;
[0031] S33, identifying the function change sensitivity of each segmented velocity model, and extracting feature points of each segmented velocity model according to the function change sensitivity of each segmented velocity model, wherein the feature points of all segmented velocity models constitute a plurality of feature points of the quartic velocity model.
[0032] Preferably, the S33 comprises:
[0033] S331, calculating the partial derivative of each time node in each segmented velocity model, and calculating the function change sensitivity of each time node according to a preset sensitivity coefficient and the partial derivative of each time node in each segmented velocity model;
[0034] S332, calculating the average of the function change sensitivities of all segmented velocity models, and taking the time nodes in all segmented velocity models with a function change sensitivity greater than the average of the function change sensitivities as the feature points;
[0035] S333, combining the feature points of all segmented velocity models to obtain the plurality of feature points of the quartic velocity model.
[0036] Preferably, the S4 comprises:
[0037] S41, acquire the time node corresponding to each feature point of the four-time velocity model, and construct a matching relationship between each time node and the preset vertical roll reduction, the preset horizontal roll reduction, and the rolling information of each segment;
[0038] S42, construct a nonlinear power function relationship between all feature points and the preset vertical roll reduction, the preset horizontal roll reduction, and the rolling information of each segment according to the matching relationship and all feature points;
[0039] S43, fit the nonlinear power function relationship to obtain a regression coefficient of the nonlinear power function relationship, and construct a deviation degree regression prediction model between multiple feature points and the vertical roll reduction, the horizontal roll reduction, and the rolling information of each segment according to the regression coefficient and the nonlinear power function relationship.
[0040] Preferably, the S5 comprises:
[0041] S51, predicting the deviation degree of the width of the target workpiece at each future time node in the vertical roll rolling according to the deviation degree regression prediction model;
[0042] S52, if the deviation degree of the width of the target workpiece at any future time node is greater than a preset width deviation threshold, determining that the future time node is an abnormal point;
[0043] S53, determining an opening degree adjustment amount of the vertical roll rolling mill according to the relationship between the deviation degree of the width of each abnormal point and the opening degree, and obtaining an opening degree control amount by superimposing the opening degree adjustment amount on a preset opening degree corresponding to the preset vertical roll reduction, so as to control the opening degree of the vertical roll rolling mill in the vertical roll rolling.
[0044] All the optional technical solutions described above can be combined arbitrarily, and the application does not describe the structures after combination in detail.
[0045] Through the above-mentioned scheme, the application has the following beneficial effects:
[0046] By acquiring the rolling information of a plurality of segments of the target workpiece in the vertical roll rolling mill, and constructing a quintic polynomial head-tail contour curve of the target workpiece, the target workpiece is segmented and processed, and the quintic polynomial head-tail contour curve is applied, so as to accurately fit the complex contour change of the target workpiece, especially the change in the head-tail area, which is more accurate than the quadratic polynomial;
[0047] A four-time velocity model is obtained by taking the first derivative of time of the quintic polynomial head-tail contour curve, a plurality of feature points in the four-time velocity model are extracted, and the change of the speed of the target workpiece in the rolling process is accurately described according to the plurality of feature points of the four-time velocity model, which can better describe the width change of the target workpiece and provide accurate width data basis for subsequent opening degree adjustment.
[0048] The deviation degree of the target workpiece width is predicted by the deviation degree regression prediction model, so as to reduce the cutting loss caused by the insufficient description of the target workpiece width increase / width loss caused by the traditional PID control, and improve the opening degree control effect of the edger mill.
[0049] The above description is only a summary of the technical solutions of the present application. In order to make the technical means of the present application clearer and can be implemented according to the content of the description, the following will be described in detail with the preferred embodiments of the present application and with the help of the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS
[0050] Figure 1 is a flowchart of the edger mill opening degree control method based on short stroke control provided by the embodiment of the present application. DETAILED DESCRIPTION
[0051] The specific embodiments of the present application will be further described in detail below in combination with the drawings and examples. The following examples are used to illustrate the present application, but not to limit the scope of the present application.
[0052] As shown in Figure 1 The edger mill opening degree control method based on short stroke control provided by the embodiment of the present application includes the following steps S1 to S5:
[0053] S1, obtaining the rolling information of a plurality of segments of the rolled part of the target workpiece in the edger mill.
[0054] Specifically, the edger mill is also called vertical mill, which is a rolling equipment for metal processing. It adopts vertically arranged rolls, adjusts the opening degree to control the roll gap between the two rolls, and then controls the width of the target workpiece in rolling. The short stroke control dynamically corrects the opening degree of the edger mill, prevents the head and tail width loss of the target workpiece generated in the large side pressure process, and controls the shape of the head and tail end of the target workpiece in the rolling process. The target workpiece in the embodiment of the present application refers to the metal material or metal product subjected to rolling processing in the edger mill. The rolling information includes rolling speed, workpiece width and workpiece thickness, etc.
[0055] In one specific embodiment, the S1 includes:
[0056] S11, according to the acquisition frequency of the monitoring device pre-configured on the edger mill and the preset moving speed of the target workpiece, obtaining the original plate width data and the original plate thickness data corresponding to each time node of the rolled part of the target workpiece.
[0057] Specifically, the monitoring device includes an optical sensor and an ultrasonic sensor, wherein the ultrasonic sensor is used to measure original plate thickness point data of the rolled part of the target workpiece, and the optical sensor is used to measure original plate width point data of the rolled part of the target workpiece. However, since the original plate width point data and the original plate thickness point data are discrete, in order to facilitate subsequent processing, the blank length between any two discrete original plate width point data is obtained by collecting the frequency and the preset moving speed, and then the blank length between the two discrete original plate width point data is filled with the average of the two original plate width point data to obtain the original plate width data.
[0058] The original plate thickness data can be obtained in the same way.
[0059] S12, segmenting the original plate width data and the original plate thickness data corresponding to each time node to obtain rolling information of each segment.
[0060] Specifically, the segmentation processing refers to dividing the rolled part of the target workpiece in the edger mill into several parts according to the time nodes, and extracting the rolling information of each part.
[0061] In one specific embodiment, the S12 includes:
[0062] S121, segmenting the original plate width data and the original plate thickness data corresponding to all time nodes according to a preset segment number to obtain several segments.
[0063] Specifically, the preset segment number is a number of segments of the rolled part of the target workpiece in the edger mill, which is determined according to historical experience values and time intervals. For example, assuming that the entire rolled part lasts for 1000 seconds, if the preset segment number is 10, then the original plate width data and the original plate thickness data of the 1000 seconds will be evenly divided into 10 segments, and each segment contains original plate width data and original plate thickness data of 100 seconds.
[0064] S122, taking the first time node in each segment in the time sequence as the origin of the workpiece standard coordinate system of the segment, and constructing the workpiece coordinate points corresponding to each time node after the origin in each segment, wherein each workpiece coordinate point contains four-dimensional information, which is: original plate width data, original plate thickness data, time node and segment; all workpiece coordinate points in each segment constitute the rolling information of each segment.
[0065] Specifically, the segment information in the four-dimensional information is represented by a unique identifier, and each segment is configured with its own identifier.
[0066] S2, constructing a quintic polynomial head-tail contour curve according to the preset edger reduction, the preset horizontal roller reduction and the rolling information of each segment.
[0067] Specifically, the preset vertical roll reduction refers to a predetermined reduction applied to the target workpiece by the vertical roll of the vertical roll mill during rolling. The preset horizontal roll reduction refers to a predetermined horizontal side reduction applied to the target workpiece by the horizontal roll of the vertical roll mill during rolling. The quintic polynomial head-tail profile curve refers to a curve representing the profile of the target workpiece obtained by fitting a quintic polynomial curve, which describes the profile change of the width of the target workpiece from the beginning to the end of rolling through a smooth curve.
[0068] In one specific embodiment, S2 comprises:
[0069] S21, calculating the plate width standard deviation of each segment of the target workpiece according to the rolling information of each segment, and taking the segment with the minimum plate width standard deviation as the stable rolling segment of the target workpiece.
[0070] Specifically, the standard deviation of the original plate width data of all time nodes in the rolling information of a segment of the target workpiece is extracted as the plate width standard deviation of the segment. The stable rolling segment refers to the segment with the minimum variation range (i.e., plate width standard deviation) of the original plate width data among all segments of the target workpiece.
[0071] S22, calculating the plate width mean of the stable rolling segment according to the rolling information of the stable rolling segment, and taking the plate width mean as the standard plate width of the target workpiece.
[0072] Specifically, the plate width mean is the mean of all original plate width data in the stable rolling segment. The standard plate width is used to compare the deviation of the original plate width data of other segments of the target workpiece from the standard plate width.
[0073] S23, calculating the difference between each workpiece coordinate point and the standard plate width in each segment of the target workpiece according to the standard plate width of the target workpiece.
[0074] Specifically, the difference between each workpiece coordinate point and the standard plate width is obtained by subtracting the standard plate width from the original plate width data in each workpiece coordinate point. The difference between each workpiece coordinate point and the standard plate width includes positive and negative relationships.
[0075] S24, determining the compensation amount of each workpiece coordinate point according to the difference between each workpiece coordinate point and the standard plate width, and fitting the compensation amounts of all workpiece coordinate points to obtain a quintic polynomial head-tail profile curve.
[0076] Specifically, the compensation amount refers to the width deviation adjustment value of the target workpiece during rolling, and the compensation amount of a workpiece coordinate point represents the width deviation adjustment value to be applied at the position of the workpiece coordinate point. The quintic polynomial head-tail profile curve obtained by fitting can describe the variation of the compensation amount of each workpiece coordinate point of the target workpiece.
[0077] In one specific embodiment, the S24 comprises:
[0078] S241, taking the inverse of the compensation amount of the two workpiece coordinate points at the head and tail of each segment as the compensation amount of the two workpiece coordinate points at the head and tail, and keeping the compensation amount of the middle workpiece coordinate point in each segment unchanged.
[0079] Specifically, by taking the inverse of the compensation amount of the two workpiece coordinate points at the head and tail of each segment, that is, reversing the compensation amount of the workpiece coordinate points at the head and tail in each segment, the deformation of the target workpiece at the head and tail of each segment in subsequent opening control is balanced.
[0080] S242, calculating an optimal coefficient set of a preset quintic polynomial function according to the compensation amounts of all workpiece coordinate points of all segments, and determining a quintic polynomial head-tail contour curve according to the optimal coefficient set and the preset quintic polynomial function.
[0081] Specifically, the preset quintic polynomial function f(x) is generally represented as: ; wherein f(x) represents the dependent variable (compensation amount), x represents the independent variable (time), and {b1, b2, b3, b4, b5, b0} represents the preset coefficient set. After obtaining the optimal coefficient set, the optimal coefficient set replaces the preset coefficient set in the preset quintic polynomial function to obtain the quintic polynomial head-tail contour curve.
[0082] In one specific embodiment, the S242, when calculating the optimal coefficient set {ai} of the preset quintic polynomial function according to the compensation amounts of all workpiece coordinate points of all segments, is realized by formula (1):
[0083] (1);
[0084] wherein ai represents the i-th coefficient in the optimal coefficient set, and i ∈ [0, 5]; N represents the number of workpiece coordinate points, p( ) represents the preset quintic polynomial function, Xk represents the k-th workpiece coordinate point sorted in time node order, C(Xk) represents the compensation amount of the k-th workpiece coordinate point, min ai represents minimizing the optimal coefficient ai.
[0085] Specifically, formula (1) obtains the optimal coefficient set by the least squares method, so that the quintic polynomial head-tail contour curve obtained from the optimal coefficient set can describe all workpiece coordinate points as much as possible.
[0086] S3, taking the first order derivative of time of the quintic polynomial head-tail contour curve to obtain a quartic velocity model, and extracting a plurality of feature points of the quartic velocity model.
[0087] Specifically, the quartic velocity model is a curve describing the speed of change of the profile of the target workpiece; the feature points are points of significance on the quartic velocity model, and all the feature points describe the key features of the quartic velocity model.
[0088] In one specific embodiment, the S3 comprises:
[0089] S31, first-order derivative operation is performed on the quintic polynomial head-tail profile curve with respect to time, and a quartic velocity model is obtained by combining the polynomial derivation rule.
[0090] Specifically, the polynomial derivation rule is as follows when performing first-order derivative operation on each polynomial in the quintic polynomial head-tail profile curve: .
[0091] S32, the quartic velocity model is segmented according to the segmentation results of all time nodes, and segmented velocity models of a preset segmentation number are obtained.
[0092] Specifically, the segmentation is to continuously segment the quartic velocity model according to the preset segmentation number, so as to obtain segmented velocity models of a preset segmentation number.
[0093] S33, the function change sensitivity of each segmented velocity model is identified, and the feature points of each segmented velocity model are extracted according to the function change sensitivity of each segmented velocity model, and the feature points of all segmented velocity models constitute a plurality of feature points of the quartic velocity model.
[0094] Specifically, the function change sensitivity is a description of the change degree of the independent variable of each segmented velocity model, and in the embodiment of the application, the function change sensitivity refers to the curve change rate of each segmented velocity model at a certain point.
[0095] In one specific embodiment, the S33 comprises:
[0096] S331, the partial derivative of each time node in each segmented velocity model is calculated, and the function change sensitivity of each time node is calculated according to a preset sensitivity coefficient and the partial derivative of each time node in each segmented velocity model.
[0097] Specifically, the preset sensitivity coefficient is a coefficient representing the sensitivity of the curve change of the segmented velocity model, which is determined according to an empirical value. When calculating the function change sensitivity of a time node in a segmented velocity model, the product of the preset sensitivity coefficient and the partial derivative of the time node is obtained.
[0098] S332, the average of the function change sensitivities of all segmented velocity models is calculated, and the time nodes in all segmented velocity models whose function change sensitivity is greater than the average of the function change sensitivities are taken as the feature points.
[0099] Specifically, when calculating the average of the function change sensitivity of all the segmented speed models, the average of the function change sensitivity of all the time nodes of all the segmented speed models is calculated.
[0100] S333, the characteristic points of all the segmented speed models are combined to obtain a plurality of characteristic points of the quartic speed model.
[0101] S4, a deviation degree regression prediction model between the plurality of characteristic points of the quartic speed model and the preset vertical roll reduction, the preset horizontal roll reduction, and the rolling information of each segment is constructed.
[0102] Specifically, the deviation degree regression prediction model is a model for predicting and quantifying the width deviation degree in the future vertical rolling process of the target workpiece.
[0103] In one specific embodiment, the S4 comprises:
[0104] S41, a time node corresponding to each characteristic point of the quartic speed model is obtained, and a matching relationship between each time node and the preset vertical roll reduction, the preset horizontal roll reduction, and the rolling information of each segment is constructed.
[0105] Specifically, the matching relationship is a relationship constructed between the preset vertical roll reduction, the preset horizontal roll reduction, and the rolling information corresponding to each characteristic point, and the preset vertical roll reduction and the preset horizontal roll reduction are both a preset value.
[0106] S42, a non-linear power function relationship between all the characteristic points and the preset vertical roll reduction, the preset horizontal roll reduction, and the rolling information of each segment is constructed according to the matching relationship and all the characteristic points.
[0107] Specifically, the non-linear power function relationship is a model describing the influence degree of the preset vertical roll reduction, the preset horizontal roll reduction, and the rolling information of each segment on each characteristic point. When constructing specifically, the value of the segmented speed model where each characteristic point is located is taken as the dependent variable, and the preset vertical roll reduction, the preset horizontal roll reduction, and the rolling information corresponding to each characteristic point are taken as a plurality of independent variables according to the matching relationship to construct the non-linear power function relationship. Wherein, the non-linear power function relationship can be expressed as: y=a×x1 c1 ×x2 c2 ×x3 c3 ×x4 c4 ; wherein {a, c1, c2, c3, c4} is a blank power coefficient set, x1 represents the preset vertical roll reduction, x2 represents the preset horizontal roll reduction, x3 represents the original plate width data in the rolling information, and x4 represents the original plate thickness data in the rolling information. y represents the value of the segmented speed model where any characteristic point is located.
[0108] S43, fitting the non-linear power function relationship to obtain the regression coefficient of the non-linear power function relationship, and constructing a deviation degree regression prediction model between the plurality of characteristic points and the roll gap of the vertical roll, the roll gap of the horizontal roll and the rolling information of each segment according to the regression coefficient and the non-linear power function relationship.
[0109] Specifically, the non-linear power function relationship is fitted by using the non-linear least squares method to obtain the regression coefficient of the non-linear power function relationship, and the regression coefficient is substituted into the non-linear power function relationship to replace the blank power coefficient set to obtain the deviation degree regression prediction model between the plurality of characteristic points and the roll gap of the vertical roll, the roll gap of the horizontal roll and the rolling information of each segment.
[0110] S5, controlling the roll gap of the vertical roll in the vertical rolling according to the deviation degree regression prediction model.
[0111] Specifically, the width deviation of the target workpiece in the future vertical rolling is predicted according to the deviation degree regression prediction model, and the roll gap of the vertical roll is controlled according to the width deviation degree.
[0112] In one specific embodiment, the S5 comprises:
[0113] S51, predicting the deviation degree of the width of the target workpiece at each future time node in the vertical rolling according to the deviation degree regression prediction model.
[0114] Specifically, each future time node, a preset standard plate thickness value, a preset horizontal roll gap and a preset vertical roll gap are input into the deviation degree regression prediction model, and the deviation degree regression prediction model outputs the deviation degree of the width of each future time node. The preset standard plate thickness value is a thickness value of the target workpiece in this rolling which is set in advance.
[0115] S52, if the deviation degree of the width of the target workpiece at any future time node is greater than a preset width deviation threshold value, determining that the future time node is an abnormal point.
[0116] Specifically, the preset width deviation threshold value is the maximum width deviation of the target workpiece determined by an empirical value.
[0117] S53, determining the opening degree adjustment amount of the vertical roll mill according to the relationship between the deviation degree of the width of each abnormal point and the opening degree, and superimposing the opening degree adjustment amount on the basis of the preset opening degree corresponding to the preset vertical roll gap to obtain the opening degree control amount, and controlling the opening degree of the vertical roll mill in the vertical rolling through the opening degree control amount.
[0118] Specifically, the relationship between the deviation degree of the width of each abnormal point and the opening degree is a rolling automation control relationship determined according to expert review and historical data, and the preset opening degree and the preset vertical roll gap are also determined by this method, so the preset opening degree can be calculated by the preset vertical roll gap.
[0119] In addition, the quintic polynomial head-tail profile curve is applied to the vertical roll rolling of the rough rolling section of the hot strip mill train, and the feedback use statistical data shows that the average monthly strip yield is 98.08% without using the quintic polynomial head-tail profile curve, and the average monthly strip yield increases to 98.19% after using the quintic polynomial head-tail profile curve, so the average strip yield is increased by 0.11%. Therefore, the application of the quintic polynomial head-tail profile curve has obtained good economic benefits for the enterprise.
[0120] Based on all the above embodiments, the vertical roll rolling mill opening control method based on short stroke control provided by the embodiments of the present application has the following beneficial effects:
[0121] Firstly, by acquiring the rolling information of the segmented rolling part of the target workpiece in the vertical roll rolling mill and constructing the quintic polynomial head-tail profile curve, the quintic polynomial head-tail profile curve can accurately describe the profile curve of the rolled part of the target workpiece in the vertical roll rolling mill, providing an accurate data basis for subsequent opening control.
[0122] Then, the first order derivative of time is obtained by the quintic polynomial head-tail profile curve to obtain a quartic speed model, and a plurality of feature points of the quartic speed model are extracted, and a deviation degree regression prediction model is constructed, which can accurately describe the speed change of the profile of the target workpiece in the rolling process, and the subsequent deviation degree regression prediction model can accurately describe the width deviation degree of the target workpiece, providing an accurate data basis for subsequent opening control.
[0123] Finally, the opening of the vertical roll rolling mill in the vertical roll rolling is controlled according to the deviation degree regression prediction model, and the width deviation degree of the deviation degree regression prediction model is combined to improve the opening control effect based on short stroke control.
[0124] The above is only the preferred embodiment of the present application, and is not used to limit the present application, it should be pointed out that for ordinary skilled in the art, without departing from the technical principles of the present application, a number of improvements and modifications can be made, these improvements and modifications should be regarded as the protection scope of the present application.
Claims
1. A method for controlling the opening degree of a vertical roll mill based on short-stroke control, characterized in that, include: S1, Obtain rolling information of several segments of the target workpiece that have been rolled in the vertical roll mill; S2, construct a fifth-order polynomial head and tail profile curve based on the preset roll reduction amount, preset horizontal roll reduction amount, and rolling information of each segment of the target workpiece. S2 includes: S21, calculate the standard deviation of plate width for each segment based on the rolling information of each segment of the target workpiece, and take the segment with the smallest standard deviation of plate width as the stable rolling segment of the target workpiece. The coordinate points of all workpieces in each segment constitute the rolling information of each segment. S22, calculate the average plate width of the stable rolling section based on the rolling information of the stable rolling section, and use the average plate width as the standard plate width of the target workpiece; S23, Calculate the difference between the coordinate point of each workpiece in each segment of the target workpiece and the standard plate width based on the standard plate width of the target workpiece; S24. Determine the compensation amount for each workpiece coordinate point based on the difference between each workpiece coordinate point and the standard plate width. Fit the compensation amounts of all workpiece coordinate points to obtain a fifth-order polynomial head and tail contour curve. S24 includes: S241, take the opposite number of the compensation amount of the two workpiece coordinate points at the beginning and end of each segment and use it as the compensation amount of the two workpiece coordinate points at the beginning and end, while keeping the compensation amount of the middle workpiece coordinate point in each segment unchanged. S242, calculate the optimal coefficient set of the preset fifth-order polynomial function based on the compensation amount of all workpiece coordinate points in all segments, and determine the fifth-order polynomial head and tail contour curves based on the optimal coefficient set and the preset fifth-order polynomial function. When S242 calculates the optimal coefficient set {ai} of the preset fifth-order polynomial function based on the compensation amount of all workpiece coordinate points in all segments, it does so through formula (1): （1）； Where ai represents the i-th coefficient in the optimal coefficients, and i∈[0,5]; N represents the number of workpiece coordinate points; p() represents the preset fifth-order polynomial function; Xk represents the k-th workpiece coordinate point sorted according to the time node order; C(Xk) represents the compensation amount of the k-th workpiece coordinate point; min ai This represents minimizing the optimal coefficient ai; S3. The first time derivative of the fifth-order polynomial head and tail contour curve is obtained to obtain the fourth-order velocity model, and multiple feature points of the fourth-order velocity model are extracted. S4, construct a regression prediction model for the deviation between multiple feature points of the four-speed model and the preset roll reduction, preset horizontal roll reduction, and rolling information of each segment; S5, the opening degree of the vertical roll mill in vertical roll rolling is controlled according to the deviation regression prediction model.
2. The method for controlling the opening degree of a vertical roll mill based on short-stroke control according to claim 1, characterized in that, S1 includes: S11. Based on the acquisition frequency of the monitoring equipment pre-configured on the vertical rolling mill and the preset moving speed of the target workpiece, obtain the original plate width data and original plate thickness data corresponding to each time node of the rolled part of the target workpiece. S12, the original plate width data and original plate thickness data corresponding to each time node are segmented and processed to obtain the rolling information of each segment.
3. The method for controlling the opening degree of a vertical roll mill based on short-stroke control according to claim 2, characterized in that, S12 includes: S121, divide the original board width data and original board thickness data corresponding to all time nodes into segments according to the preset number of segments to obtain several segments; S122, take the first time node in each segment, sorted in chronological order, as the origin of the standard coordinate system of the workpiece in that segment, and construct the workpiece coordinate points corresponding to each time node after the origin in each segment. Each workpiece coordinate point contains four-dimensional information, namely: original plate width data, original plate thickness data, time node, and segment. All workpiece coordinate points in each segment constitute the rolling information of each segment.
4. The method for controlling the opening degree of a vertical roll mill based on short-stroke control according to claim 3, characterized in that, S3 includes: S31, perform the first derivative operation of the fifth-order polynomial head and tail profile curves with respect to time, and combine it with the polynomial differentiation rule to obtain the fourth-order velocity model. S32, based on the segmentation results of all time nodes, the four-time velocity model is segmented to obtain a segmented velocity model with a preset number of segments; S33, identify the function change sensitivity of each segmented velocity model, and extract the feature points of each segmented velocity model based on the function change sensitivity of each segmented velocity model. The feature points of all segmented velocity models constitute multiple feature points of the fourth velocity model.
5. The method for controlling the opening degree of a vertical roll mill based on short-stroke control according to claim 4, characterized in that, S33 includes: S331, calculate the partial derivative of each time node in each segmented velocity model, and calculate the function change sensitivity of each time node based on the preset sensitivity coefficient and the partial derivative of each time node in each segmented velocity model; S332, calculate the mean value of the function change sensitivity of all piecewise velocity models, and take the time nodes in all piecewise velocity models where the function change sensitivity is greater than the mean value of the function change sensitivity as feature points; S333 combines the feature points of all segmented velocity models to obtain multiple feature points of the quartic velocity model.
6. The method for controlling the opening degree of a vertical roll mill based on short-stroke control according to claim 1, characterized in that, S4 includes: S41, obtain the time node corresponding to each feature point of the four speed models, and construct the matching relationship between each time node and the preset roll reduction, the preset horizontal roll reduction and the rolling information of each segment; S42, construct a nonlinear power function relationship between all feature points and the preset roll reduction, preset horizontal roll reduction and rolling information of each segment based on the matching relationship and all feature points; S43. Fit the nonlinear power function relationship to obtain the regression coefficient of the nonlinear power function relationship. Based on the regression coefficient and the nonlinear power function relationship, construct a regression prediction model for the deviation between multiple feature points and the vertical roll reduction, horizontal roll reduction and rolling information of each segment.
7. The method for controlling the opening degree of a vertical roll mill based on short-stroke control according to claim 1, characterized in that, S5 includes: S51, based on the deviation regression prediction model, predict the deviation of the width of the target workpiece at each future time node in vertical roll rolling; S52, if the deviation of the width of the target workpiece at any future time node is greater than the preset width deviation threshold, then the future time node is determined to be an abnormal point. S53. Determine the opening adjustment amount of the vertical roll mill based on the relationship between the deviation of the width of each abnormal point and the opening degree. Add the opening adjustment amount to the preset opening degree corresponding to the preset roll reduction amount to obtain the opening degree control amount. Control the opening degree of the vertical roll mill in vertical roll rolling through the opening degree control amount.