Optimized control method based on slab rough rolling width measurement

Abstract

The application discloses a kind of based on slab rough rolling width measurement optimization control method, including steps: 1, first width measuring instrument (11) of original arrangement at the first rolling mill R1 export is moved to the entrance E2 before second rolling mill R2;2, first width measuring instrument and second width measuring instrument measure slab width measurement data before and after the rolling of second rolling mill, and calculate slab width average according to slab width measurement data;3, utilize slab width average and reset rough rolling width control model;4, optimize the spread model of second rolling mill reverse pass;5, optimize the short stroke control of second rolling mill reverse pass;6, complete the rough rolling of slab, and optimize the model parameter of rough rolling horizontal rolling natural spread.The application can move the width measuring instrument at the first rolling mill export to the entrance before second rolling mill, and optimize rough rolling width model based on the measurement data of the width measuring instrument, quickly improve the precision of hot rolling width control index.

Description

Technical Field

[0001] This invention relates to a method for controlling hot rolling of strip steel, and more particularly to an optimized control method based on the measurement of slab roughing width. Background Technology

[0002] In hot strip steel production, the commonly used roughing mill configuration consists of two mills: a first mill (R1) and a second mill (R2). Width gauges are installed at the exits of both mills to measure the exit width of the slab. In existing roughing mill width control techniques, multi-pass rolling is employed, especially the second mill (R2), which typically uses 3 or 5 passes. This process presents the following problems:

[0003] 1. The width measuring instrument is installed at the exit of the second rolling mill R2. There is no measurement data during the reverse rolling of the second rolling mill R2. The variation of the roughing width includes the natural width expansion after horizontal roll pressing and the dog bone width expansion after vertical roll pressing. Multiple width expansion values ​​are superimposed and the model is uniformly genetically adjusted. It is impossible to adjust and optimize the different width expansion model parameters in a targeted and precise manner, resulting in poor control accuracy of the roughing width of the strip.

[0004] 2. The slab is relatively thick after the first pass of the first mill R1, which has a significant impact on the width measurement accuracy. The width measurement data of the first pass of the first mill R1 can generally only be used for reference and cannot be used for control. Reducing the width data is not conducive to the fine control of the roughing width.

[0005] 3. Short-stroke control of the strip head and tail during roughing is a unique control process. The width measuring instrument is generally located about 7m behind the mill exit. Hot rolling production process requires the slab to be rolled at the fastest speed. During the rolling process of the slab in non-final passes, the rolling speed of the slab will decrease and reverse rolling will begin as soon as the steel is thrown. Therefore, the width measuring instrument set after the exit of the first mill R1 and the second mill R2 can only measure the shape of the tail of the slab in the final pass. It is difficult to measure the effective data of the tail shape of the slab in non-final passes, which leads to a high incidence of tail width loss during roughing and increases the difficulty of controlling the width of the roughing. Summary of the Invention

[0006] The purpose of this invention is to provide an optimized control method based on slab roughing width measurement, which can move the width measuring instrument at the exit of the first mill to the front of the entrance of the second mill, and optimize the roughing width model based on the measurement data of the width measuring instrument, thereby rapidly improving the accuracy of hot rolling width control index.

[0007] This invention is implemented as follows:

[0008] An optimized control method based on slab roughing width measurement is applied to a roughing mill group including a first mill E1, a first mill R1, a second mill E2, and a second mill R2 arranged in sequence. A first width measuring instrument is arranged at the exit of the first mill R1, and a second width measuring instrument is arranged at the exit of the second mill R2.

[0009] The optimized control method based on slab roughing width measurement includes the following steps:

[0010] Step 1: Move the first width measuring instrument, originally located at the exit of the first rolling mill R1, to the front of the vertical rolling mill E2 of the second rolling mill;

[0011] The distance between the first width measuring instrument and the vertical roll mill E2 of the second rolling mill is 20-30m;

[0012] Step 2: The first and second width measuring instruments measure the width of the slab before and after rolling on the second rolling mill R2, and calculate the average width of the slab based on the width measurement data;

[0013] Step 3: Reset the control model for rough rolling width using the average slab width;

[0014] Step 4: Optimize the natural width spread model for horizontal rolling in the R2 reverse pass of the second rolling mill;

[0015] Step 5: Optimize the short-stroke control of the next pass in the second rolling mill R2;

[0016] Step 6: Complete the rough rolling of the slab and optimize the horizontal rolling natural width spread and dog bone width spread model parameters of the rough rolling.

[0017] In step 3, the formula for controlling the roughing width is as follows:

[0018] The slab entry width W of the i-th pass in The control model for (i) is:

[0019] W in (i)=W out (i-1)=W in (i-1)+dW Nature (i-1) Formula (1)

[0020] Among them, the reverse track vertical roller is not pressed down; W out (i-1) represents the slab exit width for the (i-1)th pass, W in (i-1) is the slab inlet width for the (i-1)th pass; dW Nature (i-1) represents the natural width of the horizontal rolling process in the (i-1)th pass;

[0021] The slab exit width W of the i-th passout The control model for (i) is:

[0022] W out (i)=W in (i)+dW Nature (i)+dW Dog (i)-dW edge (i) Formula (2)

[0023] Among them, dW Nature (i) represents the natural width of the horizontal rolling pass i; dW Dog (i) represents the width of the dog bone in the i-th pass; W in (i) represents the slab inlet width for the i-th pass; dW edge (i) represents the reduction amount of the vertical roller in the i-th pass;

[0024] Step 4 includes:

[0025] Step 4.1: Establish the measured width W of the slab exit for the (i+1)th pass. Act The model for (i+1) is as follows:

[0026] W Act (i+1)=W Act (i-1)+dW Nature (i)+dW Nature (i+1)+dW Dog (i+1)-dW edge (i+1) Formula (3)

[0027] Among them, W Act (i-1) represents the measured width of the slab exit in the (i-1)th pass, W Act (i-1) is the average width of the slab obtained from the (i-1)th pass;

[0028] dW Nature (i) represents the natural width of the horizontal rolling process in the i-th pass, dW Nature (i) Predicted by the natural width spread model of horizontal rolling;

[0029] dW Nature (i+1) represents the natural width of the horizontal rolling pass i+1, dW Nature (i+1) is predicted by the horizontal rolling natural width expansion model;

[0030] dW Dog (i+1) represents the dogbone width of the (i+1)th pass, dW Dog (i+1) is predicted by the dog bone width model;

[0031] dW edge(i+1) represents the reduction amount of the vertical roller in the (i+1)th pass;

[0032] Step 4.2: After obtaining the width of the reverse pass R2 of the second rolling mill, the horizontal natural width dW of the i-th pass is naturally expanded according to formula (4). Nature (i) model is used for learning;

[0033] W Act (i)=W Act (i-1)+dW Nature (i) Formula (4)

[0034] Among them, W Act (i) represents the measured width of the slab exit in the i-th pass, W Act (i) The average slab width obtained from the i-th pass calculation;

[0035] Step 4.3: Establish the natural width dW for horizontal rolling of the slab in the (i+1)th pass. Nature (i+1) and the natural width extension dW of the horizontal rolling of the i-th pass slab Nature (i) Model correction relation:

[0036] K Nature (i+1)=K Nature (i)*δ Formula (5)

[0037] Among them, K Nature (i+1) is the correction coefficient for the natural width expansion model of the horizontal rolling in the (i+1)th pass; K Nature (i) is the correction coefficient for the natural width spread model of the horizontal rolling in the i-th pass;

[0038] δ is a model correction parameter, ranging from 0.9 to 1.1, used to constrain K. Nature The range and direction of change of (i+1);

[0039] Step 4.4: Determine K Nature After determining the range and direction of change of (i+1), the dogbone width dW of the (i+1)th pass is obtained based on formula (6) using a nonlinear optimization method. Dog (i+1) model;

[0040] W Act (i+1)=W Act (i)+dW Nature (i+1)+dW Dog (i+1)-dW edge (i+1) Formula (6);

[0041] Step 5 includes:

[0042] Step 5.1: Obtain the tail width data of the reverse pass of the second rolling mill R2, and preprocess the tail width data to obtain statistics;

[0043] Step 5.2: Select several width feature control points within the width control range of the slab tail;

[0044] Step 5.3: Based on the width feature control points, calculate the vertical roll reduction amount for the next pass at each width feature control point;

[0045] Step 5.4: The control system combines the tail short-stroke control curve with the original tail short-stroke control curve to form a new tail short-stroke control curve, and performs width feedforward control for the next pass based on the new tail short-stroke control curve.

[0046] A hot metal detector is arranged in front of the vertical roll E2 of the second rolling mill, and the hot metal detector is located between the vertical roll E2 of the second rolling mill and the first width measuring instrument.

[0047] Step 2 includes:

[0048] Step 2.1: Perform frequency distribution processing on the slab width measurement data, and determine the slab width characteristics based on the frequency distribution characteristics;

[0049] Step 2.2: Perform amplitude limiting and speed limiting filtering on the slab width measurement data after frequency distribution processing, and remove unreliable data based on the width data characteristics;

[0050] Step 2.3: Determine the slab turning position based on the roller speed curve, and organize the slab width measurement data into symmetrical width measurement data at the same position;

[0051] Step 2.4: Compare the symmetrical width measurement data at the same location, that is, compare the width measurement data of two slabs at the same location. If the deviation between the two width measurement data of the slabs is greater than the deviation threshold, remove the symmetrical width measurement data to obtain the valid width data of the slab.

[0052] Step 2.5: Calculate the average slab width based on the valid slab width data.

[0053] The width control range is 2-5m from the tail end of the slab.

[0054] The method for selecting the width feature control points is as follows: within the width control range, take the maximum and minimum points of the slab width, and interpolate multiple points evenly between the maximum and minimum points. The maximum point, minimum point, and the evenly interpolated points are the width feature control points. The width feature control points are arranged according to their positions on the slab to form a control sequence point.

[0055] Compared with the prior art, the present invention has the following advantages:

[0056] 1. By moving the width measuring instrument at the exit of the first rolling mill to the front of the entrance of the second rolling mill, the present invention can measure as much strip width data as possible in the reverse pass of the second rolling mill through the rearward setting of the width measuring instrument. Thus, the rolling control model of the next pass can be re-set using the width data of the reverse pass of the previous pass, thereby improving the setting accuracy of roughing control.

[0057] 2. By moving the width measuring instrument at the exit of the first rolling mill to the front of the entrance of the second rolling mill, the present invention can measure the horizontal width data, and use the horizontal width data to correct the coefficients of the horizontal rolling natural width model and the dog bone width model using a nonlinear optimization method, thereby further improving the setting accuracy of roughing control.

[0058] This invention can move the width measuring instrument at the exit of the first rolling mill to the front of the entrance of the second rolling mill. By moving the width measuring instrument backward, the strip width of the reverse pass of the second rolling mill can be measured. Based on the measurement data of the width measuring instrument, the roughing width model can be re-set, and the width spread during the roughing process can be optimized, thereby establishing a brand-new roughing width control system and rapidly improving the accuracy of hot rolling width control indicators. Attached Figure Description

[0059] Figure 1 This is a schematic diagram showing the arrangement of the first width measuring instrument in the optimized control method based on slab roughing width measurement of the present invention;

[0060] Figure 2 This is a flowchart of the optimized control method based on slab roughing width measurement according to the present invention;

[0061] Figure 3 This is a flowchart of step 5 in the optimized control method based on slab roughing width measurement of the present invention;

[0062] Figure 4 This is a slab width distribution diagram for pass R22 in Example 1 of the optimized control method based on slab roughing width measurement of the present invention.

[0063] In the figure, 11 is the first width measuring instrument, 12 is the second width measuring instrument, 2 is the hot metal detector, R1 is the first rolling mill, R2 is the second rolling mill, E1 is the vertical roll mill of the first rolling mill, and E2 is the vertical roll mill of the second rolling mill. Detailed Implementation

[0064] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0065] Please see the appendix Figure 1An optimized control method based on slab roughing width measurement (hereinafter referred to as the optimized control method) is applied to a roughing mill group including a first mill vertical roll mill E1, a first mill R1, a second mill vertical roll mill E2, and a second mill R2 arranged in sequence. A first width measuring instrument 11 is arranged at the exit of the first mill R1, and a second width measuring instrument 12 is arranged at the exit of the second mill R2. Figure 1 The left side of the first rolling mill R1 shown is the inlet, and the right side is the outlet; Figure 1 The left side of the second rolling mill R2 shown is the inlet, and the right side is the outlet.

[0066] Please see the appendix Figure 2 The optimized control method includes the following steps:

[0067] Step 1: Move the first width measuring instrument 11, originally located at the exit of the first mill R1, to a position in front of the vertical roll mill E2 of the second mill. "In front" refers to the direction in which the vertical roll mill E2 of the second mill is closer to the first mill R1. After the first width measuring instrument 11 is moved to a rearward position, it can measure the full-length width data of the last pass rolled in the first mill R1, and simultaneously confirm whether the second mill R2 is biting. Therefore, the full-length width data of the last pass rolled in the first mill R1 can be used for the recalculation and setting of the second mill R2. Furthermore, after the first width measuring instrument 11 is moved to a rearward position, it can measure as much slab width data as possible along the length of the slab during reverse rolling in the second mill R2, including the width data of the slab tail.

[0068] Preferably, the distance between the first width measuring instrument 11 and the vertical roll mill E2 of the second rolling mill is 20-30m.

[0069] Preferably, the distance between the first width measuring instrument 11 and the vertical roll mill E2 of the second rolling mill is 25m.

[0070] Preferably, a hot metal detector (HMD) 2 is arranged in front of the vertical roll mill E2 of the second rolling mill, and the hot metal detector 2 is located between the vertical roll mill E2 of the second rolling mill and the first width measuring instrument 11. The distance between the hot metal detector (HMD) 2 and the vertical roll mill E2 of the second rolling mill is 6m.

[0071] During this rolling pass, before the slab head reaches the vertical roll mill E2 of the second mill, the process control system must perform a resetting calculation for this pass based on the measurement data from the previous pass. If the slab is relatively long at this point, after the slab head reaches HMD2, the process control system must collect relevant width measurement data, process the measured data, calculate the average slab width, and perform a resetting calculation for this pass. When the slab is rolled in the reverse pass R2 of the second mill, the first width measuring instrument 11 cannot measure the full length width of the slab. After the slab tail reaches HMD2, the slab begins to turn. At this time, the process control system will process the collected slab width data and perform a resetting calculation for subsequent passes, and initiate relevant model learning.

[0072] Step 2: The first width measuring instrument 11 and the second width measuring instrument 12 measure the slab width before and after rolling on the second rolling mill R2, and calculate the average slab width based on the measurement data. Since slab rolling is a reciprocating process, the slab width data includes two measurements at the same position on the strip. Utilizing the characteristic of symmetrical data at the same position, the slab width measurement data can be verified, improving the anti-interference capability of the slab width measurement data and effectively avoiding interference caused by factors such as descaling water on the mill stand to the width measuring instrument.

[0073] Step 2.1: Perform frequency distribution processing on the slab width measurement data and determine the slab width characteristics based on the frequency distribution features. Specifically, statistically, the slab width is counted according to different group intervals to form a probability distribution map. The slab width type is classified according to the slab width characteristics, mainly including T-shaped slabs, slabs with pits or bumps at the head and tail, and outliers in the width measurement data.

[0074] Step 2.2: Perform amplitude limiting and speed limiting filtering on the frequency-processed slab width measurement data, and remove unreliable data based on the width data characteristics. Specifically, a slab width threshold range and a change threshold can be set. Amplitude limiting filtering removes data in the slab width measurement data that are greater than or less than the width threshold range. Simultaneously, when the change in slab width at a certain measurement point compared to the previous measurement point exceeds the change threshold, speed limiting filtering removes the slab width measurement data at that point.

[0075] Step 2.3: Determine the slab turning position based on the roller speed curve, and organize the slab width measurement data into symmetrical width measurement data for the same position. In the existing slab rolling system, including basic automation system L1, process control system L2, and other control equipment, the roller speed curve can be directly measured by the basic automation system L1 and sent to the process control system L2. The process control system L2 defines the point where the slab speed is 0 as the slab turning position.

[0076] Step 2.4: Compare symmetrical width measurement data at the same location. That is, compare the width measurement data of two slabs at the same location. If the deviation between the two width measurement data is greater than the deviation threshold, remove the symmetrical width measurement data to obtain the valid width data of the slab. The deviation threshold is determined based on data statistics and is generally taken as 0-10mm.

[0077] Step 2.5: Calculate the average slab width based on the valid slab width data.

[0078] Step 3: Re-establish the control model for the rough rolling width using the average slab width. The model formula is as follows:

[0079] Slab entrance width (mm) W for the i-th pass in The control model for (i) is:

[0080] W in (i)=W out (i-1)=W in (i-1)+dW Nature (i-1) Formula (1)

[0081] Among them, the reverse track vertical roller is not pressed down; W out (i-1) is the slab exit width (mm) for the (i-1)th pass; W in (i-1) is the slab entry width (mm) for the (i-1)th pass; dW Nature (i-1) represents the natural width (mm) of the horizontal rolling pass i-1. W out (i-1) can be calculated using step 2, which is the average width of the slab in the (i-1)th pass, and is used to improve the setting accuracy.

[0082] Slab exit width (mm) W of the i-th pass out The control model for (i) is:

[0083] W out (i)=W in (i)+dW Nature (i)+dW Dog (i)-dW edge (i) Formula (2)

[0084] Among them, dW Nature (i) represents the natural width (mm) of the horizontal rolling pass i.

[0085] dW Dog (i) represents the width of the dog bone in the i-th pass (mm);

[0086] dW edge (i) represents the vertical roll reduction (mm) for the i-th pass.

[0087] Step 4: Optimize the natural width development model of the horizontal rolling process in the R2 reverse pass of the second rolling mill. The slab width measurement data for the R2 reverse pass of the second rolling mill is obtained by moving the first width measuring instrument 11 backward, thereby enabling the optimization of the natural width development model of the horizontal rolling process in the R2 reverse pass. Furthermore, since the vertical rolls are not pressed down during the reverse pass rolling, the optimization of the reverse pass only includes the natural width development model of the horizontal rolling process. Depending on the amount of data, optimization can be performed for each reverse pass of the R2 rolling mill.

[0088] Step 4.1: Establish the measured width W of the slab exit for the (i+1)th pass. Act The model for (i+1) is as follows:

[0089] W Act (i+1)=W Act (i-1)+dW Nature (i)+dW Nature (i+1)+dW Dog (i+1) - dW edge (i+1) Formula (3)

[0090] Among them, W Act (i-1) represents the measured width (mm) of the slab exit in the (i-1)th pass, W Act (i-1) can be calculated in step 2, which is the average width of the slab in the (i-1)th pass.

[0091] dW Nature (i) represents the natural width (mm) of the horizontal rolling pass i; dW Nature (i+1) represents the natural width (mm) of the horizontal rolling pass (i+1), dW Nature (i) and dW Nature (i+1) are all predicted by the width control module in the process control system L2 based on the natural width expansion model of horizontal rolling.

[0092] dW Dog (i+1) represents the dogbone width (mm) of the (i+1)th pass, dW Dog (i+1) is predicted by the width control module in process control system L2 based on the dog bone width expansion model.

[0093] dW edge (i+1) represents the reduction amount of the vertical roller in the (i+1)th pass.

[0094] Step 4.2: After obtaining the width of the reverse pass R2 of the second rolling mill, the natural width dW of the horizontal rolling of the i-th pass is expanded according to formula (4). Nature (i) is used to learn the model.

[0095] W Act(i)=W Act (i-1)+dW Nature (i) Formula (4)

[0096] Among them, W Act (i) represents the measured width (mm) of the slab exit in the i-th pass; W Act (i) can be calculated in step 2, that is, the average width of the slab in the i-th pass.

[0097] Step 4.3: Due to the natural width expansion dW of the horizontal rolling in the (i+1)th pass. Nature The property of (i+1) is related to the natural width dW of the horizontal rolling in the i-th pass. Nature (i) has similar properties. A modified relationship is established between the (i+1)th pass and the i-th pass for the natural width expansion model of horizontal rolling:

[0098] K Nature (i+1)=K Nature (i)*δ Formula (5)

[0099] Among them, K Nature (i+1) is the correction coefficient for the natural width spread model of horizontal rolling in the (i+1)th pass;

[0100] K Nature (i) is the correction coefficient for the natural width spread model of the horizontal rolling in the i-th pass;

[0101] δ is the model correction parameter, and K is... Nature (i) Multiply by a constant, namely the model correction parameter δ, and apply this constant to K. Nature The range and direction of change of (i+1) are restricted. Since the natural width expansion properties of horizontal rolling in adjacent passes are similar, δ is generally taken as 0.9-1.1.

[0102] Step 4.4: Determine K Nature After determining the range and direction of change of (i+1), based on formula (6), a more accurate dogbone width dW for the (i+1)th pass can be obtained through nonlinear optimization. Dog (i+1) model.

[0103] W Act (i+1)=W Act (i)+dW Nature (i+1)+dW Dog (i+1) -dW edge (i+1) Formula (6)

[0104] Among them: W Act (i+1) represents the measured width of the slab exit in the (i+1)th pass, W Act (i) represents the measured width of the slab exit in the i-th pass, dWedge (i+1) represents the reduction amount of the vertical roller in the (i+1)th pass.

[0105] Step 5: Optimize the short-stroke control of the next pass of the second rolling mill R2.

[0106] Please see the appendix Figure 3 Step 5.1: Obtain the tail width data of the reverse pass of the second rolling mill R2, and preprocess the tail width data to obtain statistical quantities such as the tail width mean and tail width variance.

[0107] Step 5.2: Select several width feature control points within the width control range of the slab tail.

[0108] The width control range is 2-5m from the tail end of the slab, preferably 4m.

[0109] The method for selecting the width feature control points is as follows: within the width control range, take the maximum and minimum points of the slab width, and interpolate multiple points evenly between the maximum and minimum points (preferably 10 points). The maximum point, minimum point, and the evenly interpolated points are the width feature control points. The width feature control points are arranged according to their positions on the slab to form a control sequence point.

[0110] Step 5.3: Based on the width characteristic control points, calculate the vertical roll reduction amount for the next pass at each width characteristic control point to form the tail short-stroke control curve. The vertical roll reduction amount can be calculated using the vertical roll reduction setting technique in short-stroke control. The vertical roll reduction setting technique is one of the common techniques in short-stroke control, and will not be elaborated here.

[0111] Step 5.4: The control system combines the tail short-stroke control curve with the original tail short-stroke control curve to form a new tail short-stroke control curve, and performs width feedforward control for the next pass based on the new tail short-stroke control curve.

[0112] Step 6: Complete the rough rolling of the slab and optimize the horizontal rolling natural width spread and dog bone width spread model parameters of the rough rolling.

[0113] Step 6.1: Establish the dog bone width expansion model after horizontal rolling. The model formula is: dW Dog =f(W in W Out H in ,R).

[0114] Among them, W in H represents the slab entrance width (mm). in R represents the thickness of the rolled piece before side pressing (mm), and R represents the radius of the horizontal roll (mm).

[0115] Step 6.2: Establish a natural width spread model for horizontal rolling. The model formula is:

[0116] Among them, T a T is the temperature effect coefficient. a = -0.0004*T+K T T represents the rolling temperature (°C), K represents the temperature of the rolled piece. T For model parameters. H in H represents the thickness of the rolled piece before side pressing. out H represents the thickness of the rolled piece after side pressing. in and H out All were obtained through model calculations from the process control system. W in The width of the slab entrance (mm).

[0117] α is the size influence coefficient. K A K B K C These are the model parameters.

[0118] Step 6.3: The width W of the slab after being rolled by the vertical and horizontal rolling mill rolls. out It can consist of the following three items:

[0119] W out =W in -dW edge +dW Nature +dW Dog Formula (7)

[0120] Among them, dW edge This represents the amount of pressure applied to the vertical roller.

[0121] Example 1:

[0122] Taking a strip of steel rolled on-site as an example, the rolling mode of this strip of steel is 3-3 passes.

[0123] Strip slab width: Width = 1162.56 mm;

[0124] Strip slab thickness: Thick = 229.27 mm;

[0125] Rough rolling target width: RmTarget = 1062.4652 mm.

[0126] The horizontal roll reduction (mm) for each pass of R2 is: R2Redu={35.726,34.767,28.501};

[0127] The heritability coefficients of natural spread in each pass of horizontal rolling in R2 are: R2Spreadinher={1.1298311,1.3415148,1.3423822}.

[0128] After steel extraction, the model's pre-calculation results are as follows:

[0129] SpTarget width: 1034.7926mm;

[0130] SP dog bone width (SpDog): 22.848873mm;

[0131] SP reduction SpRedu: 127.76926mm.

[0132] Target width (mm) for each positive pass in R1: R1Target = {1055.4652, 1058.7299};

[0133] R2 target width (mm) for each positive pass: R2Target={1053.5704,1064.4652}.

[0134] R1 vertical roll reduction (mm) for each pass: E1Draft={15, 26.233};

[0135] R2 vertical roll reduction (mm) for each pass: E2Draft={42.888,29.843}.

[0136] Natural width spread (mm) of horizontal rolling in each pass of R1: dW Nature (R1)={6.343, 8.893, 6.742};

[0137] Natural width spread (mm) of each pass of horizontal rolling in R2: dW Nature (R2)={9.942, 11.093, 9.125};

[0138] R1 lane dogbone width (mm): dW Dog (R1) = {6.48, 13.96};

[0139] R2 lane dogbone width (mm): dW Dog (R2) = {28.21, 21.28}.

[0140] After the R21 pass rolling is completed, the measured width W of the slab exit for the R21 pass is obtained. Act (R21) = 1054.2363mm. Since the measured width is reliable, recalculation is triggered to calculate the natural width spread, dog-bone width spread, vertical roll reduction, and target width of subsequent passes. The calculation results are as follows:

[0141] R22 pass horizontal rolling natural width extension: dW Nature (R22) = 10.9589 mm;

[0142] R23 pass horizontal rolling natural width extension: dW Nature (R23) = 8.8235 mm;

[0143] R23 pass vertical roll reduction: dW edge (R23) = 32.02 mm;

[0144] R23 Lane Dogbone Wide Extension: dW Dog (R23) = 22.79 mm.

[0145] This recalculation shows that after the R22 rolling passes are completed, the exit width W calculated by the model is... Act (R22)=W Act (R21)+dW Nature (R22) = 1065.1952 mm.

[0146] Appendix: Slab width measurement data for R22 pass Figure 4 As shown, data processing is performed on the slab width measurement data for pass R22. The data is symmetrical about the slab width; based on this characteristic, a dedicated data processing module is used to process the width measurement data.

[0147] After the data processing module removes outliers, performs frequency domain processing, and compares data from the same location on the slab width measurement data of pass R22, the average slab width W of pass R22 is obtained. Act (R22)=1062.5mm, then the actual horizontal rolling width of the strip in R22 passes is dW(R22)=W Act (R22)-W Act (R21) = 8.264 mm.

[0148] Based on the average width W of the slab Act (R22) Recalculate and control the R23 passage. W Act Substituting (R22) into the width load distribution model and recalculating, the vertical roll reduction and dog-bone width expansion for pass R23 are calculated as follows:

[0149] R23 pass vertical roll reduction: dW edge (R23) = 15.8862mm, which can greatly improve the control accuracy of the roughing mill exit width.

[0150] R23 Lane Dogbone Wide Extension: dW Dog (R23) = 10.965 mm.

[0151] The R22 pass can only obtain the width data of the head of the reverse pass. This width data is the tail width data of the strip during forward rolling. Using this width data, width feedforward control is performed to optimize the short stroke curve at the tail.

[0152] During the preliminary calculation, within the strip width control range, the maximum point of 1500mm and the minimum point of 0mm of the slab width are taken, and 8 points are evenly interpolated between the maximum and minimum points to form 10 width characteristic control points. The short stroke setting value (mm) of the R23 strip is shown in Table 1:

[0153] Table 1. Short-stroke setting values ​​(mm) for R23 strip passes

[0154] Width feature control points Short stroke setting value Width feature control points Short stroke setting value 1500 4 666.667 3 1333.33 4 500 1 1166.67 3 333.333 1 1000 3 166.667 1 833.333 3 0 0

[0155] From the appendix Figure 4 It can be seen that there is a width loss at the tail of the strip, and short-stroke feedforward control is needed at the tail.

[0156] Rolling data was collected continuously for one year. Based on data such as steel grade, horizontal roll reduction of R22 pass, actual horizontal rolling natural width spread, and dog-bone width spread, nonlinear optimization methods were used to optimize the coefficients of the roughing rolling horizontal width spread model.

[0157] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An optimization control method based on slab rough rolling width measurement, which is applied to a rough rolling mill train comprising a edger mill El, a first rolling mill Rl, a second edger mill E2 and a second rolling mill R2 arranged in sequence, a first width gauge (11) is arranged at the outlet of the first rolling mill Rl, and a second width gauge (12) is arranged at the outlet of the second rolling mill R2. The application is characterized in that The optimization control method based on slab rough rolling width measurement comprises the following steps: Step 1: moving the first width gauge (11) originally arranged at the outlet of the first rolling mill Rl to the front of the second edger mill E2; The distance between the first width gauge (11) and the second edger mill E2 is 20-30 m; Step 2: measuring the slab width measurement data before and after rolling of the second rolling mill R2 by the first width gauge (11) and the second width gauge (12), and calculating the average value of the slab width according to the slab width measurement data; Step 3: resetting the control model of the rough rolling width by using the average value of the slab width; Step 4: optimizing the horizontal rolling natural spread model of the reverse pass of the second rolling mill R2; Step 5: optimizing the short stroke control of the next pass of the second rolling mill R2; Step 6: completing the rough rolling of the slab, and optimizing the horizontal rolling natural spread and dog bone spread model parameters of the rough rolling; In step 3, the control model formula of the rough rolling width is as follows: the slab entry width W of the i-th pass in The control model of (i) is: W in (i) = W out (i-1) = W in (i-1) + dW Nature (i-1) Equation (1) wherein the reverse pass vertical roll is not pressed down; W out (i-1) is the slab exit width of the i-1th pass, W in (i-1) is the slab entry width of the i-1th pass; dW Nature (i-1) is the horizontal rolling natural spread of the i-1th pass; The slab exit width W of the i-th pass out The control model of (i) is: W out (i) = W in (i) + dW Nature (i) + dW Dog (i) - dW edge (i) Equation (2) wherein dW Nature (i) is the natural spread of the i-th pass of horizontal rolling; dW Dog (i) is the dog-bone spread of the i-th pass; W in (i) is the slab entry width of the i-th pass; dW edge (i) is the stand roll reduction of the i-th pass; Step 4 comprises: Step 4.1: Establish the measured width W of the slab at the exit of the i+1 pass Act (i+1) of the model, the model formula is: Act (i+1) = W Act (i-1) + dW Nature (i) + dW Nature (i+1) + dW Dog (i+1) - dW edge (i+1) Formula (3) wherein W Act (i-1) is the measured width of the slab at the exit of the i-1th pass, W Act (i-1) is the average value of the slab width calculated using the i-1th pass. dW Nature (i) is the natural spread of the i-th pass of horizontal rolling, dW Nature (i) is predicted from the model of natural spread of horizontal rolling; dW Nature (i+1) is the horizontal rolling natural spread of the i+1 pass, dW Nature (i+1) is predicted by the horizontal rolling natural spread model; dW Dog (i+1) is the dog bone width spread for the i+1 pass, dW Dog (i+1) is predicted from the dog bone width spread model; dW edge (i+1) is the roll setting of the i+1 pass Step 4.2: After obtaining the second rolling mill R2 reverse pass width, the horizontal natural spread dW of the i-th pass is calculated according to formula (4) Nature learning from the model of (i); W Act (i) = W Act (i-1) + dW Nature (i) Equation (4) wherein W Act (i) is the measured slab exit width for the i-th pass, W Act (i) is the average slab width calculated using the i-th pass Step 4.3: Establishing the horizontal rolling natural spread dW of the (i+1)th pass slab Nature (i+1)th pass slab with the horizontal rolling natural spread dW of the ith pass slab Nature Model correction relationship of (i): K Nature (i+1) = K Nature (i) * δ Equation (5) wherein K Nature (i+1) is the horizontal rolling natural spread model correction factor for the i+1 pass; K Nature (i) is the horizontal rolling natural spread model correction factor for the i pass; δ is a model correction parameter, δ takes 0.9-1.1, used to limit K Nature the change range and change direction of (i+1); Step 4.4: Determining K Nature After the range and direction of variation of dW(i+1) are determined, the dog bone width spread dW(i+1) of the i+1th pass is obtained by a non-linear optimization method based on equation (6) Dog (i+1) model; W Act (i+1) = W Act (i) + dW Nature (i+1) + dW Dog (i+1) - dW edge (i+1) Equation (6); Step 5 comprises: Step 5.1: obtaining the tail width data of the reverse pass of the second rolling mill R2, and pre-processing the tail width data to obtain statistical quantities; Step 5.2: selecting several width characteristic control points within the control range of the tail width of the slab; Step 5.3: calculating the edger roll reduction of the next pass at each width characteristic control point based on the width characteristic control points; Step 5.4: the control system synthesizes the tail short stroke control curve with the originally set tail short stroke control curve to obtain a new tail short stroke control curve, and performs width feedforward control of the next pass according to the new tail short stroke control curve.

2. The slab rough rolling width measurement based optimization control method according to claim 1, characterized in that in A hot metal detector (2) is arranged in front of the second edger mill E2, and the hot metal detector (2) is located between the second edger mill E2 and the first width gauge (11).

3. The slab rough rolling width measurement based optimization control method according to claim 1, characterized in that: Step 2 comprises: Step 2.1: performing frequency distribution processing on the slab width measurement data, and determining the width characteristics of the slab according to the frequency distribution characteristics; Step 2.2: performing amplitude limiting filtering and speed limiting filtering processing on the slab width measurement data after frequency distribution processing, and removing unreliable data according to the width data characteristics; Step 2.3: determining the slab turning position according to the roll speed curve, and arranging the slab width measurement data into symmetrical width measurement data at the same position; Step 2.4: comparing the symmetrical width measurement data at the same position, i.e. comparing the two slab width measurement data at the same position, if the deviation of the two slab width measurement data is greater than the deviation threshold, removing the symmetrical width measurement data, and obtaining the effective slab width data; Step 2.5: calculating the average value of the slab width according to the effective slab width data.

4. The slab rough rolling width measurement based optimization control method according to claim 1, characterized in that: The width control range is 2-5 m from the tail end of the slab.

5. The slab rough rolling width measurement based optimization control method according to claim 1, characterized by: The selection method of the width feature control points is as follows: in the width control range, the maximum point and the minimum point of the slab width are taken, and multiple points are evenly interpolated between the maximum point and the minimum point, and the maximum point, the minimum point and the evenly interpolated points are the width feature control points, and the width feature control points are sorted according to the positions on the slab to form the control sequence points.

Images