Snake control device

Abstract

This invention provides a serpentine control device that can control the flow of raw material while maintaining wedge quality and achieving stable plate passage even when a continuous rolling mill is continuously rolling raw material. The serpentine control device includes: a prediction calculation unit that predicts the serpentine amount and wedge amount of the raw material tail end after passing through each of the multiple mill stands based on a rolling model for each of the multiple mill stands, the rolling model using an initial serpentine amount, a final serpentine amount, a reduction leveling setpoint for each of the multiple mill stands, and rolling information; a setpoint update unit that updates the reduction leveling setpoint for each of the multiple mill stands based on the serpentine amount and wedge amount of the raw material tail end predicted by the prediction calculation unit; and multiple reduction leveling control units that control the reduction leveling of the multiple mill stands based on the updated reduction leveling setpoints from the setpoint update unit.

Description

TECHNICAL FIELD

[0001] The present application relates to a snake control device. BACKGROUND

[0002] In a continuous rolling mill such as a hot finishing mill having a plurality of rolling stands, a snake phenomenon in which the width direction center of a workpiece (rolling material) in rolling moves in the direction of the operation side or the drive side of the continuous rolling mill from the width direction center of the roll is known.

[0003] In the continuous rolling mill, the tail end of the workpiece loses tension after being rolled by the next rolling stand downstream when passing through an arbitrary rolling stand. At this time, the tail end of the workpiece rotates due to the difference in extension between the left and right (operation side and drive side) at the entry side of the rolling stand, moves in the width direction just below the roll, and the tail end generates a snake phenomenon.

[0004] Just below the roll, the load on the side where the snake phenomenon has occurred increases, the elastic deformation of the rolling mill on the side where the snake phenomenon has occurred increases, the gap of the roll on the side where the snake phenomenon has occurred opens, and the thickness of the plate increases. As a result, the difference in extension between the left and right of the workpiece is amplified, and the snake phenomenon is further exacerbated.

[0005] On the exit side of the rolling stand, the workpiece that has occurred the snake phenomenon bends and enters the downstream rolling stand. In the continuous rolling mill, the snake phenomenon occurs in each rolling stand, and thus the snake phenomenon can develop rapidly. If the tail end collides with the side guide plate provided at the entry side of the rolling stand, the workpiece that has occurred the snake phenomenon is bent and directly rolled, and a fold is generated. The fold can damage the roll and cause the work to be interrupted such as roll replacement, and thus the production efficiency is reduced.

[0006] The conventional snake control technology for the rolling mill is broadly classified into a method of dynamically controlling the roll gap based on the rolling difference load and the amount of snake detected by a camera provided between the rolling stands, and a method of setting the roll gap in advance based on the past rolling results.

[0007] As the method of dynamically controlling the roll gap, a method of detecting or estimating the snake using the rolling difference load for a long time ago and using the camera provided between the stands in recent years, and controlling the roll gap by feedback calculation is generally used.

[0008] For example, in Patent Literature 1, a method using an optimal regulator is proposed, which estimates the amount of snake and the entry angle of the workpiece using a state observer based on the rolling difference load, and controls the roll gap in such a manner that an evaluation function having the amount of snake, the integral value of the amount of snake, the entry angle, and the roll gap as variables becomes minimum.

[0009] Further, the difference in the extension of the rolling material to the right and left that causes the snake is generated due to a change in the wedge ratio (a difference in the thickness of the plate on the working side and the driving side is referred to as a wedge, and a value obtained by dividing the wedge by the thickness of the plate is referred to as a wedge ratio) of the entry side and the exit side of the rolling stand, and thus a control method that aims at the change in the wedge ratio to achieve suppression of the snake has been proposed.

[0010] For example, in Patent Literature 2, a method is proposed in which the wedge is measured on the exit side of the final rolling stand, and the reduction schedule is controlled so that the measured wedge ratios become equal for each rolling stand, and the snake is suppressed while reducing the wedge.

[0011] As a method of presetting the reduction schedule, for example, in Patent Literature 3, a method is proposed in which, based on the relationship between the reduction schedule and the amount of warping (an amount of bending of the main body with respect to the head and tail ends is referred to as the amount of warping) of each rolling stand, a reduction schedule correction amount of each rolling stand is calculated for the material (a rolling material that is the object of setting of the reduction schedule) and the preceding material (a rolling material that is rolled immediately before), and the reduction schedule is set based on the difference.

[0012] Patent Literature 1: Japanese Patent No. 5790636

[0013] Patent Literature 2: Japanese Patent No. 6044194

[0014] Patent Literature 3: Japanese Patent No. 3664068

[0015] In the method described in Patent Literature 1, the reduction schedule changes the exit-side wedge so as to rotate the rolling material in a direction in which the snake is reduced by the change in the wedge ratio.

[0016] However, the changed exit-side wedge is the entry-side wedge from the viewpoint of the next rolling stand downstream. When the next rolling stand downstream rolls the rolling material, rotation in a direction in which the reduced snake is restored occurs.

[0017] Therefore, if the influence from the upstream rolling stand or the influence on the downstream rolling stand is not taken into account, the snake control as a whole of the rolling stands becomes unstable.

[0018] Further, in the method described in Patent Literature 2, in the case where the wedge ratio is not uniform in the longitudinal direction of the rolling material, the wedge ratio measured on the exit side of the final rolling stand is different from the wedge ratio of the rolling material located directly below the roll of each rolling stand, and thus it can be impossible to perform effective control.

[0019] In the case of the tail end portion of the rolled material, the time of stabilization by feedback control or the length of the rolled material is not sufficient, and therefore the snake cannot be suppressed. In addition, the wedge ratio is the ratio of the plate thickness to the wedge, and therefore a large reduction schedule is required in the upstream rolling stands in which the plate thickness is thick. In this case, due to mechanical limitations of the device, the reduction schedule cannot be controlled sufficiently, and the target wedge ratio cannot be achieved. As a result, the wedge ratio can be inconsistent with the downstream rolling stands, and the snake can occur.

[0020] In addition, in the method described in Patent Document 3, the reduction schedule of each rolling stand is determined independently with respect to the amount of warping, and the influence of the upstream rolling stands on the downstream rolling stands cannot be considered. In addition, in the snake phenomenon at the tail end, the snake develops exponentially with respect to time or the rolling distance, and the snake and the wedge interference occur in each rolling stand.

[0021] In order to predict more accurately, it is desirable to explicitly calculate the snake and the wedge after rolling for each rolling stand using a rolling model, and to simulate the development based on time or the rolling distance.

[0022] In addition, in the above-described related art, there is no means to automatically correct the influence coefficient table between the models and the variables used based on the rolling results, and it can be impossible to cope with changes over time, and manual-based model adjustment needs to be repeated. SUMMARY

[0023] The present application was made in order to solve the above-described problems, and an object thereof is to provide a snake control device that can control to achieve stable plate passing while maintaining the wedge quality even if a continuous rolling mill continuously rolls a raw material.

[0024] The snake control device of one aspect of the present application is a device that controls the snake of a raw material in a continuous rolling mill that has a plurality of rolling stands through which the raw material passes and to which a reduction force is applied in order, and is characterized by including: a prediction calculation section that predicts the amount of snake and the amount of wedge of each of tail end portions of the raw material after passing through each of the plurality of rolling stands based on a rolling model of each of the plurality of rolling stands, the rolling model using an initial amount of snake of the raw material before rolling, a final amount of snake of the raw material after rolling by all of the rolling stands, a reduction schedule setting value for each of the plurality of rolling stands, and rolling information that determines a setting of a rolling process; a setting value update section that updates the reduction schedule setting value for each of the plurality of rolling stands based on the amount of snake and the amount of wedge of each of the tail end portions of the raw material predicted by the prediction calculation section; and a plurality of reduction schedule control sections that control the reduction schedule of each of the plurality of rolling stands based on the updated reduction schedule setting value of each of the plurality of rolling stands by the setting value update section.

[0025] Further, the snake control device of one aspect of the present application is characterized in that, preferably, the set value updating section repeatedly evaluates, using an evaluation function based on the snake amount and the wedge amount predicted by the prediction calculation section, the set value for the roll gap adjustment of each of the plurality of rolling stands until a prescribed end condition is satisfied, and updates the set value for the roll gap adjustment of each of the plurality of rolling stands so as to become a roll gap adjustment set pattern of a time series or a rolling distance series.

[0026] Further, the snake control device of one aspect of the present application is characterized in that, preferably, the set value updating section repeatedly evaluates, using an evaluation function based on the snake amount and the wedge amount predicted by the prediction calculation section, the set value for the roll gap adjustment of each of the plurality of rolling stands until a prescribed end condition is satisfied, and updates the set value for the roll gap adjustment of each of the plurality of rolling stands so as to become a roll gap adjustment set pattern of a time series or a rolling distance series.

[0027] Further, the snake control device of one aspect of the present application is characterized in that, preferably, the set value updating section repeatedly evaluates, using an evaluation function based on the snake amount and the wedge amount predicted by the prediction calculation section, the set value for the roll gap adjustment of each of the plurality of rolling stands until a prescribed end condition is satisfied, and updates the set value for the roll gap adjustment of each of the plurality of rolling stands so as to become a roll gap adjustment set pattern of a time series or a rolling distance series.

[0028] Further, the snake control device of one aspect of the present application is characterized in that, preferably, the set value updating section repeatedly evaluates, using an evaluation function based on the snake amount and the wedge amount predicted by the prediction calculation section, the set value for the roll gap adjustment of each of the plurality of rolling stands until a prescribed end condition is satisfied, and updates the set value for the roll gap adjustment of each of the plurality of rolling stands so as to become a roll gap adjustment set pattern of a time series or a rolling distance series.

[0029] Further, the snake control device of one aspect of the present application is characterized in that preferably, further comprising: a learning value storage section that stores a learning value used for adjustment of the rolling model; and a learning value update section that updates the learning value stored in the learning value storage section, the plurality of roll gap adjustment control sections respectively acquire a time series or a rolling distance series of a roll gap adjustment actual pattern of the work from the plurality of rolling stands, the prediction calculation section predicts the amount of snake and the amount of wedge of each of the tail end portions of the work after passing through each of the plurality of rolling stands based on the rolling model of each of the plurality of rolling stands, the rolling model uses the initial amount of snake, the final amount of snake, each of the roll gap adjustment actual patterns acquired by the plurality of roll gap adjustment control sections, the rolling information, and the learning value stored in the learning value storage section, and the learning value update section updates the learning value stored in the learning value storage section using the amount of snake of the work after passing through the most downstream rolling stand predicted by the prediction calculation section and the final amount of snake.

[0030] Further, the snake control device of one aspect of the present application is characterized in that preferably, the learning value update section repeatedly evaluates the learning value for each of the plurality of rolling stands using an evaluation function that uses the amount of snake of the work after passing through the most downstream rolling stand predicted by the prediction calculation section and the final amount of snake until a prescribed end condition is satisfied.

[0031] Further, the snake control device of one aspect of the present application is characterized in that preferably, the learning value update section updates the learning value so that the error between the amount of snake of a prescribed point of the tail end portion of the work after passing through the most downstream rolling stand predicted by the prediction calculation section and the final amount of snake becomes 0.

[0032] Further, the snake control device of one aspect of the present application is characterized in that preferably, the learning value update section updates the learning value by evaluating the amount of snake detected by an inter-stand snake amount detection section that detects the amount of snake of the work between the rolling stand and another rolling stand and the amount of snake predicted by the prediction calculation section until a prescribed end condition based on the evaluation result is satisfied.

[0033] Effects of the Invention

[0034] According to the present application, even if a continuous rolling mill continuously rolls a work, it is possible to control so as to realize stable plate passing while maintaining the quality of the wedge. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is a side view schematically showing a configuration example of a rolling system of one embodiment.

[0036] Figure 2 is a diagram showing a configuration example of a snake control device and its peripheral configuration of one embodiment.

[0037] Figure 3 is a flowchart showing a process in which the snake control device repeatedly performs calculation in the setting step of the first operation example.

[0038] Figure 4 is a diagram schematically showing a press-down leveling setting pattern in the first operation example of the snake control device.

[0039] Figure 5 is a flowchart showing a process in which the snake control device repeatedly performs calculation in the learning step of the first operation example.

[0040] Figure 6 is a diagram schematically showing a tension state at the time of rolling in the third operation example of the snake control device.

[0041] Figure 7 is a diagram schematically showing characteristics at each position of a rolled material after rolling in the third operation example of the snake control device.

[0042] Figure 8 is a diagram showing a configuration example of a snake control device and its peripheral configuration in a modification example of a rolling system.

[0043] Figure 9 is a flowchart showing a process in which the snake control device repeatedly performs calculation in the learning step.

[0044] Figure 10 is a conceptual diagram showing a hardware configuration example of a processing circuit possessed by the snake control device.

[0045] Explanation of symbols

[0046] 1: rolled material, 2: initial snake amount detection unit, 3: final snake amount detection unit, 4: inter-stand snake amount detection unit, 10: continuous rolling mill, 11: work roll, 12: backup roll, 13: press-down control unit, 20, 20a: snake control device, 21: prediction calculation unit, 22: set value update unit, 23: press-down leveling control unit, 24, 24a: learning value update unit, 25: learning value storage unit, 30: setting device, 91: processor, 92: memory, 93: hardware, 100, 100a: rolling system, F1 to F N : rolling stand DETAILED DESCRIPTION

[0047] Hereinafter, one embodiment of a rolling system will be described with reference to the drawings. Figure 1 is a side view schematically showing a configuration example of a rolling system 100 of one embodiment. As shown in FIG. 1, the rolling system 100 includes a continuous rolling mill 10, a setting device 30, and a snake control device 20. Figure 1As shown, the rolling system 100 includes, for example, a continuous rolling mill 10, a serpentine control device 20, and a setting device 30.

[0048] The multi-stage continuous rolling mill 10 is equipped with mill stands F1, F2, ..., F... N To ensure that the rolled material (raw material) 1 passes through sequentially and a reducing force is applied to it at each stand, N is a natural number greater than 2. That is, the rolled material 1 passes through the mill stands F1, F2, ..., F1 sequentially. N Move in the way (in) Figure 1 (Moving from left to right), it is rolled to the specified plate thickness.

[0049] Each rolling mill stand F i (1≦i≦N) It ​​has two upper and lower working rollers 11 and two upper and lower support rollers 12. The gap between the upper and lower working rollers 11 can be adjusted by pressing devices (not shown) provided on the working side and driving side of the support rollers 12 respectively.

[0050] The pressing control unit 13 obtains the pressing leveling setting value from the serpentine control device 20 and changes the pressing leveling (the difference in pressing amount between the working side and the drive side) performed by the pressing device to the obtained pressing leveling setting value. In addition, the pressing control unit 13 outputs the actual pressing leveling performed by the pressing device as a pressing leveling actual pattern of time series or rolling distance series to the serpentine control device 20 (the pressing leveling control unit 23 described later).

[0051] The initial snake amount detection unit 2 is located upstream of the mill stand F1. It detects the snake amount, i.e., the initial snake amount, of the rolled material 1 before rolling and outputs the detected initial snake amount to the snake control device 20. For example, the initial snake amount detection unit 2 detects the deviation of the center position of the rolled material 1 in the width direction from the center position of the continuous mill 10 in the width direction, and uses this as the initial snake amount.

[0052] The final snake-like quantity detection unit 3 is installed on the specific rolling mill stand F. N At a downstream position, the final snake-like amount (or "snake-like amount") of the rolled material 1 after being rolled by all the mill stands is detected, and the detected final snake-like amount is output to the snake-like control device 20. The final snake-like amount detection unit 3 detects the deviation of the center position of the rolled material 1 in the width direction relative to the center position of the continuous mill 10 in the width direction, and uses this as the final snake-like amount. Furthermore, the location of the final snake-like amount detection unit 3 is not limited to the exit side of the continuous mill 10, and can also be set in other downstream processes.

[0053] The snake control device 20 calculates the press leveling set value for each roll stand based on the rolling information input from the setting device 30, the initial snake amount input from the initial snake amount detection section 2, and the final snake amount input from the final snake amount detection section 3, and outputs the same to the press control section 13, respectively.

[0054] Next, a configuration example of the snake control device 20 of one embodiment will be described. Figure 2 is a view showing a configuration example of the snake control device 20 of one embodiment and its surrounding configuration. Hereinafter, the same symbol is given to the configuration substantially the same as the above-described configuration.

[0055] The snake control device 20 includes a prediction calculation section 21, a set value update section 22, N number of press leveling control sections 23, a learning value update section 24, and a learning value storage section 25.

[0056] The prediction calculation section 21 predicts the snake amount and the wedge amount of the tail end portion of the rolling material 1 after each of the plurality of roll stands based on a rolling model for each of the plurality of roll stands, which uses the snake amount of the rolling material 1 before rolling, i.e., the initial snake amount, the snake amount of the rolling material 1 after rolling by all of the roll stands, i.e., the final snake amount, the press leveling set value for each of the plurality of roll stands, and the rolling information that determines the settings of the rolling process.

[0057] Further, the prediction calculation section 21 has a function of predicting the snake amount and the wedge amount of the tail end portion of the rolling material 1 after each of the plurality of roll stands based on a rolling model for each of the plurality of roll stands, which uses the initial snake amount, the final snake amount, the respective press leveling actual patterns acquired by the plurality of press leveling control sections 23, the rolling information, and the learning value stored in the learning value storage section 25.

[0058] The set value update section 22 updates the press leveling set value for each of the plurality of roll stands based on the snake amount and the wedge amount of the tail end portion of the rolling material 1 predicted by the prediction calculation section 21.

[0059] Further, the set value update section 22 has a function of repeatedly evaluating the press leveling set value for each of the plurality of roll stands based on an evaluation function using the snake amount and the wedge amount predicted by the prediction calculation section 21 until a prescribed end condition is satisfied, so that the press leveling set value for each of the plurality of roll stands becomes a press leveling set pattern in a time series or a rolling distance series.

[0060] Further, the set value updating section 22 has a function of setting the roll gap adjustment set value to a certain value, repeatedly evaluating using an evaluation function based on the snake amount and the wedge amount of the tail end portion of the rolled material 1 after passing through the most downstream rolling stand predicted by the prediction operation section 21, and updating the roll gap adjustment set value for each of the plurality of rolling stands until a prescribed end condition is satisfied.

[0061] Further, the set value updating section 22 has a function of sequentially updating the roll gap adjustment set value for each of the plurality of rolling stands from the upstream side so that the snake amount of a prescribed point of the tail end portion of the rolled material 1 after passing through the most downstream rolling stand predicted by the prediction operation section 21 becomes 0.

[0062] Further, the set value updating section 22 has a function of sequentially evaluating, for each of the plurality of rolling stands from the upstream side, using an evaluation function based on the snake amount and the wedge amount of a prescribed point of the tail end portion of the rolled material 1 after passing through the most downstream rolling stand predicted by the prediction operation section 21, and updating the roll gap adjustment set value for each of the plurality of rolling stands based on the evaluation result until a prescribed end condition is satisfied.

[0063] The roll gap adjustment control section 23 controls the roll gap adjustment of the plurality of rolling stands respectively based on the respective roll gap adjustment set values updated by the set value updating section 22.

[0064] Further, the roll gap adjustment control section 23 has a function of controlling the roll gap adjustment of the rolling stands based on the roll gap adjustment pattern, the time, and the entry side speed of the rolled material 1 updated by the set value updating section 22.

[0065] The learning value updating section 24 updates the learning value used in the adjustment of the rolling model stored in the learning value storage section 25.

[0066] Further, the learning value updating section 24 has a function of updating the learning value stored in the learning value storage section 25 using the snake amount of the rolled material 1 after passing through the most downstream rolling stand and the final snake amount predicted by the prediction operation section 21.

[0067] Further, the learning value updating section 24 has a function of repeatedly evaluating using an evaluation function using the snake amount of the rolled material 1 after passing through the most downstream rolling stand and the final snake amount predicted by the prediction operation section 21, and updating the learning value for each of the plurality of rolling stands until a prescribed end condition is satisfied.

[0068] Further, the learning value updating section 24 has a function of updating the learning value so that the error between the snake amount of a prescribed point of the tail end portion of the rolled material 1 after passing through the most downstream rolling stand predicted by the prediction operation section 21 and the final snake amount becomes 0.

[0069] Next, the functions and actions of each part of the snake control device 20 will be described in detail. The actions of the snake control device 20 include a setting step, a control step, and a learning step.

[0070] In the setting step, the snake control device 20 starts to act after the tail end of the rolled material 1 passes through the initial snake amount detection section 2, and the setting value updating section 22 calculates the roll gap leveling setting value, and outputs the calculation result to the roll gap leveling control section 23 to end the action.

[0071] In the control step, the snake control device 20 controls each rolling stand at different timings.

[0072] For example, the roll gap leveling control section 23 corresponding to the rolling stand Fl starts to act after the position of the rolled material 1 at a distance Lt from the tail end passes directly below the work roll 11 of the rolling stand Fl. Then, the roll gap leveling control section 23 operates (controls) the roll gap control section 13 according to the roll gap leveling setting value or the roll gap leveling setting pattern, and ends the action after the tail end of the rolled material 1 passes directly below the work roll 11 of the rolling stand Fl.

[0073] The roll gap leveling control section 23 corresponding to the rolling stand F j (2 ≦ j ≦ N) starts to act after the tail end of the rolled material 1 passes through the immediately upstream rolling stand F j-1 and then starts to change the roll gap leveling.

[0074] In the learning step, the snake control device 20 starts to act after the tail end of the rolled material 1 passes through the final snake amount detection section 3, and the learning value updating section 24 calculates the update of the learning value, and stores the calculation result in the learning value storage section 25 to end the action.

[0075] (First Action Example)

[0076] Figure 3 is a flowchart showing the process of repeated calculation by the snake control device 20 in the setting step. In addition, Figure 3 the dashed box shown in the figure indicates the process performed by the prediction calculation section 21.

[0077] At the start of the setting step, when the setting value updating section 22 issues a read command to the prediction calculation section 21, the prediction calculation section 21 acquires the rolling information from the setting device 30, acquires the initial snake amount y c0 from the learning value storage section 25, and calculates each coefficient of the rolling model for each rolling stand (S100).

[0078] The coefficients of the rolling model are coefficients of influence of the input or current state on the future state. The rolling model uses, for example, a dynamic model represented by a state space model shown by the following equations (1) to (5).

[0079] [Num 1]

[0080] x i [k+1] = A i x i [k] + B i u i [k]... (1)

[0081] [Num 2]

[0082] z i [k] = C i x i [k] + D i u i [k]... (2)

[0083] [Num 3]

[0084]

[0085] [Num 4]

[0086]

[0087] [Num 5]

[0088]

[0089] x i : by rolling mill stand F i state vector related to the rotational motion in rolling

[0090] u i : by rolling mill stand F i input vector related to the rolling material before rolling and the roll gap adjustment

[0091] z i : by rolling mill stand F i output vector related to the rolling material after rolling

[0092] Ay ct : amount of wobble at the roll gap changed by rolling mill stand F i

[0093] A0 i : entry side rotational angle of the rolling material changed by rolling mill stand F i

[0094] y ci : rolling mill stand F i ​​wandering amount of the work roll directly below

[0095] δh i : rolling mill stand F i wedge amount of the work roll directly below

[0096] δs t : rolling mill stand F i set value of the gap adjustment

[0097] A i , B i , C i , D i : coefficient matrix calculated from the rolling information of the rolling mill stand F i and the learning value

[0098] Here, k is a parameter that is incremented by 1 from a value 1 indicating the start of the operation of the rolling mill stand F i to a value M indicating the end of the operation. i

[0099] (2) The output vector z i of the formula is represented by (4) and is constituted by the wandering amount and the wedge amount of the work roll 11 directly below. The input vector u i is represented by (5) and is constituted by the output vector z i-1 of the immediately preceding rolling mill stand F i-1 upstream calculated by (2) and the gap adjustment set value δS i of the rolling mill stand F i .

[0100] However, in the case of the rolling mill stand Fl located at the most upstream, the initial wandering amount y i-1 is brought to z c0 , and the initial wedge amount is assumed to be 0.

[0101] Here, the gap adjustment set pattern δS i is a gap adjustment set value that is varied in accordance with the parameter k shown in (3). Figure 4

[0102] Further, the state equation of (1) is not limited to a differential equation or a difference equation that develops in accordance with time, but can also adopt a differential equation or a difference equation that develops in accordance with the rolling distance. In the case of the state equation that develops in accordance with the rolling distance, there are advantages that useless time due to the transportation and the time variation of the entry side speed can be avoided, and the state space model can be linearized.

[0103] Here, with respect to the learning value, a stable modeling error can also be considered, for example, is imparted by a vector added to the right side of (2). Further, the rolling model is not limited to a dynamic model, but can also be a linear model or the like.​​

[0104] The set value update section 22 updates the set values δS1, δS2,..., δS ini，1 , δS ini，2 ,..., δS ini，N to the prediction operation section 21 and issues a prediction command (S102). As the initial set value, for example, an average value of the set down leveling actual pattern of an arbitrary section of the stable section in each rolling stand is given.

[0105] The prediction operation section 21 receives the prediction command and performs operation using the rolling model from the rolling stand Fl in order based on the input condition and the rolling model, and outputs the snake amount y c1 , y c2 ,..., y cN and the wedge amount δ h1 , δ h2 ,..., δ hN to the set value update section 22 after the rolling of each rolling stand (S104).

[0106] The set value update section 22 evaluates the obtained snake amount after the rolling of each rolling stand and the wedge amount after the rolling of the final rolling stand (S106). The set value update section 22, for example, uses an evaluation function shown in the following formula (6) in the evaluation.

[0107]

Formula 6

[0108]

[0109] α: evaluation weight of the snake amount

[0110] w i : evaluation weight of the rolling stand F i

[0111] Then, the set value update section 22 determines whether the evaluation value satisfies an end condition. The end condition is, for example, a condition that is judged based on whether the evaluation value of the previous time and this time is a value or less that is set in advance (S108).

[0112] The set value update section 22, in a case where the end condition is not satisfied, updates the set pattern δS1, δS2,..., δS N of the set down leveling of each rolling stand based on the evaluation value, and issues a prediction command again to the prediction operation section 21 (S110).

[0113] For example, the update based on the evaluation value can be realized by calculation using a quasi-Newton method or the like that is known in an optimization method.

[0114] ​Then, the snake control device 20 repeatedly calculates the prediction, evaluation, and update until a prescribed end condition is satisfied, to determine the press-down leveling set patterns δS1, δS2,..., δS N .

[0115] Further, in the repeated calculation, the mechanical limit of the press-down device can also be considered, and the optimization with a limit can be performed, or a compensation term related to the press-down leveling can be added in the equation (6).

[0116] The set value update section 22 outputs the determined press-down leveling set patterns to the press-down leveling control sections 23 corresponding to the respective rolling stands.

[0117] The press-down leveling control section 23 corresponding to the rolling stand Fi stores the press-down leveling set pattern acquired in the setting step, and reads out the press-down leveling set value δS i [1] at the start of the control step.

[0118] Then, the press-down leveling control section 23 calculates the elapsed time or the rolling distance from the start of the control, and follows the press-down leveling set pattern δS i [1], δS i [2],..., δS i [M i ] to sequentially operate the press-down control section 13.

[0119] Further, the sequential operation of the press-down control section 13 can be advanced, or compensated by a filter based on a transfer function of the dynamic characteristics, considering the response delay or the dynamic characteristics of the press-down device.

[0120] Figure 5 is a flowchart showing the process of the repeated calculation of the snake control device 20 in the learning step. Further, Figure 5 The process performed by the prediction operation section 21 is shown in the dashed box.

[0121] The press-down leveling control section 23 corresponding to the respective rolling stands acquires the press-down leveling actual patterns δS1, δS2,..., δS N of the time series or the rolling distance series from the press-down control section 13 at the start of the learning step, and outputs them to the learning value update section 24.

[0122] The learning value update section 24 acquires the initial snake amount y c0 from the initial snake amount detection section 2, the final snake amount y cM from the final snake amount detection section 3, and the learning value from the learning value storage section 25, after acquiring the press-down leveling actual patterns from the press-down leveling control section 23 and the rolling information from the setting device 30, and issues a read command to the prediction operation section 21.

[0123] Furthermore, the learning value update unit 24 inputs the actual pressure leveling pattern obtained from the pressure leveling control unit 23 to the prediction calculation unit 21 and issues a prediction command.

[0124] The prediction calculation unit 21 first receives the read command, obtains the learning value from the learning value storage unit 25, and calculates each coefficient of the rolling model of each mill stand based on the obtained learning value and the rolling information input from the setting device 30 (S200).

[0125] However, in subsequent read commands, the prediction calculation unit 21 uses the learning value assigned by the learning value update unit 24 instead of obtaining the learning value from the learning value storage unit 25.

[0126] Next, the prediction calculation unit 21 receives the prediction command and, based on the input conditions and the rolling model, performs calculations using the rolling model sequentially, starting from mill stand F1, to predict the rolling stand F1. N The serpentine amount y after rolling cN Output to the learning value update unit 24 (S202).

[0127] Learning value update unit 24 uses the acquired rolling mill stand F N The serpentine amount y after rolling cN and the final serpentine amount y measured at the corresponding position of the rolled material. cM The prediction error is evaluated (S204). For example, the learning value update unit 24 uses the evaluation function shown in equation (7) for evaluation.

[0128]

Number 7

[0129]

[0130] Then, the learning value update unit 24 determines whether the evaluation value meets the termination condition. The termination condition is, for example, a condition based on whether the previous and current evaluation values ​​are below a preset value (S206).

[0131] If the termination condition is not met, the learning value update unit 24 updates the learning value based on the evaluation value, inputs it into the prediction calculation unit 21, and issues a read command and a prediction command again (S208).

[0132] Updates based on evaluation values ​​can be achieved, for example, by using quasi-Newton methods, which are well-known in optimization methods.

[0133] Then, the snake control device 20 repeatedly calculates the prediction, evaluation, and update until the specified termination conditions are met, thereby determining the learning value.

[0134] Finally, the learning value update unit 24 proportionally allocates the learning value determined as described above and the first acquired learning value, and updates the learning value stored in the learning value storage unit 25 as the final learning value. In this way, the serpentine control device 20 determines the optimal learning value for this rolling result.

[0135] (Example of the second action)

[0136] In the second action example of the snake control device 20, instead of the pressure leveling setting pattern in the first action example, a certain pressure leveling setting value is set in the control step.

[0137] The dynamic model changes equation (5) to equation (8).

[0138]

Number 8

[0139]

[0140] In the setting step, the setting value update unit 22 updates the setting value according to the pressing and leveling setting patterns δS1, δS2, ..., δS of each mill stand as described in the first operation example. N The decision also determines to lower the leveling setpoints δS1, δS2, ..., δS N It outputs to the pressing and leveling control unit 23 of each rolling mill stand.

[0141] With rolling mill stand F i The corresponding pressure-down leveling control unit 23, at the start of the control step, determines the pressure-down leveling setting value δS obtained from the setting value update unit 22. i For rolling mill stand F i The pressing control unit 13 controls the operation. Alternatively, the operation of the pressing control unit 13 can be initiated earlier, taking into account the response delay of the pressing device.

[0142] (Example of the third action)

[0143] In the third action example of the serpentine control device 20, the calculations in the setting steps described in the first action example are performed sequentially from the upstream mill stand, thereby reducing the calculation cost.

[0144] Figure 6 This is a diagram schematically illustrating the tension state during rolling in the third action example of the serpentine control device 20. Figure 6 In the example of mill stands F1 to F3, the range of rolled material 1 is represented by each mill stand of the continuous mill 10 rolling material without back tension.

[0145] Rolled material 1 is rolled from upstream to downstream to increase the thickness of the inlet plate by H. i With the thickness h of the side plate i The ratio of Hi / h i The elongation λ represents i Extending this further, the tension state of the rolled material 1 varies depending on the position of its tail end during rolling by each mill stand.

[0146] First, the sections rolled by each mill stand in a manner without back tension ( Figure 6 The diagonal interval in the rolling mill stands (F2) is from the tail end to L, and in the rolling mill stands (F1) it is from the tail end to Lt.

[0147] Next, regarding the length Lt on the inlet side of mill stand F1, since the length extends to λ1×Lt on the outlet side, the interval on the inlet side of mill stand F2, from a position λ1×Lt away from the tail end to a position L away from the tail end ( Figure 6 The middle horizontal section is rolled without rear tension.

[0148] Similarly, the section L from the tail end is rolled by mill stand F2 in a manner without back tension ( Figure 6 The diagonal interval in the middle is also, since it extends to λ2×L, therefore, the interval on the F3 side of the rolling mill stand, starting from a position λ2×L from the tail end and ending at a position L from the tail end ( Figure 6 The middle horizontal section is rolled without rear tension.

[0149] As a result, the tension state of each mill stand changes at various locations of the rolled material 1. Furthermore, the description has been given up to mill stand F3, but the description continues up to mill stand F... N The same phenomenon will continue to occur until then.

[0150] Figure 7 This diagram schematically illustrates the characteristics (tension state) of the rolled material 1 at various positions in the third operation example of the serpentine control device 20. The number of mill stands is N as described above. Intervals 1 to N are intervals distinguished by the tension state of each position of the rolled material 1 during rolling at each mill stand.

[0151] Interval i is from mill stand F1 to F i Rolling is performed without back tension. Therefore, the serpentine movement in interval i is affected by the mill stands F1 to F2. i The effect of the pressure reduction leveling can be ignored, while the effect of the rolling mill stand F can be ignored. i+1 ~F N The effect of pressure reduction and leveling.

[0152] As a result of the above relationship, the roll reduction leveling setting value δS1 of the mill stand F1 is determined to be, for the mill stand F predicted by the prediction calculation unit 21 N The serpentine amount y after rolling cNy = 0 at the position of the boundary between the interval 1 and the interval 2, and cN becomes 0. That is, the setting value updating section 22 performs the process of updating to the decided roll gap leveling setting pattern δS1, and issues the prediction command again.

[0153] Here, the prediction calculation section 21 performs the prediction by changing equation (2) to equation (9) assuming that the rotation will not occur due to the change in the wedge rate in the rolling mill stand that performs the rolling with the rear tension.

[0154] [Equation 9]

[0155] z i [k] = D i u i [k]... (9)

[0156] Thus, for the interval 1, the rolling is performed with the rear tension after the rolling mill stand F2, and thus the predicted value of the amount of the serpentine of each rolling mill stand becomes y c1 = y c2 =... = y cN .

[0157] Further, the roll gap leveling of the rolling mill stand before the start of the control step shows the roll gap leveling initial setting value explained in the first example of operation.

[0158] Next, the roll gap leveling setting value δS2 of the rolling mill stand F2 is set to, for the rolling mill stand F N the amount of the serpentine y cN , at the position of the boundary between the interval 2 and the interval 3, and cN becomes 0. The setting value updating section 22 decides the roll gap leveling setting values δS1,..., δS N .

[0159] (Fourth example of operation)

[0160] In the fourth example of operation of the serpentine control device 20, the repeated calculation in the setting step explained in the third example of operation is performed for the rolling mill stands from the upstream, and the roll gap leveling setting value is decided also considering the amount of the wedge after the rolling of the final rolling mill stand, thereby improving the quality of the wedge.

[0161] The setting value updating section 22 evaluates the amount of the serpentine after the rolling of the final rolling mill stand and the amount of the wedge predicted by the prediction calculation section 21 for the roll gap leveling δS i of the rolling mill stand Fi, focusing on the boundary position between the interval i and the interval i+1 of the rolled material 1, using the evaluation function of equation (10).

[0162] [Equation 10]

[0163]

[0164] y cN，i The snake-like amount of the final mill stand after rolling at the boundary position between interval i and interval i+1.

[0165] δh N，i The wedge shape after final mill stand rolling at the boundary position between interval i and interval i+1.

[0166] Then, the serpentine control device 20 continues to operate on the mill stand F until the specified termination condition is met. i The pressure leveling setting value δS i The decision is made through changes and repeated re-prediction. Furthermore, i is in the order of 1 to N.

[0167] For example, when i=3, the setting value update unit 22 inputs the determined setting values ​​to the prediction calculation unit 21 for the pressure-lowering leveling setting values ​​δS1 and δS2, and inputs the determined setting values ​​to the prediction calculation unit 21 for the pressure-lowering leveling setting values ​​δS4 to δS2. N The initial settings are input into the prediction calculation unit 21.

[0168] (Example of the fifth action)

[0169] In the fifth operation example of the serpentine control device 20, the computational cost is reduced by repeatedly performing the learning steps described in the first operation example, starting from the upstream mill stand. For example, the learning value update unit 24 calculates the learning value sequentially from the mill stand F1, starting from interval 1, using the error between the serpentine amount predicted by the prediction calculation unit 21 and the detected serpentine amount.

[0170] The learning value for mill stand F1 is determined as follows: for the mill stand F predicted by the prediction calculation unit 21... N The serpentine amount y after rolling cN The serpentine amount at the boundary between interval 1 and interval 2 and y cM As long as they are consistent. Then, the learning value update unit 24 updates the learning value of the determined mill stand F1 and issues the read command and prediction command again.

[0171] Next, the learning value for mill stand F2 is determined as follows: for mill stand F... N The serpentine amount y after rolling cN The serpentine amount at the boundary between interval 2 and interval 3 and y cM Consistency is required. Then, the learning value update unit 24 determines the learning value sequentially from the upstream mill stand using the same steps. Finally, the learning value update unit 24 proportionally allocates the learning values ​​of each mill stand determined as described above, along with the first acquired learning value, as the final learning value, and updates the learning value stored in the learning value storage unit 25.

[0172] Thus, the snake control device 20 controls the roll gap adjustment of the plurality of roll stands based on the updated individual roll gap adjustment set values of the set value updating unit 22, and thus even if the continuous rolling mill 10 continuously rolls the rolled material 1, it is possible to control so as to achieve stable plate passing while maintaining the wedge quality.

[0173] Next, a modified example of the rolling system 100 will be described. Figure 8 is a drawing showing a configuration example of a snake control device 20a of a modified example (rolling system 100a) of the example rolling system 100 and its periphery.

[0174] The rolling system 100a differs from the rolling system 100 (refer to Figure 1 , 2 ) in that, for example, there is one or more inter-stand snake amount detection units 4 between prescribed roll stands. For example, the rolling system 100a is provided with one inter-stand snake amount detection unit 4 between the roll stands F i-1 and F i . Further, the position of the inter-stand snake amount detection unit 4 can be any inter-stand position of the roll stands F1 to F N . Further, the number of the inter-stand snake amount detection units 4 is at most N-1.

[0175] Then, the snake control device 20a determines the learning value in a manner to improve the accuracy of the rolling model using the inter-stand snake amount detected by the inter-stand snake amount detection unit 4.

[0176] For example, the learning value updating unit 24a further has a function of evaluating the snake amount detected by the inter-stand snake amount detection unit 4 and the snake amount predicted by the prediction operation unit 21 by an evaluation function, and updating the learning value until a prescribed end condition based on the evaluation result is satisfied.

[0177] In the first operation example of the above snake control device 20, the prediction operation unit 21 predicts the snake amount y c0 from the initial snake amount y N to the snake amount y cN after the rolling of the roll stand F cN , and the learning value updating unit 24a calculates the learning value in a manner to make the snake amount y cM consistent with the final snake amount y i-1 .

[0178] In this case, the error of the rolling model of the upstream roll stand propagates to the downstream, and it is possible to affect the prediction of the downstream roll stand.

[0179] The snake control device 20a calculates the learning value in a manner to make the snake amount y i-1 consistent with the final snake amount y iIn the case between, the rolling mill stands F1 to F i-1 Divided into the upstream side, the rolling mill stand F i ~F N The learning value is repeatedly calculated by dividing the learning value into downstream sides.

[0180] The F detected by the inter-rack snake detection unit 4 i-1 -F i The serpentine quantity is represented by y si F i-1 -F i Inter-serpentine quantity y si Used in rolling mill stand F i The position of the rolled material 1, which is rolled with rear tension, for example... Figure 7 The interval i-1 in the middle, and the rolling mill stand F i-1 The serpentine amount y after rolling ci-1 Compare them.

[0181] Figure 9 This is a flowchart illustrating the iterative calculation process performed by the serpentine control device 20a during the learning process. In this iterative calculation, firstly, for the upstream mill stands F1 to F... i-1 Using rolling mill stand F i-1 The serpentine amount y after rolling ci-1 And the F measured at the corresponding position of rolled material 1 i-1 -F i Inter-serpentine quantity y si The learned value is updated in the same way as the first action example (upstream side processing: S300~S308).

[0182] Here, the evaluation function is changed from Equation (7) to Equation (11) and used.

[0183]

Number 11

[0184]

[0185] Here, for F i-1 -F i Inter-serpentine quantity y si Pre-matched with rolling mill stand F i-1 The parameter k is resampled using linear interpolation and based on the mill stand F. i-1 The distance between the work roll 11 directly below and the snake-like detection position, and the mill stand F i-1 -F i The speed of transport between them causes it to deflect.

[0186] The evaluation interval is between the rolling mill stand F i-1In the relevant rolling model, the learning value is calculated from k = 1 corresponding to the boundary of the interval i-2 and the interval i-1 to k = M corresponding to the boundary of the interval i-1 and the interval i i-1’ until M i-1’ is calculated by the following expression (12). Here, Int() is a function of discarding the decimal point. Further, the interval 0 means the stable section.

[0187] [Num 12]

[0188]

[0189] Then, the learning value updating section 24a updates the learning value for the rolling mill stands F1 to F i-1 Using the learning value obtained as described above, the learning value for the rolling mill stands on the downstream side is calculated. The learning value updating section 24a updates the learning value for the rolling mill stands F i to F N , using the learning value for the rolling mill stands F N the amount of the snake y cN after the rolling, and the final amount of the snake y cM , and the learning value is updated as in the first example of operation. Finally, the learning value updating section 24a proportionally distributes the learning value for each rolling mill stand determined as described above and the learning value obtained for the first time, and updates the learning value stored in the learning value storage section 25 as the final learning value (downstream processing: S400 to S408).

[0190] Thus, the snake control device 20a updates the learning value stored in the learning value storage section 25 using the amount of the snake of the rolled material 1 after passing through the most downstream rolling mill stand and the final amount of the snake predicted by the prediction calculating section 21, and therefore, even when the continuous rolling mill 10 continuously rolls the rolled material 1 in a case where a change over time is likely to occur, it is possible to control so as to achieve stable passage while maintaining the wedge quality.

[0191] Next, a hardware configuration example of the snake control device 20 (or the snake control device 20a) will be described. Figure 10 is a conceptual diagram showing a hardware configuration example of a processing circuit possessed by the snake control device 20 (or the snake control device 20a). Each section in the snake control device 20 represents a part of a function, and each function is realized by a processing circuit.

[0192] As one scheme, the processing circuit is provided with at least one processor 91 and at least one memory 92. As another scheme, the processing circuit is provided with at least one dedicated hardware 93.

[0193] In a case where the processing circuitry has the processor 91 and the memory 92, each function is implemented by software, firmware, or a combination of software and firmware. At least one of the software and the firmware is described as a program. Furthermore, at least one of the software and the firmware is stored in the memory 92.

[0194] The processor 91 implements each function by reading out and executing the program stored in the memory 92.

[0195] In a case where the processing circuitry has the dedicated hardware 93, the processing circuitry is, for example, a single circuit, a composite circuit, a programmed processor, or a combination of these. Then, each function is implemented by the processing circuitry.

[0196] The above describes the embodiments of the present application, but the present application is not limited to the above-described embodiments, and can be implemented in various modifications without departing from the spirit of the present application.

Claims

1. A serpentine control device for controlling the serpentine movement of raw material in a continuous rolling mill, the continuous rolling mill having multiple mill stands through which the raw material passes and for which a reducing force is applied sequentially, the serpentine control device being characterized by having: The prediction calculation unit predicts the meandering amount and wedge amount of the raw material tail end after passing through each of the multiple rolling mill stands, based on the rolling models of each of the multiple rolling mill stands. The rolling model uses the meandering amount of the raw material before rolling, i.e. the initial meandering amount, the meandering amount of the raw material after being rolled by all the multiple rolling mill stands, i.e. the final meandering amount, the reduction and leveling setting value for each of the multiple rolling mill stands, and the rolling process setting to determine the rolling information. The setting update unit updates the reduction and leveling setting values ​​for each of the multiple mill stands based on the serpentine amount and wedge amount of the raw material tail end predicted by the prediction calculation unit; and Multiple reduction and leveling control units control the reduction and leveling of the multiple rolling mill stands respectively, based on the reduction and leveling setting values ​​updated by the aforementioned setting value update unit. The aforementioned setting update unit repeatedly evaluates the snake-like amount and wedge-shaped amount predicted by the aforementioned prediction calculation unit using an evaluation function, and updates the reduction and leveling setting values ​​for each of the multiple mill stands until a predetermined termination condition is met, so that the reduction and leveling setting values ​​for each of the multiple mill stands become a reduction and leveling setting pattern of a time series or rolling distance series. The aforementioned pressure reduction and leveling control unit controls the pressure reduction and leveling of the mill stand based on the pressure reduction and leveling setting pattern, time, and raw material feed speed updated by the aforementioned setting value update unit.

2. A serpentine control device for controlling the serpentine movement of raw material in a continuous rolling mill, the continuous rolling mill having multiple mill stands through which the raw material passes and for which a reducing force is applied sequentially, the serpentine control device being characterized by having: The prediction calculation unit predicts the meandering amount and wedge amount of the raw material tail end after passing through each of the multiple rolling mill stands, based on the rolling models of each of the multiple rolling mill stands. The rolling model uses the meandering amount of the raw material before rolling, i.e. the initial meandering amount, the meandering amount of the raw material after being rolled by all the multiple rolling mill stands, i.e. the final meandering amount, the reduction and leveling setting value for each of the multiple rolling mill stands, and the rolling process setting to determine the rolling information. The setting update unit updates the reduction and leveling setting values ​​for each of the multiple mill stands based on the serpentine amount and wedge amount of the raw material tail end predicted by the prediction calculation unit; and Multiple reduction and leveling control units control the reduction and leveling of the multiple rolling mill stands respectively, based on the reduction and leveling setting values ​​updated by the aforementioned setting value update unit. The aforementioned setting update unit sets the pressure reduction and leveling setting value to a certain value, and repeatedly evaluates the snake amount and wedge amount predicted by the aforementioned prediction calculation unit, updating the pressure reduction and leveling setting value for each of the aforementioned mill stands until the specified termination condition is met.

3. A serpentine control device for controlling the serpentine movement of raw material in a continuous rolling mill, the continuous rolling mill having multiple mill stands through which the raw material passes and for which a reducing force is applied sequentially, the serpentine control device being characterized by having: The prediction calculation unit predicts the meandering amount and wedge amount of the raw material tail end after passing through each of the multiple rolling mill stands, based on the rolling models of each of the multiple rolling mill stands. The rolling model uses the meandering amount of the raw material before rolling, i.e. the initial meandering amount, the meandering amount of the raw material after being rolled by all the multiple rolling mill stands, i.e. the final meandering amount, the reduction and leveling setting value for each of the multiple rolling mill stands, and the rolling process setting to determine the rolling information. The setting update unit updates the reduction and leveling setting values ​​for each of the multiple mill stands based on the serpentine amount and wedge amount of the raw material tail end predicted by the prediction calculation unit; and Multiple reduction and leveling control units control the reduction and leveling of the multiple rolling mill stands respectively, based on the reduction and leveling setting values ​​updated by the aforementioned setting value update unit. The aforementioned setting update unit evaluates each of the multiple mill stands sequentially from the upstream side using an evaluation function based on the snake-like amount and wedge-shaped amount predicted by the aforementioned prediction calculation unit at a predetermined point after the raw material tail end of the most downstream mill stand. Based on the evaluation results, the unit updates the pressure reduction and leveling setting values ​​for each of the multiple mill stands until the predetermined termination conditions are met.

4. The serpentine control device as described in any one of claims 1 to 3, characterized in that, It also has: The learning value storage unit stores the learning values ​​used for adjusting the aforementioned rolling model; and The learning value update unit updates the learning values ​​stored in the learning value storage unit. The aforementioned reduction and leveling control units respectively obtain the actual reduction and leveling patterns of the time series or rolling distance series from the aforementioned rolling mill stands. The aforementioned prediction calculation unit predicts the meandering amount and wedge shape of the raw material tail end after passing through each of the multiple mill stands based on the rolling models of each of the multiple mill stands. The rolling model uses the initial meandering amount, the final meandering amount, the actual reduction and leveling patterns obtained by the multiple reduction and leveling control units, the rolling information, and the learning values ​​stored in the learning value storage unit. The learning value update unit updates the learning value stored in the learning value storage unit using the snake-like amount of the raw material after passing the downstream mill stand and the final snake-like amount predicted by the prediction calculation unit.

5. A serpentine control device for controlling the serpentine movement of raw material in a continuous rolling mill, the continuous rolling mill having multiple mill stands through which the raw material passes and for which a reducing force is applied sequentially, the serpentine control device being characterized by having: The prediction calculation unit predicts the meandering amount and wedge amount of the raw material tail end after passing through each of the multiple rolling mill stands, based on the rolling models of each of the multiple rolling mill stands. The rolling model uses the meandering amount of the raw material before rolling, i.e. the initial meandering amount, the meandering amount of the raw material after being rolled by all the multiple rolling mill stands, i.e. the final meandering amount, the reduction and leveling setting value for each of the multiple rolling mill stands, and the rolling process setting to determine the rolling information. The setting update unit updates the reduction and leveling setting values ​​for each of the multiple mill stands based on the snake-like amount and wedge-shaped amount of the raw material tail end predicted by the prediction calculation unit. Multiple reduction leveling control units control the reduction leveling of multiple mill stands respectively based on the reduction leveling setting values ​​updated by the aforementioned setting value update unit. The learning value storage unit stores the learning values ​​used for adjusting the above-mentioned rolling model; as well as The learning value update unit updates the learning values ​​stored in the learning value storage unit. The aforementioned reduction and leveling control units respectively obtain the actual reduction and leveling patterns of the time series or rolling distance series from the aforementioned rolling mill stands. The aforementioned prediction calculation unit predicts the meandering amount and wedge shape of the raw material tail end after passing through each of the multiple mill stands based on the rolling models of each of the multiple mill stands. The rolling model uses the initial meandering amount, the final meandering amount, the actual reduction and leveling patterns obtained by the multiple reduction and leveling control units, the rolling information, and the learning values ​​stored in the learning value storage unit. The learning value update unit repeatedly evaluates the learning value for each of the multiple mill stands by using the evaluation function of the serpentine amount of the raw material after passing the downstream mill stand, which is predicted by the prediction calculation unit, and the final serpentine amount, until the specified termination condition is met.

6. A serpentine control device for controlling the serpentine movement of raw material in a continuous rolling mill, the continuous rolling mill having multiple mill stands through which the raw material passes and for which a reducing force is applied sequentially, the serpentine control device being characterized by having: The prediction calculation unit predicts the meandering amount and wedge amount of the raw material tail end after passing through each of the multiple rolling mill stands, based on the rolling models of each of the multiple rolling mill stands. The rolling model uses the meandering amount of the raw material before rolling, i.e. the initial meandering amount, the meandering amount of the raw material after being rolled by all the multiple rolling mill stands, i.e. the final meandering amount, the reduction and leveling setting value for each of the multiple rolling mill stands, and the rolling process setting to determine the rolling information. The setting update unit updates the reduction and leveling setting values ​​for each of the multiple mill stands based on the snake-like amount and wedge-shaped amount of the raw material tail end predicted by the prediction calculation unit. Multiple reduction leveling control units control the reduction leveling of multiple mill stands respectively based on the reduction leveling setting values ​​updated by the aforementioned setting value update unit. The learning value storage unit stores the learning values ​​used for adjusting the above-mentioned rolling model; as well as The learning value update unit updates the learning values ​​stored in the learning value storage unit. The aforementioned reduction and leveling control units respectively obtain the actual reduction and leveling patterns of the time series or rolling distance series from the aforementioned rolling mill stands. The aforementioned prediction calculation unit predicts the meandering amount and wedge shape of the raw material tail end after passing through each of the multiple mill stands based on the rolling models of each of the multiple mill stands. The rolling model uses the initial meandering amount, the final meandering amount, the actual reduction and leveling patterns obtained by the multiple reduction and leveling control units, the rolling information, and the learning values ​​stored in the learning value storage unit. The learning value update unit updates the learning value so that the error between the snake-like amount predicted by the prediction calculation unit as the specified point of the raw material tail end after passing the downstream mill stand and the final snake-like amount becomes 0.

7. A serpentine control device for controlling the serpentine movement of raw material in a continuous rolling mill, the continuous rolling mill having multiple mill stands through which the raw material passes and for which a reducing force is applied sequentially, the serpentine control device being characterized by having: The prediction calculation unit predicts the meandering amount and wedge amount of the raw material tail end after passing through each of the multiple rolling mill stands, based on the rolling models of each of the multiple rolling mill stands. The rolling model uses the meandering amount of the raw material before rolling, i.e. the initial meandering amount, the meandering amount of the raw material after being rolled by all the multiple rolling mill stands, i.e. the final meandering amount, the reduction and leveling setting value for each of the multiple rolling mill stands, and the rolling process setting to determine the rolling information. The setting update unit updates the reduction and leveling setting values ​​for each of the multiple mill stands based on the snake-like amount and wedge-shaped amount of the raw material tail end predicted by the prediction calculation unit. Multiple reduction leveling control units control the reduction leveling of multiple mill stands respectively based on the reduction leveling setting values ​​updated by the aforementioned setting value update unit. The learning value storage unit stores the learning values ​​used for adjusting the above-mentioned rolling model; as well as The learning value update unit updates the learning values ​​stored in the learning value storage unit. The aforementioned reduction and leveling control units respectively obtain the actual reduction and leveling patterns of the time series or rolling distance series from the aforementioned rolling mill stands. The aforementioned prediction calculation unit predicts the meandering amount and wedge shape of the raw material tail end after passing through each of the multiple mill stands based on the rolling models of each of the multiple mill stands. The rolling model uses the initial meandering amount, the final meandering amount, the actual reduction and leveling patterns obtained by the multiple reduction and leveling control units, the rolling information, and the learning values ​​stored in the learning value storage unit. The learning value update unit evaluates the snake-like amount detected by the inter-stand snake-like amount detection unit, which detects the snake-like amount of raw material between the above-mentioned mill stand and the snake-like amount predicted by the above-mentioned prediction calculation unit, using an evaluation function, until the specified termination condition based on the evaluation result is met, and then updates the learning value.

Images