WIDTH CONTROL DEVICE FOR HOT ROLLING MILL AND CONTROL METHOD THEREOF
TECHNICAL FIELD
The present invention relates to a width control device for a hot rolling mill and a control method of the width control device.
BACKGROUND ART
As a conventional technology of performing a width control for a hot rolling mill, JP2007-50413A (literature 1) discloses a technology of storing rolling-case data for each operation factor, highly-accurately predicting a finish width from data on an operation factor similar to a rolling-target material, and performing a control based on the prediction information.
Moreover, JP2002-224723A (literature 2) discloses a technology of defining a width-prediction-formula vias item as a function of a chemical composition of a rolled material, and of using the vias item for a width control.
Further, JPH09-225513A (literature 3) discloses a technology of learning a relationship between a rougher-rolling actual data after a vertical rolling and a width actual data after a horizontal rolling using information before a rolling, and of repeatedly predicting a width of a rolling material already having undergone rolling after rolling using the learning result, thereby increasing the accuracy of the setup control for a vertical rolling mill for following rolling materials.
For example, JP2003-211212A (literature 4) discloses a width control method for a finishing roll. In a width control using a width gauge (intermediate width gauge) provided between arbitral stands of a hot finishing mill comprising plural stands or at a delivery side of the finishing mill and a width gauge (finish width gauge) provided right ahead of a coiling
device (down coiler), the method predicts the finishing width from an actual width detected by the intermediate width gauge and changes an intermediate width target based on a difference between the predicted and a target finishing width.
DISCLOSURE OF THE INVENTION
PROBLEM TO BE SOLVED BY THE INVENTION
The foregoing technologies, however, have the following problems.
According to literature 1, it is possible to determine the similarity using a factor detectable as an operation factor, but it is consequently difficult to determine a similarity related to an undetectable factor. The variation in the width of a slab prior to edger rolling affects a finish width, but since there is no means for directly measuring the width of the slab prior to edger rolling in a normal hot rolling line, the prediction accuracy of a finish width after edger rolling may decrease due to the absence of such means.
Moreover, according to the control method disclosed in literature 2, it is possible to deeply consider the elastic-plastic characteristic of a steel product to be subjected to edger rolling using information on a chemical composition, but various factors affecting the finish width, in particular, any affects to a slab width prior to edger rolling is not in consideration. Accordingly, due to the variation in a slab width, the finish width accuracy may decrease.
In order to apply the control method disclosed in literature 3, it is necessary to measure widths before and after vertical rolling, a width and a thickness after horizontal rolling. Accordingly, it is necessary to provide individual detectors for measuring those, so that the system becomes expensive. Moreover, even if those detectors are provided beforehand, since a descaling process of eliminating a scale on the surface of a rolling material is normally carried out in a rougher rolling, due to steams generated through this process, the measurement accuracy may decrease or measurement itself may become difficult, resulting in a difficulty of application of such control method. Furthermore, the method needs a repeat
calculation of predicting a width after a horizontal rolling while hypothetically setting the width of the slab after a vertical rolling, so that the setup calculation becomes complex.
The conventional technology disclosed in literature 4 cannot be applied to a steckel mill, which is a reversing hot rolling mill comprised of one stand. That is, unlike a hot finishing mill having tandem sOtands, a steckel mill cannot perform a tension control between finishing mills, resulting in a difficulty of an online controlling of a finish width. Moreover, a large width shrinkage of a slab occurs in a finish rolling. Accordingly, this conventional technology cannot be directly applied to a rolling equipment having a steckel mill.
It is an object of the present invention to overcome at least one of the foregoing problems.
For example, it is an object of the present invention to provide a width control device for a hot rolling mill and a control method of the width control device which can control a width of a rolling-target material highly accurately with a simple calculation in a width controlling in a hot rolling.
For example, the present invention provides a width control technology for a hot rolling mill which enables a width control independent of a finish rolling by estimating a finish width shrinkage amount highly accurately at the time of presetting.
MEANS FOR SOLVING THE PROBLEM
To achieve the objects, a width control device according to the first aspect of the present invention is a width control device for a hot rolling mill, the width control device controlling a slab of a rolling-target material, which is casted by a continuous caster and delivered into a rougher rolling mill, to have a predetermined slab width using an edger, and then controlling the rolling-target material after rolled by a finishing mill to have a width matching a target value, the width control device includes: a width prediction model that includes a relationship between a width of the slab and widths of the rolling-target material during a rolling and after
a rolling; a width estimation unit that estimates a width of the rolling-target material after rolled by the finishing mill using the width prediction model from a primary value of a width of the slab and rolling information on the slab loaded from the hot rolling mill; a width deviation storing unit that stores a deviation between an estimation result of the width estimation unit and a width actual measured value of the rolling-target material after rolling for a rolled rolling-target material; an adaptation unit that corrects a target value of a width of the rolling-target material after rolling based on a content of the width deviation storing unit; and a preset control unit that calculates and outputs a control instruction to the edger through a calculation using the target value of the width corrected by the adaptation unit.
A width control device according to the second aspect of the present invention is a width control device for a hot rolling mill, the width control device controlling a slab of a rolling-target material, which is casted by a continuous caster and delivered into a rougher rolling mill, to have a predetermined slab width using an edger, and then controlling the rolling-target material after rolled by a finishing mill to have a width matching a target value, the width control device includes: a width prediction model that includes a relationship between a width of the slab and widths of the rolling-target material during a rolling and after the rolling; a slab width estimation unit that estimates a width of the slab of the rolling-target material before a rolling by a back-calculation of the width prediction model from a width actual measured value of the rolling-target material after rolling and rolling information on the rolling-target material loaded from the hot rolling mill; a slab width deviation storing unit that "stores a deviation between an estimation result of the slab width estimation unit and a primary value of a width of the slab; a training unit that corrects the primary value of the width of the slab based on a content of the slab width deviation storing unit; and a preset control unit that calculates and outputs a control instruction to the edger to set a width of the rolling-target material after rolling to be a target value through a calculation using the width of the slab corrected by the training unit.
A width control device according to the third aspect of the present invention is a width control device for a hot rolling mill, the width control device controlling a slab of a rolling-target material, which is casted by a continuous caster and delivered into a rougher rolling mill, to have a predetermined slab width using an edger, and then controlling the rolling-target material after rolled by a finishing mill to have a width matching a target value, the width control device includes: a width prediction model that includes a relationship between a width of the slab and widths of the rolling-target material during a rolling and after the rolling; a width estimation unit that estimates a width of the rolling-target material after rolled by the finishing mill using the width prediction model from a primary value of a width of the slab and rolling information on the slab loaded from the hot rolling mill; a width deviation storing unit that stores a deviation between an estimation result of the width estimation unit and a width actual measured value of the rolling-target material after rolling for a rolled rolling-target material; an adaptation unit that corrects a target value of an after-rolling width of the rolling-target material to be rolled based on a content of the width deviation storing unit; a slab width estimation unit that estimates a width of the slab of the rolling-target material before rolling by a back-calculation of the width prediction model from a width actual measured value of the rolling-target material after rolling and rolling information on the rolling-target material loaded from the hot rolling mill; a slab width deviation storing unit that stores a deviation between an estimation result of the slab width estimation unit and a primary value of a width of the slab; a training unit that corrects the primary value of the width of the slab based on a content of the slab width deviation storing unit; and a preset control unit that calculates and outputs a control instruction to the edger to set a width of the rolling-target material after rolling to be a target value through a calculation selectively using either one of an output from the training unit or an output from the adaptation unit.
A width control method according to the fourth aspect of the present invention is a width
control method of a width control device for a hot rolling mill, the width control device controlling a slab of a rolling-target material, which is casted by a continuous caster and delivered into a rougher rolling mill, to have a predetermined slab width using an edger, and then controlling the rolling-target material after rolled by a finishing mill to have a width matching a target value, the method includes: storing a relationship between a width of the slab and widths of the rolling-target material during a rolling and after the rolling; estimating a width of the slab of the rolling-target material before rolling from a width actual measured value of the rolling-target material after rolling and rolling information on the rolling-target material loaded from the hot rolling mill by a back-calculation of the relationship between the width of the slab and the widths of the rolling-target material during a rolling and after the rolling; storing a deviation between an estimation result and a primary value of the width of the slab in a first table; estimating a width of the slab, the relationship between widths of the rolling-target material during a rolling and after the rolling, and a width of the rolling target material rolled by the finishing mill, using the primary value of the width of the slab and rolling information on the slab loaded from the hot rolling mill; calculating a deviation between the width estimation result and the width actual measured value for a rolled rolling-target material, and storing the deviation in a second table; acquiring a number of slabs rolled after the continuous caster starts casting or after the continuous caster changes a width of a mold, and correcting a primary value of a width of the slab to be rolled based on a content of the first table when the number of rolled slabs is not greater than a predetermined number; correcting a width target value of the rolling-target material to be rolled based on a content of the second table when the number of rolled slabs is greater than the predetermined number; and calculating and outputting a control instruction to the edger to control a width of the rolling-target material after rolling to be a predetermined value in accordance with a correction result.
Further, to overcome the foregoing problems, a width control device according to the
fifth aspect of the present invention is a width control device for a hot rolling mill including an edger mill that rolls a heated steel product in a width direction; a rougher rolling mill which is provided adjacent to the edger mill and rolls the steel product rolled by the edger mill in a thickness direction; and a finishing mill that further rolls the steel product rolled by the edger mill and the rougher mill in a thickness direction to produce a hot coil, the width control device controlling a width of the hot coil to be a target width given by a control instruction, the width control device includes: a fmish-width-shrinkage estimation amount calculation unit that calculates and estimates a width shrinkage amount at the finishing mill; and a first width target value correction unit that corrects the target width using an output from the finish-width-shrinkage estimation amount calculation unit, wherein the control instruction calculated based on a target width corrected by the first width target value correction unit is output to the edger mill.
EFFECT OF THE INVENTION
According to the present invention, as one of the effects, the width of a rolling-target material can be controlled highly accurately with a simple calculation in a width controlling in a hot rolling.
As another effect, a finish width shrinkage amount can be accurately estimated at the time of presetting, resulting in a width control independent from a finish rolling.
BRIEF DESCRIPTION OF DRAWINGS
FIG. 1 is a conceptual diagram showing a structure of a control system having a width control device according to a first embodiment of the present invention:
FIG. 2 is a diagram showing an example structure of a PDI table according to the first embodiment;
FIG. 3 is a flowchart showing a process executed by width estimation means according
to the first embodiment;
FIG. 4 is a diagram showing a structure of width deviation storing means according to the first embodiment;
FIG. 5 is a flowchart showing a process executed by slab width back-calculation means according to the first embodiment;
FIG. 6 is a diagram showing a structure of slab width deviation storing means according to the first embodiment;
FIG 7 is a flowchart showing a process executed by change-over means according to the first embodiment;
FIG. 8 is a flowchart showing a process executed by preset control means according to the first embodiment;
FIG. 9 is an exemplary diagram showing contents of processes from steps S84 to S86 by preset control means according to the first embodiment;
FIG. 10 is a conceptual diagram showing a structure of a control system that has adaptive gain calculation means in a width control device according to a second embodiment of the present invention;
FIG. 11 is a flowchart showing a process executed by adaptive gain calculation means according to the second embodiment;
FIG. 12 is a diagram showing an adaptive error sequence table according to the second embodiment;
FIG. 13 is a diagram for explaining a thickness control device according to a third embodiment of the present invention;
FIG. 14 is a diagram for explaining an example of a finish-width-shrinkage adaptive amount table;
FIG. 15 is a diagram showing an example of a fmish-width-shrinkage adaptive gain table;
FIG. 16 is a diagram showing an example of a relationship between a number of rolling coils and a model error;
FIG. 17 is a diagram showing a relationship between a slab width before and after width-shrinkage correction and a model error;
FIG. 18 is a diagram for explaining a process executed by first width target value correction means up to calculation of a finish width correction amount;
FIG 19 is a diagram showing an example of a rougher width adaptive amount storing table;
FIG. 20 is a diagram for explaining how to determine a target width at each path of a rougher edger mill;
FIG. 21 is a diagram for explaining how to correct a target width at each path of a rougher edger mill; and
FIG. 22 is a diagram for explaining a process until an under-edger mill pressure primary value at each rougher path is determined by correcting a rougher target width 1.
DESCRIPTION OF REFERENCE NUMERALS
100 Width control device
101 Preset control means
102 Width estimation means

104 Width deviation storing means (second table)
105 Adaptation means
106 Slab width back-calculation means

108 Slab width deviation storing means (first table)
109 Training means
110 Change-over means
115 Width prediction model
150 Hot rolling mill
151 Edger

154 Rougher rolling mill
155 Finishing mill
156 Slab (rolling-target material)
15 7 Steel plate (rolling-target material)
161 Thin slab continuous caster
180 Thickness control means
190 PDI table
10001 Adaptive gain calculation means
P Adaptive gain
AWact Deviation
1100 Width control device
1110 Production instruction unit
1120 First width target value correction means
1121 Finish-width-shrinkage estimation amount calculation means
1122 Finish-width-shrinkage adaptive amount calculation means
1123 Finish-width-shrinkage-adaptive-amount effectiveness quantification means

1131 Finish-width-shrinkage adaptive amount storing table
1132 Finish-width-shrinkage adaptive gain storing table

1140 Second width target value correction means
1141 Rougher width adaptive amount calculation means
1142 Rougher-width-adaptive-amount effectiveness quantification means

1150 Width expansion model
1151 Rougher width adaptive amount storing table
1152 Rougher width adaptive gain storing table
1160 Edger mill setup means
1170 Rougher width actual data gathering means
1180 Finish width actual data gathering means
1200 Control target
1210 S lab extraction unit
1215 Furnace
1220 Edger rolling unit
1205 Slab
1225 Edger mill
1235 Rougher rolling mill
1240 Finish rolling unit
1245 Finishing mill (steckel mill)
1250 Coiling unit
1255 Downcoiler
1261 Rougher delivery-side width gauge
1262 Finish delivery-side width gauge
BEST MODE FOR CARRYING OUT THE INVENTION
«Overview of Width Control Device for Hot Rolling Mill and Control Method for Same»
A width control device of the present invention for a hot rolling mill and a control method of the width control device are appropriate for conforming a width of a rolling-target material after rolling to a target value with a simple calculation focusing on variation in a slab width prior to a rolling of the rolling-target material like a steel plate, which is successively rolled in a hot rolling.
In particular, the present invention is appropriate for a mini hot having a large correlation of variation in a width of a rolling-target material after rolling, and the present
invention is particularly effective for a so-called mini mill in which a slab casted by a continuous caster is directly conveyed to a mill.
Hereinafter, an explanation will be given of the embodiments of the present invention with reference to accompanying drawings. Note that the components will be individually named "means" in the following explanation, but can be also called "unit" individually. «First Embodiment»
FIG 1 is a conceptual diagram showing a structure of a control system S having a width control device 100 according to the first embodiment of the present invention.
The control system S having the width control device 100 of the first embodiment has a hot rolling mill 150 which is a hot ro'lling equipment and the width control device 100 which receives various signals from the hot rolling mill 150, outputs a control signal to an edger 151 of the hot-rolling mill 150 subjected to a control, and controls a roll gap between rolls 1 Sir of the edger 151.
First, an explanation will be given of a structure of the hot rolling mill 150 subjected to a control by the width control device 100.
The hot rolling mill 150 is an equipment so-called a mini mill which carries out a successive casting/rolling process.
The hot rolling mill 150 has a thin slab continuous caster 161 for cooling down a molten steel of a rolling-target material like a high-temperature iron supplied from a tundish 164 and for forming a slab having a uniform horizontal cross-section, a torch cutter 162 which cuts the rolling-target material conveyed from the thin slab continuous caster 161 to a predetermined length, a tunnel furnace 163 which heats the rolling-target material cut by the torch cutter 162, a rougher rolling mill 154 which forms a slab 156 conveyed from the tunnel furnace 163 to have a predetermined width by vertical rolls of the edger 151 and performs horizontal rolling
on the slab 156 using horizontal rolling mills Rl, R2, and a finishing mill 155 which performs finish rolling using tandem rolling mills Fl to F5 with five stands.
Note that in the first embodiment, an explanation will be given of a case where a rolling-target material is a steel plate 157.
The thin slab continuous caster 161 shown in FIG 1 has the tundish 164 which retains a high-temperature molten steel of the rolling-target material like iron, and a casting mold 165 which cools down the high-temperature molten steel supplied from the tundish 164 to form the slab 156 and which has a size corresponding to the thickness of the slab 156 and the width thereof.
The rougher rolling mill 154 has the edger 151 which forms the width of the rolling-target material to a predetermined width, and two-stand type horizontal rolling mills Rl, R2 which roll the rolling-target material conveyed from the edger 151 in a thickness direction to form the rolling-target material having a predetermined thickness.
The edger 151 indicated by a symbol El in FIG. 1 has two vertical rolls 151r each having a rotational axis in a vertical direction and provided at both sides (fron^ack direction of the plane of FIG 1) of the conveyed slab 151.
By adjusting a distance (roll gap) between those rolls 151r (a distance between the pair of rolls 151r provided in the front/back direction of the plane of FIG 1), the width of the slab 151 pressed by the two vertical rolls 151r can be controlled.
The finishing mill 155 is a five-stand type and comprises tandem rolling mills Fl to F5. The finishing mill 155 successively rolls a steel plate 157, which is the rolling-target material conveyed from the rougher rolling mill 154, using rolls 159 of the individual tandem rolling mills Fl to F5 to finish the steel plate 157 to have a desired thickness.

The rolling process of the hot rolling mill 150 is carried out as follows.
As shown in FIG 1, the high-temperature molten steel of the rolling-target material retained in the tundish 164 is supplied to the mold 165 by its own weight and cooled down in the mold 165 from the external portion of the molten steel to be solidified, so that a steel plate having a thickness of 70 to 85 mm or so is formed. The steel plate is cut to pieces each having a length of 10 to 30 m or so by the torch cutter 162 to form a slab 156, which is a rolling-target material prior to rolling.
The slab 156 is kept warm and heated by the tunnel furnace 163, and then rolled in the width direction by the pair of rolls 1 Sir of the edger 151 which rotate around the vertical axis (vertical direction of the plane of FIG 1) in the rougher rolling mill 154 and rolled in the thickness direction by the horizontal rolling mills Rl, R2, and further rolled in the thickness direction by the respective rolls 159 of the tandem rolling mills Fl to F5 in the finishing rolling mill 155, so that a steel plate having a thickness of 2 to 10 mm or so is formed.
A width gauge 158 is provided at the delivery side of the finishing mill 155 and a width of the rolling-target material like the steel plate 157 is measured.
It is not illustrated in FIG 1, but the steel plate 157 is then cooled by a run-out table (not shown), and is coiled up by a down coiler (not shown) as a hot coil.
Next, an explanation will be given of the structure of the width control device 100 which controls the roll gap between the rolls 1 Sir of the edger 151 in the hot rolling mill 150.
The width control device 100 shown in FIG 1 is realized by using, for example, a PLC (Programmable Logic Controller).
Prior to a rolling by the hot rolling mill 150, the width control device 100 reads out a target width of the steel plate 157 of the rolling-target material from a PDI (Primary Data Input) table 190 (see FIG. 2), calculates a roll gap primary value between the rolls 15 Ir of the
edger 151 to achieve the target width, and outputs a control instruction to the edger 151 via preset control means 101.
A distance between the rolls 15 Ir of the edger 151 is controlled to be a roll gap in accordance with the roll gap primary value transmitted from the width control device 100, and then a process is carried out.
FIG. 2 shows an example structure of the PDI table 190.
As shown in FIG. 2, the PDI table 190 stores data of each item, such as a steel kind 190a corresponding to a steel plate number 190o, or a slab width 190b.
FIG. 2 indicates that the steel kind 190a is SS400, the slab width 190b is 1220 mm, and the target width 190c is 1200 mm, correspondingly to the steel plate number HX0012352.
An order 190i after a charging is started or after a width is changed shown in FIG .2 indicates what number the HXOO12352 is in all steel plates 157 produced after the thin slab continuous caster 151 starts charging (casting) or after a casting width is changed.
Note that "after a width is changed" means after a width setting value of the thin slab continuous caster 161 or the like is changed to set the steel plate 157 after the rolling to have a desired width.
The width control device 100 shown in FIG 1 has a width prediction model 115 for predicting the width of the steel plate 157 after rolling from the roll gap of the edger 151 and a rolling state such as the transition of a thickness of the rolling-target material at the rougher rolling mill 154 and the finishing mill 155. The content of the width prediction model 115 can be expressed by equations (1) to (6) to be discussed later.
The width control device 100 has width estimation means 102 which acquires, from the hot rolling mill 150 subjected to the control, the rolling actual data of the rougher rolling mill 154 and the finishing mill 155 and the actual data like the thickness of the slab 156 prior to a rolling by the rougher rolling mill 154 and estimates a width of the steel plate 157 after rolling using the width prediction model 115, width deviation calculation means 103 which calculates
a deviation between an output of the width estimation means 102 and an actual measured width value corresponding to the steel plate 157 (see FIG. 1) actually measured by the width gauge 158 after rolling, width deviation storing means 104 which stores a value of a width deviation calculated by the width deviation calculation means 103 with respect to a steel plate 157 rolled recently, and adaptation means 105 which calculates an adaptation control amount for correcting the width target value so that the rolling-target material (steel plate 157) after rolling can have a desired width using a value of the width deviation stored in the width deviation storing means 104.
The width control device 100 shown in FIG 1 further has slab width back-calculation means 106 which acquires, from the hot rolling mill 150 to be controlled, rolling actual data of the rougher rolling mill 154 and the finishing mill 155 and a width of the rolling target material (steel plate 157) after rolling measured by the width gauge 158, and acquires a width of the slab 156 prior to rolling at the rougher rolling mill 154 by a back calculation, slab width deviation calculation means 107 which calculates a slab width deviation from the back calculation result obtained by calculating back the width prediction model 115 by the slab width back-calculation means 106 and the slab width 190b instructed from the PDI table 190, slab width deviation storing means 108 which stores the value of the slab width deviation calculated by the slab width deviation calculation means 107 for a long period, and training means 109 which calculates a training control amount for correcting the slab width used for a preset calculation by the preset control means 101 using the value of the slab width deviation stored in the slab width deviation storing means 108.
Further, the width control device 100 has change-over means 110 which determines, from the PDI table 190, what number the slab 156 is in all slabs conveyed after a charging (casting) is started or after the thin slab continuous caster 161 changes the width setting value, and outputs information for changing over an output from the adaptation means 105 and the output from the training means 109 to the preset control means 101.
Utilizing the foregoing components, the width control device 100 shown in FIG 1 calculates a roll gap primary value between the rolls 1 Sir of the edger 151, and outputs a control instruction to the edger 151 from the preset control means 101.
Next, the individual components will be explained in detail.
An explanation will be given of the width estimation means 102 which estimates the width of the steel plate 157 that is the rolling-target material having undergone rolling with reference to FIG 3. FIG 3 is a flowchart showing the process executed by the width estimation means 102.
In a step S31 shown in FIG 3, a width shrinkage amount ΔW1 after a rolling at the El edger 151 (see FIG 1) relative to a width BO of the slab 156 prior to a rolling at the rougher rolling mill 154 is calculated. The width shrinkage amount ΔW1 can be acquired from the following equation:
ΔW1=BO-Se (1)
where Se is the roll gap of the edger 151.
In steps S32, S33 in FIG 3, a width expansion amount at the horizontal rolling mill Rl in the rougher rolling mill 154 is calculated. That is, in the step S32 in FIG 3, a width expansion amount ΔW2 (rectangle amount ΔW2) of the rolling-target material due to a horizontal rolling is calculated, and in the step S33 in FIG 3, a width expansion amount (dog-bone width expansion amount, i.e., dog-bone recovery amount ΔW3) in the width direction due to the recovery of an edge portion of the rolling-target material like the steel plate 157, which is pressed toward the middle portion by a rolling by the edger 151 and built up in the thickness direction, is calculated.
A width expansion amount of the steel plate 157 of the rolling-target material in the horizontal rolling mill Rl can be expressed by a sum of the foregoing values.
The rectangle amount ΔW2 in the step S32 in FIG 3 can be acquired from equation (2),
and the dog-bone recovery amount ΔW3 in the step S33 in FIG. 3 can be acquired from equation (3).
(Equation Removed)
where:
Bo is a slab width;
B is a width at the horizontal rolling mill Rl entry side (= Se);
RE is a roll diameter of the edger 151;
Se is a roll gap of the edger 151;
R is a work roll radius of the horizontal rolling mill Rl;
H is a thickness at the horizontal rolling mill Rl entry side; and
h is a thickness at the horizontal rolling mill Rl delivery side.
In a step S34 in FIG. 3, a width expansion amount AW4 of the steel plate 157 which is the rolling-target material at the horizontal rolling mill R2 shown in FIG 1 is acquired from equation (4):
(Equation Removed)
where:
B is a width at the horizontal rolling mill R2 entry side;
R is a work roll radius of the horizontal rolling mill R2;
H is a thickness at the horizontal rolling mill R2 entry side; and
h is a thickness at the horizontal rolling mill R2 delivery side.
In a step S35 in FIG. 3, a width shrinkage amount ΔW5 between both-side edges of the steel plate 157 of the rolling-target material at the tandem rolling mills Fl to F5 of the finishing mill 155 is acquired from equation (5).
(Equation Removed)
bp: target width
tb: backward tension which is tension of the rolling-target material entering into the tandem rolling mill F1
t: tension elapsed period (period when tension is affecting to the steel plate 157)
A, B, C, m, n: constants
Thereafter, in a step S36, a width Wc of the rolling-target material which is a steel plate 157 or the like after final finishing is estimated from equation (6). Note that BO is the slab width 190b acquired from the PDI table 190.
(Equation Removed)
The above-explained procedures are the process executed by the width estimation means 102, and equations (1) to (6) are called the width prediction model 115 (see FIG 1).
Note that equations (2) to (5) are experimental formulas well-known in the art.
The width deviation calculation means 103 shown in FIG 1 calculates a deviation ΔWact between Wc which is a predicted value of the width of the steel plate 157 of the rolling-target material after rolling and an actual width Wa of the steel plate 157 (rolling-target material) corresponding to the predicted value Wc and measured by the width gauge 158.

FIG. 4 shows a structure of the width deviation storing means 104.
The width deviation storing means 104 shown in FIG 4 stores a width deviation ΔWact and ΔWadap (adaptive value) which is a corrected amount of the width target value in a width setup calculation, in association with information for specifying the steel plate 157 of the rolling-target material rolled recently.
For example, in FIG 4, regarding a coil (steel plate 157) which is the rolling-target material rolled at last, it is indicated that a setup calculation is carried out while performing -11 mm (adaptive value ΔWadap) correction on the width target value, and as a result, the width deviation ΔWact is + 3 mm. Moreover, regarding a coil (steel plate 157) which is the rolling-target material rolled at last but one, it is indicated that a setup calculation is carried out while performing -10 mm (adaptive value AWadap) correction on the width target value, and as a result, the width deviation ΔWact is -1 mm.
Next, an explanation will be given of the slab width back-calculation means 106 that calculates back the width of the slab 156 prior to a rolling from rolling actual data of the rougher rolling mill 154 and the finishing mill 155 shown in FIG 1 and a width value of the steel plate 157 which is the rolling-target material having undergone rolling measured by the width gauge 158.
FIG. 5 is a flowchart showing the process executed by the slab width back-calculation means 106.
In a step S51 in FIG 5, a width Wf,n of the rolling-target material at the entry side of the tandem rolling mill Fl of the finishing mill 155 (see FIG 1) is acquired by a back calculation from the width Wa of the steel plate 157 which is the rolling target material measured by the width gauge 158 (see FIG 1).
The width Wfm of the steel plate 157 of the rolling-target material at the entry side of the
tandem rolling mill Fl can be acquired by adding AW5 from equation (5) to the width Wa of the steel plate 157 of the rolling-target material having undergone rolling.
Next, in a step S52 in FIG 5, a width Wr2 of the steel plate 157 which is the rolling-target material at the entry side of the horizontal rolling mill R2 (see FIG 1) is acquired by a back calculation. The width Wr2 at the entry side of the horizontal rolling mill R2 can be acquired by subtracting AW4 (see equation (4)) from the width Wfin at the entry side of the rolling mill Fl.
In a step S53 in FIG. 5, the slab width Bo is acquired by a back calculation.
First, as shown in following equation (7), the dog-bone width expansion amount (dog-bone recovery amount) ΔW3 can be acquired by subtracting the gap Se of the rolls 151r of the edger 151 and the horizontal rolling width expansion amount (rectangle amount) ΔW2 indicated in equation (2) from the width Wr2 of the rolling-target material at the entry side of the horizontal rolling mill R2 (see FIG. 1).
(Equation Removed)
ΔW3 is substituted into equation (3), and the equation is solved for B0, thereby acquiring the slab width BO by back calculation.
Equation (3) cannot be algebraically solved for BO, but by successively substituting values which are possible values of the slab width into the slab width Bo at the right side, and by acquiring B0 having a calculation result which is the most closest to the substituted ΔW3, the equation can be easily solved as numerical analysis.
It is needless to say that Bo acquired at this stage does not directly correspond to the difference between an actual slab width and the slab width 190b instructed from the PDI table 190, but is a value which summarizes variation containing prediction errors of equations (2) to (5) into the difference between the slab width B0 and the primary value (190b) of the PDI table 190.
FIG. 6 is a diagram showing a structure of the slab width deviation storing means 108.
As shown in FIG 6, the slab width deviation storing means 108 stores slab width deviations (deviations each between the acquired slab width BO and the slab width primary value 190b of the PDI table 190) 108c corresponding to the steel plates 157 which are the rolling-target material having undergone rolling, added with newer numbers 108b from the steel plate 157 which is the recent rolling-target material, for each target width 108a of the steel plate 157 of the rolling-target material. The slab width deviation storing means 108 stores such data for a long period.
For example, as shown in FIG 6, it is indicated that a slab width deviation is -6 mm at the last rolling actual data, and -3 mm at last but one actual data for the steel plate 157 having a target width of 900 mm.
Next, an explanation will be given of the change-over means 110 which outputs information for changing over an output to the preset control means 101 shown in FIG. 1 between the output of the adaptation means 105 and the output of the training means 109.
The change-over means 110 executes a process of selecting either one of a result of the adaptation means 105 (see FIG. 1) or a result of the training means 109 (see FIG 1) that is used when the preset control means 101 of the width control device 100 calculates a gap between the rolls 151r of the edger 151 which rolls the slab 156 in the width direction.
FIG. 7 is a flowchart showing the process executed by the change-over means 110.
First, in a step S71 in FIG 7, the number of steel plates 157 of the rolling-target materials after a charging (casting) is started or after the width of the mold 165 (see FIG. 1) is changed is acquired from the PDI table 190. Note that the width of the mold 165 corresponds to the casting width of the thin slab continuous caster 161, so that the width of the slab 156 can be determined from the casting width.
The reason why the condition after a charging (casting) is started or after the width of
the mold 165 (see FIG 1) is changed is selected is that the width setting of the rolling-target material may be changed during a charging, i.e., during a casting.
In a step S72 in FIG. 7, a correlation between the acquired value and a width of the slab 156 to be rolled next is determined. That is, it is determined whether or not the width of the slab 156 to be rolled next has a large correlation with a width of the slab 156 rolled recently (e.g., last time or several times before).
The largeness of the correlation corresponds to the similarity of widths of the slabs 156, i.e., the temporal approximation, and is determined based on whether or not the number of the steel plates 157 of the rolling-target materials acquired at the step S71 in FIG 7 is smaller than a predetermined number.
More specifically, when the number of steel plates 157 of the rolling-target material counted from after charging is started or after the width of the mold 165 is changed is small, it is determined that the correlation is small, and when the number of the steel plates 157 is not small, it is determined that the correlation is large.
For example, in a case where the steel plate 157 is the first one after a casting is started, there is no correlation in a width deviation with the steel plate of the rolling-target material rolled last time, so that the prediction errors and the like contained in equations (2) to (5) must be compensated in accordance with a long-term tendency. That is, when the number of the steel plate 157 of the rolling-target material is the first one counted after a charging is started or after the width of the mold 165 is changed, previous rolling is one at a charging carried out at a previous time or one before the width of the mold 165 is changed, so that a correlation between data cannot be expected. Accordingly, the prediction errors must be compensated in accordance with actual result regarding how much errors are contained based on data stored beforehand for a long period.
Accordingly, the training means 109 shown in FIG 1 is activated in a step S73 in FIG 7. The training means 109 refers to the slab width deviations 108c for the corresponding
respective target widths 108a stored in the slab width deviation storing means 108 (see FIG 6), acquires a weighted average thereof in accordance with a process following equation (8), and calculates a training result Wt.
(Equation Removed)
where coi is a weighted value, and ASi is a slab width deviation.
coi is a weighted value for each slab width deviation ASi, can all be 1, or a weighted value of the steel plate 157 of the rolling-target material having undergone rolling recently may be set large to calculate a training result Wt while putting an weight on the recent slab width deviation ΔSi. Conversely, in a case of a rolling-target material determined as non-first ones, i.e., second or later ones, since the slab 156 is casted with the same width of the mold 165 as that of the steel plate 157 of the rolling-target material rolled at a previous time, it is expected that there is a large correlation in a width deviation with the steel plate 157 of the rolling-target material to be rolled next.
Therefore, the adaptation means 105 (see FIG 1) is activated in a step S74 in FIG 7.
The adaptation means 105 refers to the width deviation ΔWact calculated with the steel plate 157 of the rolling-target material rolled recently and stored in the width deviation storing means 104 (see FIG 4), and calculates an adaptive result Wa.
The adaptation means 105 may refer to the width deviation storing means 104 only for the width deviation of the steel plate 157 of the rolling-target material rolled last time, and may refer to the width deviation ΔWact of the recent several steel plates 157 of the rolling-target material for the purpose of stabilizing an adaptive result. A weighted average is acquired through a process in accordance with equation (9) for the former case, and equation (10) for the latter case, and an adaptive result Wa is acquired.
In the setup calculation, since the width target value is corrected beforehand in a width setup calculation, the effect thereof can be cancelled by subtracting the adaptive value ΔWadap.
(Equation Removed)
where ΔCpre is a width deviation of the steel plate 157 rolled last time, and (Wadap)pre is an adaptive value ΔWadap applied to the setup calculation of the steel plate 157 rolled at last.
(Equation Removed)
where coi is a weighted value, ΔCi is a width deviation and (WadaP)i is an adaptive value.
An acquired result is output to the preset control means 101 in a step S75 in FIG 7.
Note that in the embodiment, it is determined whether the training means 109 is used or the adaptation means 105 is used in the step S72 in FIG. 7 based on the number of rolled steel plates 157 after a charging (casting) is started or after the width of the mold 165 is changed, the determination may be based on either one of the number of the rolled steel plates 157 after a charging (casting) is started or the number of the rolled steel plates 157 after the width of the mold 165 is changed.
Normally, the rolled number used in the determination in the step S72 is set to 1, and the output of the training means 109 is used for the first slab after charging is started or after the width of the mold 165 is changed, and the output of the adaptation means 105 is used for other cases.
Next, an explanation will be given of the process executed by the preset control means 101 which outputs a roll gap primary value between the rolls 1 Sir to the edger 151 shown in FIG. 1.
FIG. 8 is a flowchart executed by the preset control means 101.
In a step S81 in FIG 8, the preset control means 101 receives the output from the change-over means 110 (see FIG 1), and determines whether the setup of the steel plate 157 of the rolling-target material to be rolled next is a training preset for a compensation by the training means 109 or an adaptive preset for a compensation by the adaptation means 105.
When determining in the step S81 in FIG 8 that it is the training preset, in a step S82 in FIG. 8, the preset control means 101 acquires Ws_cont by adding a value acquired by
multiplying the training result Wt (refer to equation (8)) by a predetermined gain a to the slab width (190b) Ws stored in the PDI table 190 in accordance with equation (11), and executes the following setup calculation.
The gain a is normally 0 to 1, and the better the reliability of the training result Wt is, the larger the gain can be set. When the reliability of the training result Wt is extremely poor, the gain is set to 0.
(Equation Removed)
Conversely, when determining in the step S81 in FIG 8 that it is the adaptive preset, in a step S83, the preset control means 101 acquires Wc cont by adding a value acquired by multiplying the adaptive result Wa (refer to equation (9)) by a predetermined adaptive gain p to a target width (190c) Wcjarget stored in the PDI table 190 (see FIG 2) in accordance with equation (12), and executes the following setup calculation.
Like the gain a, the adaptive gain p is generally 0 to 1, and the larger the correlation of (ΔCi - (Wadap)O (refer to equation (9)) between coils (steel plates 157 of rolling-target material) of an error detected from the recent coil (steel plate 157 of rolling-target material) is, the larger the adaptive gain P can be set.
(Equation Removed)
The preset control means 101 acquires a rolling schedule (setting result of from what mm to what mm the steel plate 157 of the rolling-target material is rolled by the horizontal rolling mill Rl, R2 and the tandem rolling mill Fl to F5, respectively), and a value of the tension of the steel plate 157 of the rolling-target material measured by a tension gauge (not shown) between the stands of the tandem rolling mill Fl to F5 from thickness control means 180 in a step S84 in FIG. 4.
The preset control means 101 calculates an entry side finish width estimation value (a width estimation value at the entry side of the finishing mill 155 shown in FIG. 1) from the target width Wc cont using the width prediction model 115 (see FIG 1) represented by
equations (1) to (6) in a step S85 in FIG 8.
The entry side finish width estimation value can be acquired by adding ΔW5 from equation (5) to the target width Wc_COnt-
The preset control means 101 calculates a delivery-side width estimation value Weib of the edger 151 (see FIG 1) in a step S86 in FIG 8.
The delivery-side width estimation value Weib of the edger 151 can be acquired by subtracting the rectangle amount ΔW2 acquired from equation (2) at the horizontal rolling mill Rl, the dog-bone recovery amount ΔW3 acquired from equation (3) at the horizontal rolling mill Rl, and the width expansion amount AW4 acquired from equation (4) at the horizontal rolling mill R2, from the entry side finish width estimation value acquired in the step S85.
FIG. 9 exemplary shows the foregoing processes. Note that FIG 9 is an exemplary diagram showing the contents of the processes from S84 to S86 executed by the preset control means 101.
The entry side finish width estimation value of the rolling-target material input into the tandem rolling mill Fl is calculated from the target width Wc cont, and an El (edger 151) delivery-side width estimation value is further calculated.
Next, in a step S87 in FIG 8, a delivery-side width estimation value Weif of the edger 151 which can be acquired by applying an adjustable maximum width reduction to the width of the slab 156 is calculated.
In a step S88 in FIG 8, it is determined whether or not width buildup is possible based on the magnitude correlation between two El (edger 151) delivery-side width estimation values Weib and Weif.
For example, when a steel plate 157 having a width of 910 mm after a finish rolling is desired, if the target width can be set to 900 mm at most, width buildup is impossible.
As shown in FIG 9, when Weib > We|f, width buildup is possible and when Weib < Weif, a
target width cannot be obtained under the maximum edger rolling, so that width buildup is impossible.
When it is determined in the step S88 in FIG. 8 that width buildup is possible, the roll gap of the edger 151 (distance between rolls 151r) and a load applied to the edger 151 are calculated in a step S89 in FIG 8, and are output to the edger 151.
Conversely, when it is determined in the step S88 in FIG 8 that width buildup is impossible, the thickness control means 180 (see FIG 1) is instructed to re-calculate a rolling schedule (setting for from what mm to what mm the steel plate 157 of the rolling-target material is rolled at individual horizontal rolling mills Rl and R2, and tandem rolling mills Fl to F5 shown in FIG 1) in a step S810 in FIG 8.
In the first embodiment, the explanation has been given of the case where there is one thin slab continuous caste 161, but there may be a case where a plurality of casting equipments (strands) (plurality of thin-slab continuous casters 161 are provided in the fron^ack direction of the plane of FIG 1) are provided, and slabs 156 are casted concurrently. In this case, it is necessary that each strand has width deviation storing means 104 (see FIG. 1), and the adaptation means 105 determines from which strand a slab 156 is formed to perform a calculation using data stored in the width deviation storing means 104 corresponding to that strand. The present invention can also be applied to such a case with the foregoing structure.
Note that the explanation has been given of the case where the width control device 100 has both adaptation means 105 and training means 109 and changes over those in the first embodiment, but the width control device 100 may employ a structure of always using either one of those. «Second Embodiment»
Next, an explanation will be given of a control system 2S according to the second embodiment of the present invention.
FIG. 10 is a conceptual diagram showing the control system 2S where an adaptive gain
calculation means 10001 is included in the width control device 100 according to the second embodiment.
In the second embodiment, the width control device 100 of the first embodiment has the adaptive gain calculation means 10001 which calculates an adaptive gain p (refer to equation (12)).
The other structures are the same as those of the first embodiment, the same reference numerals will be denoted to the components, and the detailed explanations thereof will be omitted.
The adaptive gain calculation means 10001 in the width control device 100 of the second embodiment acquires a width deviation ΔWact of the steel plate 157 rolled recently, i.e., last time or several times before, and an adaptive value AWadap from the width deviation storing means 104 (see FIG 4) and automatically optimizes an adaptive gain (3 of equation (12) in accordance with the largeness of a correlation. How to calculate the adaptive gain P will be discussed later.
Next, an explanation will be given of the process executed by the adaptive gain calculation means 10001 of the width control device 100.
FIG. 11 is a flowchart showing the process executed by the adaptive gain calculation means 10001 of the second embodiment.
The adaptive gain calculation means 10001 acquires the content of the width deviation storing means 104 (see FIG 4), and creates an adaptive error sequence table shown in FIG. 12 in a step Sill in FIG 11. Note that FIG 12 is a diagram showing the adaptive error sequence table of the second embodiment.
A first adaptive error sequence shown in FIG 12 is a value acquired by subtracting an adaptive value ΔWadap from a width deviation ΔWact of the width deviation storing means 104 shown in Fig. 4, and a second adaptive error sequence shown in FIG 12 is a value acquired by
shifting the value of the first adaptive error sequence by one steel plate 157.
For example, in the width deviation storing means 104 shown in FIG 4, since the width deviation is +3 mm and the adaptive value is -11 mm for the coil rolled at last (steel plate 157 of rolling-target material rolled at last), the value of the corresponding first error adaptive sequence becomes 14 mm. Likewise, the first adaptive error sequences can be calculated as 9 mm, 17 mm. In contrast, the second adaptive error sequences are calculated as 9 mm, 17 mm, and the like by shifting the respective first adaptive error sequences.
The correlation between the errors of width estimation related to the steel plates 157 of the rolling-target material rolled contiguously, i.e., successively can be expressed by a cross correlation between the first adaptive error sequence and the second adaptive error sequence.
In general, when the correlation is large, it is possible to expect the same estimation error for a next rolling as that of the previous rolling, so that the adaptive gain p can be set large and the width estimation error can be positively compensated. Conversely, when the correlation is small, the adaptive gain P cannot be set large. The adaptive gain calculation means 10001 sets the adaptive gain p while considering this characteristic.
The adaptive gain calculation means 10001 calculates a correlation of the width deviation of a coil (steel plate 157 of rolling-target material) rolled recently in a step SI 12 in FIG 11.
That is, from equation (13), a cross correlation Conv between the first adaptive error sequence and the second adaptive error sequence shown in FIG 12 is calculated.
(Equation Removed)
where:
Cov(x, y) is a covariance of the fist adaptive error sequence and the second adaptive error sequence;
ax is a variance of the first adaptive error sequence; and
ay is a variance of the second adaptive error sequence.
When the cross correlation Conv is large, the adaptive gain p can be set large.
The adaptive gain calculation means 10001 determines the adaptive gain p in a step Sll3 in FIG. 11 based on the acquired cross correlation.
The adaptive gain p can be calculated from, for example, equation (14).
(Equation Removed)
where K is a constant.
Note that the adaptive gain p can be 0 to 1 as explained above, and is set large as there is a large correlation, and K functions as an adjustment.
Through the foregoing calculation, it is possible to automatically set the adaptive gain p to an appropriate value based on the correlation between the adaptive error sequences of the width estimation model 115 in a recent rolling.
As shown in FIG 1 and FIG 10, the width of the steel plate 157 of the rolling-target material are controlled by the edger 151 provided at the rougher rolling mill 154. In general, the width of the slab 156 casted by a continuous caster (thin slab continuous caster 151) is determined by the width of the mold 165 of the caster, and there is little variation in the widths of the slabs 156 successively casted.
Accordingly, in a case where a casted slab 156 is directly conveyed to a mill like a mini mill, there is a correlation between the widths of the slabs 156 successively rolled. In contrast, such a correlation is lost at a timing when the width of the mold 165 is changed.
Focusing on this characteristic, and in order to realize a highly-accurate width control for a rolling-target material, the width control device 100 has the slab width deviation calculation means 107 which acquires, by back calculation, the width of the slab 157 that is the rolling-target material prior to rolling from the width of the rolling-target material like the steel plate 157 having undergone rolling detected by the width gauge 158 and rolling actual data, and calculates a deviation from the primary value of the width of the slab 156, the slab width deviation storing means 108 which stores the calculated slab width deviation, the width
estimation means 102 which estimates a width of the steel plate 157 having undergone rolling from the width of the slab 156 and rolling actual data, the width deviation storing means 104 which stores a deviation between an estimated value and an actually-obtained width, the training means 109 which trains the tendency of the width deviation of the slab 156 from the content of the slab width deviation storing means 108 and outputs a training result, the adaptation means 105 which estimates a width estimation error from the content of the width deviation storing means 104 and outputs an estimation result, and the preset control means 101 which selectively uses the output of the training means and the output of the adaptation means to conduct width estimation operation based on a correlation between the width of the slab 156 to be rolled and the width of the slab 156 rolled recently and calculates the set value of the edger 151 that is an operation end for a width controlling in accordance with an estimation result.
Accordingly, in a width controlling in a hot rolling, it is possible to highly-accurately control the width of a rolling-target material like the steel plate 157 by a simple calculation focused on the correlation of the widths of the slabs 156 rolled previously and recently.
The present invention can be widely applied to a width controlling for a hot rolling mill. «Third Embodiment»
FIG. 13 is a diagram for explaining a width control device according to the third embodiment. A width control device 1100 receives various signals from a control target 1200 and outputs a control signal to the control target 1200. First, an explanation will be given of the structure of the control target 1200.
The control target 1200 is a width control line of a hot rolling mill, rolls a slab 1205 having a width of 1000 to 1600 mm ejected from a furnace 1215, which is a slab ejection unit 1210, in a width direction using an edger mill 1225 of a edger rolling unit 1220, and performs reversing rolling of three to seven paths using a rougher rolling mill 1235 of a rougher rolling unit 1230. The edger mill 1225 and the rougher rolling mill 1235 are disposed adjacently,
and a reverse rolling is carried out simultaneously at the edger mill 1225 and the rougher rolling mill 1235.
Thereafter, front/back-direction horizontal rolling of three to seven paths is performed by a steckel mill finishing mill 1245 of a finish rolling unit 1240, and a rolling-target material is coiled up by a down coiler 1255 which is a coiling device of a coiling unit 1250. A rougher delivery-side width gauge 1261 measures a width of the slab 1205 immediately after rolled by the edger mill 1225 and the rougher rolling mill 1235. A finish delivery-side width gauge 1262 measures a width of a steel plate immediately after rolled by the finish rolling unit 1240. Here, a target width of the slab 1205 after a final path rolling by the finishing mill 1245 is called a finish target width. Moreover, a target width after a final path rolling by the rougher rolling mill 1235 is called a rougher target width.
The purpose of a width controlling is to conform a width measured by the finish delivery-side width gauge 1261 to a target width instructed by a production instruction unit 1110.
Next, an explanation will be given of the structure of a width control device 1100. The width control device 1100 has the production instruction unit 1110 which instructs a finish target width of a coil, finish-width-shrinkage estimation amount calculation means 1121 which estimates and calculates a width shrinkage amount at the finishing mill 1245, and first width target value correction means 1120 which corrects a finish target width using an output from the finish-width-shrinkage estimation amount calculation means 1121.
Moreover, the width control device 1100 has rougher width actual data gathering means 1170 which gathers width actual data of the rougher delivery-side width gauge 1261, finish width actual data gathering means 1180 which gathers width actual data of the finish delivery-side width gauge 1262, finish-width-shrinkage adaptive amount calculation means 1122 which estimates a finish width shrinkage amount from a difference between a width measured by the rougher delivery-side width gauge 1261 and a width measured by the finish
delivery-side width gauge 1262, and a finish-width-shrinkage adaptive amount storing table 1131 which stores a finish-width-shrinkage adaptive amount calculated by the finish-width-shrinkage adaptive amount calculation means 1122.
The width control device 1100 further has a width expansion model 1150 which estimates a width at the delivery side of the rougher rolling mill from each reduction and roll gap of the edger mill 1225 and the rougher rolling mill 1235, or hardness information of a steel product, edger mill setup means 1160 which determines a control instruction for the edger mill using the width expansion model 1150, rougher width adaptive amount calculation means 1141 which quantify a model error of the width expansion model 1150 from a difference between a width estimated by the width expansion model 1150 and a corresponding width detected by the rougher delivery-side width gauge 1261, a rougher width adaptive amount storing table 1151 which stores a rougher width adaptive amount calculated by the rougher width adaptive amount calculation means 1141, and second target width correction means 1140 which corrects a target width using an output from the rougher width adaptive amount calculation means 1141.
Still further, the width control device 1100 has finish-width-shrinkage-adaptive-amount effectiveness quantification means 1123 which acquires, for each steel plate rolled most recently, a difference between a width measured by the rougher delivery-side width gauge 1261 and a width measured by the finish delivery-side width gauge 1262 and quantifies a correlation of differences for each steel plate, and a finish-width-shrinkage adaptive gain storing table 1132 which stores a finish-width-shrinkage adaptive gain quantified by the finish-width-shrinkage-adaptive-gain effectiveness quantification means 1123. Yet further, the width control device 1100 has rougher-width-adaptive-amount effectiveness quantification means 1142 which acquires, for each steel plate rolled most recently, a difference between a width measured by the rougher delivery-side width gauge 1261 and a width estimated by the width expansion model 1150 and quantifies a correlation of differences for each steel plate,
and a rougher-width-adaptive-gain storing table 1152 which stores a rougher width adaptive gain quantified by the rougher-width-adaptive-amount effectiveness quantification means 1142.
In the finish-width-shrinkage estimation amount calculation means 1121, a fmish-width-shrinkage estimation amount Shrink_pre can be acquired from, for example, equation (15).
(Equation Removed)
where ph is a finishing mill target thickness and A, B, C and D are width shrinkage amount calculation parameters.
The parameters A, B, C and D in equation (15) are classified for a kind of rolled steel and for a slab thickness after finish rolling. The finish-width-shrinkage adaptive amount calculation means 1122 corrects an error between the width-shrinkage estimation amount Shrink_pre calculated from equation (15) and an actual width shrinkage amount Shrink_act.
A width-shrinkage-amount error AShrink(Br_act) when an actual width of the slab having undergone rolling by the rougher rolling mill, i.e., an actual width of the slab prior to a finish rolling is Br_act is stored in a table (see FIG. 13) classified for the kind of rolled steel, for a slab thickness after finish rolling, and for the number of finish total paths. Once width-shrinkage-amount errors ΔShrink(Br_act, i) (i = 1, 2, ... n) for n number of coils corresponding to some classification are stored, the width-shrinkage-amount errors ΔShrink are approximated by an m-th order approximation equation expressed as equation (16) as a function of a width Br of the slab prior to finish rolling. The approximated width-shrinkage-amount correction amount is called Shrink_comp. Note that the calculation of the width-shrinkage-amount correction amount Shrink_comp is independently executed for each classification.
(Equation Removed)
where Br is a width of the slab prior to finish rolling, and ak is a parameter.
The parameter ak is stored in a finish-width-shrinkage adaptive amount storing table 1131, which is classified for the kind of rolled steel, the slab thickness after finish rolling, and the number of finish total paths.
FIG. 14 is a diagram for explaining an example of the table (finish-width-shrinkage adaptive amount table) when m = 2.
As an adaptive amount effectiveness is quantified by the
finish-width-shrinkage-adaptive-amount effectiveness quantification means 1123, i.e., as an adaptive gain a is determined, a finish-width-shrinkage adaptive amount calculated from equation (16) is optimally used. How to determine the adaptive gain a will now be explained. The adaptive gain a is determined based on an autocorrelation of actual data error sequences.
A width-shrinkage-amount model error Cm which is a difference between a corrected width shrinkage amount, which is acquired by adding a width shrinkage correction amount acquired from equation (16) to a finish width shrinkage amount Shrink_pre calculated from equation (15), and an actual width shrinkage amount Shrink_act is expressed as equation (17).
(Equation Removed)
Accordingly, if a model error at the time of previous coil rolling is (Cm)n-l and a model error at the time of last but one coil rolling is (Cm)n-2, an error vector can be expressed as equation (18).
(Equation Removed)
where (Cm)i is a width shrinkage amount model error observed from a coil previous to an i-th coil.
Here, (Cm)i is set as Xi, and (Cm)i-l is set as (Yl)i, and a data sequence for Xi and (Yl)i are created. Accordingly, model errors acquired from adjacent coils are to be associated with each other as a data sequence. Likewise, a data sequence for Xi and (Y2)i are created with (Cm)i being set as Xi and (Cm)i-2 being set as (Y2)i. Accordingly, model errors for coils adjacent to each other with one intervening coil are constructed as a corresponding data sequence. Next, using the individual data sequences, cross correlations Covl, Cov2 between Xi and (Yl)i, and Xi and (Y2)i are calculated from equations (19), (20).
(Equation Removed)
Cov(x, y) can be expressed by equation (21).
(Equation Removed)
Covl represents a largeness of the relativity of model errors for adjacent coils, and if Covl is large, it means that the correlation of errors is also large. Likewise, if Cov2 is large, it means that there is a large correlation between coils adjacent to each other with one intervening coil. When the correlation is large, a model error and a correction amount
Shrink_comp can be directly associated with each other. Conversely, when the correlation is small, the relativity between model errors is small, and the gain cannot be set large. Thus, in accordance with the values of Covl, Cov2, a value acquired by multiplying those values by appropriate constants bl, b2 is set as the gain vector a (refer to equation (22)).
(Equation Removed)
According to the foregoing calculation, a correction amount Bf_comp for a finish target width for calculating a rougher target width by the first width target value correction means 1120 can be expressed as equation (23).
(Equation Removed)
Therefore, the rougher target width Br_tgt can be expressed as equation (24) using a finish target width Bfjgt of the production instruction unit 1110.
(Equation Removed)
The gain a is stored in the finish-width-shrinkage adaptive gain storing table 1132 which is classified for the kind of rolled steel, for the slab thickness after finish rolling, and for the number of finish total paths.
FIG. 15 is a diagram showing an example of the table (finish-width-shrinkage adaptive gain table) when n = 2.
FIG. 16 is a diagram showing an example of a relationship between the number of rolled coils and a model error. As shown in the figure, by performing correction using actual data of the large number of rolled coils, a model error decreases. FIG 17 is a diagram showing a
relationship between a width before and after width shrinkage correction and a model error.
FIG. 18 is a diagram for explaining the process executed by the first width target value correction means 1120 up to a calculation of a finish width correction amount.
The first width target value correction means 1120 acquires a finish target thickness which is a production instruction in a step S106-1. The first width target value correction means 1120 calculates a finish-width-shrinkage estimation amount Shrink_pre using the acquired finish target thickness and equation (15) in a step S106-2. The first width target value correction means 1120 checks the current number of rolled slabs in a step SI 06-3, and if the number of rolled slabs matches a number N which is an adaptive amount calculation rolling number set beforehand, the process progresses to an adaptive amount calculation flow starting from a step SI06-4. The first width target value correction means 1120 calculates a finish-width-shrinkage adaptive amount Shrink_comp in accordance with a calculation scheme of equation (16) in a step SI 06-4, and calculates an adaptive gain a in accordance with a calculation scheme of equation (22) in a step SI 06-5. The first width target value correction means 1120 initializes the rolling count, i.e., returns it to 0, and terminates the adaptive amount calculation flow.
The process progresses to a step SI06-7, the first width target value correction means 1120 calculates a finish width correction amount Bf_comp using equation (23) in the first width target value correction means 1120 from the finish-width-shrinkage estimation amount Shrink_pre calculated in the step SI06-2, the finish-width-shrinkage adaptive amount Shrink_comp calculated in the step SI06-4, and the adaptive gain a calculated in the step SI06-5, and terminates the calculation.
When the number of rolled stabs is less than the number N in the step SI 06-2, the process does not progress to the adaptive amount calculation flow, but directly progresses to the step SI06-7, and the first width target value correction means 1120 calculates a finish width correction amount Bf_comp using equation (23) from the finish-width-shrinkage
estimation amount Shrink_pre calculated in the step SI 06-2, a previous value of the adaptive amount Shrink_comp, and a previous value of the adaptive gain a, and terminates the calculation.
Next, the rougher width adaptive amount calculation means 1141 quantify a model error of the width expansion model 1150 using a width estimated by the width expansion model 1150 and a rougher actual width acquired by the rougher width actual data gathering means 1170, and calculates a correction value with the quantified value being a rougher target width 1001. The width expansion model 1150 comprises two model formulas for estimating a width expansion amount of a portion called a dog-bone and a portion called a rectangle. There are several model formulas known as such model formulas, and examples of those are equations (25) and (26).
(Equation Removed)
where,
(Equation Removed)
where:
dwl is a dog-bone width expansion amount (mm); dw2 is a rectangle width expansion amount (mm); BO is an edger mill entry side width (mm); B is an edger mill delivery-side width (mm); RE is an edger mill roll diameter (mm); Se is an edger mill gap (mm);
Ld is (R-(H-h))l/2 (mm);
α, ß,  and ξ, are width expansion calculation constant parameters (-);
R is a rougher rolling mill work roll radius (mm);
H is a rougher entry side thickness (mm); and
h is a rougher delivery-side thickness (mm).
A rougher width error difference Br_comp which is a difference between the width estimated by the width expansion model 1150 and actual width is stored in a table which is classified for the kind of rolled steel, for the slab width that is a rougher first path entry side width, and for the slab thickness after rougher rolling. Once rougher width errors Bri_comp (i = 1, 2,... n) of n number of coils corresponding to a classification are stored, the rougher width errors, i.e., a rougher width correction amount Br_comp is approximated by an m-th order approximation equation shown as equation (27) as a function of a rougher target width Br_tgt. The approximation of the rougher width correction amount Br_comp is independently executed for each classification.
(Equation Removed)
where Br_tgt is a target width after rougher final path rolling, and ck is a parameter.
The parameter ck is stored in the rougher width adaptive amount storing table 1151 which is classified for the kind of rolled steel, for the slab width, and for the slab thickness after rougher rolling.
FIG. 19 is a diagram showing an example of the table (rougher width adaptive amount storing table) when n =2.
Moreover, by quantifying the adaptive amount effectiveness by the
rougher-width-adaptive-amount effectiveness quantification means 1142, i.e., by determining an adaptive gain γ, an adaptive amount calculated from equation (27) is optimally used.
Regarding how to determine the adaptive gain y, like the
finish-width-shrinkage-adaptive-amount effectiveness quantification means 1123, the adaptive gain y is determined based on an autocorrelation of actual error sequences. The determined adaptive gain y is stored in the rougher width adaptive gain storing table 1152 classified for the kind of rolled steel, for the slab width, and for the slab thickness after rougher rolling. Accordingly, a correction value B'r_comp for a target width at a rougher final path is calculated from equation (28) by the second target value correction means 1140.
(Equation Removed)
Using the calculated rougher target width correction amount B'r_comp, a corrected rougher target width 1002 Br_tgt_new can be expressed as equation (29).
(Equation Removed)
The edger mill setup means 1150 determines a gap primary value and a rougher width target value Bri_tgt (i is the number of paths) at each rougher path from the corrected rougher target width 1002 Br_tgt_new and the model formulas expressed as equation (25) and equation (26). The maximum reduction amount and the minimum reduction amount of the edger mill determined in advance and the rougher target width Br_tgt are used for such determination. First, using only the maximum reduction amount of the edger mill, rolling is carried out from a first path to a final path. Next, using only the minimum reduction amount of the edger mill, rolling is gone back from a target width Br_tgt of the final path to one of the first path.
FIG. 20 shows examples where rolling is performed using the maximum reduction amount of the edger mill and where rolling is gone back using only the minimum reduction amount. When there is an intersection point P between an m-th path and (m+l)-th path, as shown in Fig. 21, the reduction amount at the time of (m+l)-th path rolling is corrected. The
corrected reduction amount is output to the edger mill 1225 as the control primary value.
FIG. 22 is a diagram for explaining a process up to a determination of the edger mill reduction amount control primary value for each rougher path by correcting a rougher target width 1001.
In a step S1010-1, a finish target width Bfjgt and a rougher target width 1001 ((Br_tgt) calculated by the first width target value correction means 1120 are acquired. In a step S1010-2, the current number of rolled slabs is checked, and when the number of rolled slabs matches a number N which is an adaptive amount calculation rolling number set in advance, the process progresses to an adaptive amount calculation flow starting from a step S1010-3. In the step S1010-3, a rolling count is initialized. In a step S1010-4, a rougher width adaptive amount Br_comp is calculated in accordance with a calculation scheme of equation (27), and in a step S1010-5, an adaptive gain y is calculated in accordance with a calculation scheme of equation (22). In a step SI 010-6, a rougher target width correction amount B'r_comp is calculated using equation (27) of the second width target value correction means 1140 from the rougher width adaptive amount Br_comp calculated in the step S1010-4 and the adaptive gain y calculated in the step S1010-5. In a step S1010-7, the rougher target width correction amount calculated in the step S1010-6 is added to the rougher target width 1001 acquired in the step S1010-1, thereby calculating a new rougher target width 1002. In a step SI010-8, a gap primary value for the edger mill is calculated using the rougher target width 1002 calculated in the step SI010-7 and the edger mill setup means 1160, and then the calculation is terminated.
When the number of rolled slabs is less than the number N in the step S1010-2, the process does not progress to the adaptive amount calculation flow, but progresses to the step S1010-7, and a rougher target width 1002 is calculated from the rougher target width 1001 acquired in the step S1010-1, the previous value of the adaptive amount Br_comp, and the previous value of the adaptive gain y, and then the calculation is terminated.
As explained above, according to the third embodiment, the width control device 1100 has the finish-width-shrinkage estimation amount calculation means which estimates and outputs a width shrinkage amount at the finishing mill, and the first width target value correction means which corrects the finish target width using an output from the finish-width-shrinkage estimation amount calculation means. Moreover, the width control device 1100 has a rougher width actual data gathering means which gathers width actual data of the rougher delivery-side width gauge, the finish width actual data gathering means which gathers width actual data of the finish delivery-side width gauge, the finish-width-shrinkage adaptive amount calculation means which estimates a width shrinkage amount from a difference between a width measured by the rougher delivery-side width gauge and a width measured by the finish delivery-side width gauge, the edger mill setup means which determines a control instruction of the edger mill using the width expansion model that estimates a width at the rougher rolling mill delivery-side from the reduction amounts of the edger mill and the rougher rolling mill, the roll gap, and the hardness information of a steel product, the rougher width adaptive amount calculation means which quantifies the model error of the width expansion model from a difference between a width estimated by the width expansion model and a corresponding width detected by the rougher delivery-side width gauge, and the second width target value correction means which corrects a target width using an output from the rougher width adaptive amount calculation means.
Moreover, the width control device 1100 has the
finish-width-shrinkage-adaptive-amount effectiveness quantification means which acquires a difference between a width measured by the rougher delivery-side width gauge and a width measured by the finish delivery-side width gauge for each steel plate rolled most recently and quantifies a correlation of differences for each steel plate, and a rougher-width-adaptive-amount effectiveness quantification means which acquires a difference between a width measured by the rougher delivery-side width gauge and a width
estimated by the width expansion model for each steel plate rolled most recently and quantifies a correlation of differences for each steel plate.
According to the third embodiment, a finish-width-shrinkage estimation amount calculated beforehand is corrected with actual width shrinkage amount which is a difference between an actual rougher delivery-side width gathered by the rougher delivery-side width gauge and an actual finish delivery-side width gathered by the finish delivery-side width gauge, and a finish-width-shrinkage adaptive amount which is the corrected amount is corrected by the quantification means which quantifies the effectiveness of an actual width shrinkage amount. That is, a finish width shrinkage amount is accurately estimated at the stage of presetting, and the estimated value is set to be highly accurate by an online adaptive process. Accordingly, the most appropriate rougher target width after rougher rolling can be determined in such a way that a width after finish rolling can be a target width. Moreover, since the reduction amount at each path of the edger mill is determined in order to accomplish the determined rougher target width, a width controlling independent from a finish rolling is enabled.
Further, since a rougher width primary value is set appropriately, it is possible to improve the accuracy of the finish width. In addition, a highly-accurate width can be achieved for various kinds of steel products and steel plates having various widths. This results in elimination of an excessive width and an inadequately-small width, thereby improving the yield of a product.

CLAIMS
1. A width control device for a hot rolling mill, the width control device controlling a
slab of a rolling-target material, which is casted by a continuous caster and delivered into a
rougher rolling mill, to have a predetermined slab width using an edger, and then controlling
the rolling-target material after rolled by a finishing mill to have a width matching a target
value, the width control device comprising:
a width prediction model that comprises a relationship between a width of the slab and widths of the rolling-target material during a rolling and after the rolling;
a width estimation unit that estimates a width of the rolling-target material after rolled by the finishing mill using the width prediction model from a primary value of a width of the slab and rolling information on the slab loaded from the hot rolling mill;
a width deviation storing unit that stores a deviation between an estimation result of the width estimation unit and a width actual measured value of the rolling-target material after rolling for a rolled rolling-target material;
an adaptation unit that corrects a target value of an after-rolling width of the rolling-target material to be rolled based on a content of the width deviation storing unit; and
a preset control unit that calculates and outputs a control instruction to the edger through a calculation using the target value of the width corrected by the adaptation unit.
2. A width control device for a hot rolling mill, the width control device controlling a
slab of a rolling-target material, which is casted by a continuous caster and delivered into a
rougher rolling mill, to have a predetermined slab width using an edger, and then controlling
the rolling-target material after rolled by a finishing mill to have a width matching a target
value, the width control device comprising:
a width prediction model that comprises a relationship between a width of the slab and
widths of the rolling-target material during a rolling and after the rolling;
a slab width estimation unit that estimates a width of the slab of the rolling-target material before a rolling by a back-calculation of the width prediction model from a width actual measured value of the rolling-target material after rolling and rolling information on the rolling-target material loaded from the hot rolling mill;
a slab width deviation storing unit that stores a deviation between an estimation result of the slab width estimation unit and a primary value of a width of the slab;
a training unit that corrects the primary value of the width of the slab based on a content of the slab width deviation storing unit; and
a preset control unit that calculates and outputs a control instruction to the edger to set a width of the rolling-target material after rolling to be a target value through a calculation using the width of the slab corrected by the training unit.
3. A width control device for a hot rolling mill, the width control device controlling a slab of a rolling-target material, which is casted by a continuous caster and delivered into a rougher rolling mill, to have a predetermined slab width using an edger, and then controlling the rolling-target material after rolled by a finishing mill to have a width matching a target value, the width control device comprising:
a width prediction model that comprises a relationship between a width of the slab and widths of the rolling-target material during a rolling and after the rolling;
a width estimation unit that estimates a width of the rolling-target material after rolled by the finishing mill using the width prediction model from a primary value of a width of the slab and rolling information on the slab loaded from the hot rolling mill;
a width deviation storing unit that stores a deviation between an estimation result of the width estimation unit and a width actual measured value of the rolling-target material after rolling for a rolled rolling-target material;
an adaptation unit that corrects a target value of a width of the rolling-target material after rolling based on a content of the width deviation storing unit;
a slab width estimation unit that estimates a width of the slab of the rolling-target material before rolling by a back-calculation of the width prediction model from a width actual measured value of the rolling-target material after rolling and rolling information on the rolling-target material loaded from the hot rolling mill;
a slab width deviation storing unit that stores a deviation between an estimation result of the slab width estimation unit and a primary value of a width of the slab;
a training unit that corrects the primary value of the width of the slab based on a content of the slab width deviation storing unit; and
a preset control unit that calculates and outputs a control instruction to the edger to set a width of the rolling-target material after rolling to be a target value through a calculation selectively using either one of an output from the training unit or an output from the adaptation unit.
4. The width control device according to claim 3, further comprising a change-over
unit which changes over an output to be input to the preset control unit between an output
from the training unit and an output from the adaptation unit, and outputs the changed-over
output to the preset control unit,
wherein the change-over unit acquires a number of rolled slabs after the continuous caster starts casting, selects the output from the training unit when the number of rolled slabs is not greater than a predetermined number, and selects the output from the adaptation unit when the number of rolled slabs is greater than the predetermined number.
5. The width control device according to claim 3, further comprising a change-over
unit which changes over an output to be input to the preset control unit between an output
from the training unit and an output from the adaptation unit, and outputs the changed-over output to the preset control unit,
wherein the change-over unit acquires a number of rolled slabs after the continuous caster changes a width of a mold, selects the output from the training unit when the number of rolled slabs is not greater than a predetermined number, and selects the output from the adaptation unit when the number of rolled slabs is greater than the predetermined number.
6. The width control device according to claim 3, further comprising a change-over
unit which changes over an output to be input to the preset control unit between an output
from the training unit and an output from the adaptation unit, and outputs the changed-over
output to the preset control unit,
wherein the change-over unit acquires a number of rolled slabs after the continuous caster starts casting or after the continuous caster changes a width of a mold, selects the output from the training unit when the number of rolled slabs is not greater than a predetermined number, and selects the output from the adaptation unit when the number of rolled slabs is greater than the predetermined number.
7. The width control device according to claim 3, further comprising a change-over
unit which changes over an output to be input to the preset control unit between an output
from the training unit and an output from the adaptation unit, and outputs the changed-over
output to the preset control unit,
wherein the change-over unit determines a degree of similarity between a width of the slab rolled recently and a width of the slab to be rolled next, selects the output from the training unit when the degree of similarity is not greater than a predetermined value, and selects the output from the adaptation unit when the similarity is greater than the predetermined value.
8. The width control device according to claim 1 or any one of claims 3 to 7, further
comprising an adaptive gain calculation unit which acquires a largeness of a correlation
between width deviations of the rolling target materials rolled successively from a content of
the width deviation storing unit, and determines an adaptive gain based on the largeness of the
correlation,
wherein the adaptation unit calculates a correction amount of a width target value of the rolling-target material to be rolled next using a value acquired by multiplying a width deviation extracted from the width deviation storing unit by the adaptive gain;
wherein the preset control unit calculates and outputs a control instruction to the edger to set a width of the rolling-target material after rolling to be a width target value using a target width corrected based on an output from the adaptation unit.
9. The width control device according to claim 8,
wherein the adaptive gain calculation unit outputs a large adaptive gain when the correlation between the width deviations of the rolling-target materials rolled successively is large, and outputs a small adaptive gain when the correlation is small.
10. A width control method of a width control device for a hot rolling mill, the width
control device controlling a slab of a rolling-target material, which is casted by a continuous
caster and delivered into a rougher rolling mill, to have a predetermined slab width using an
edger, and then controlling the rolling-target material after rolled by a finishing mill to have a
width matching a target value, the method comprising:
storing a relationship between a width of the slab and widths of the rolling-target material during a rolling and after the rolling;
estimating a width of the slab of the rolling-target material before rolling from a width
actual measured value of the rolling-target material after rolling and rolling information on the rolling-target material loaded from the hot rolling mill by a back-calculation of the relationship between the width of the slab and the widths of the rolling-target material during a rolling and after the rolling;
storing a deviation between an estimation result and a primary value of the width of the slab in a first table;
estimating a width of the slab, a relationship between widths of the rolling-target material during a rolling and after the rolling, and a width of the rolling target material rolled by the finishing mill, using the primary value of the width of the slab and rolling information on the slab loaded from the hot rolling mill;
calculating a deviation between a width of the estimation result and the width actual measured value for a rolled rolling-target material, and storing the deviation in a second table;
acquiring a number of slabs rolled after the continuous caster starts casting or after the continuous caster changes a width of a mold, and correcting a primary value of a width of the slab to be rolled based on a content of the first table when the number of rolled slabs is not greater than a predetermined number;
correcting a width target value of the rolling-target material to be rolled based on a content of the second table when the number of rolled slabs is greater than the predetermined number; and
calculating and outputting a control instruction to the edger to control a width of the rolling-target material after rolling to be a predetermined value in accordance with a correction result.
11. A width control device for a hot rolling mill comprising an edger mill that rolls a heated steel product in a width direction; a rougher rolling mill which is provided adjacent to the edger mill and rolls the steel product rolled by the edger mill in a thickness direction; and
a finishing mill that further rolls the steel product rolled by the edger mill and the rougher mill in a thickness direction to produce a hot coil, the width control device controlling a width of the hot coil to be a target width given by a control instruction, the width control device comprising:
a finish-width-shrinkage estimation amount calculation unit that calculates and estimates a width shrinkage amount at the finishing mill; and
a first width target value correction unit that corrects the target width using an output from the fmish-width-shrinkage estimation amount calculation unit,
wherein the control instruction calculated based on a target width corrected by the first width target value correction unit is output to the edger mill.
12. A width control device for a hot rolling mill comprising an edger mill that rolls a heated steel product in a width direction; a rougher rolling mill which is provided adjacent to the edger mill and rolls the steel product rolled by the edger mill in a thickness direction; a finishing mill that further rolls the steel product rolled by the edger mill and the rougher mill in a thickness direction to produce a hot coil; a rougher delivery-side width gauge provided at a delivery-side of the rougher rolling mill; a finish delivery-side width gauge provided at the delivery-side of the finishing mill; a rougher width actual data gathering unit that gathers width actual data of the rougher delivery-side width gauge; and a finish width actual data gathering unit that gathers width actual data of the finish delivery-side width gauge, the width control device controlling a width of the hot coil to be a target value given by a control instruction, width control device comprising:
a finish-width-shrinkage estimation amount calculation unit that calculates and estimates a width shrinkage amount at the finishing mill;
a finish-width-shrinkage adaptive amount calculation unit that estimates a width shrinkage amount from a difference between a width measured by the rougher delivery-side
width gauge and a width measured by the finish delivery-side width gauge; and a first width target value correction unit that corrects the target width, wherein the first width target value correction unit corrects the target width using an output from the finish-width-shrinkage estimation amount calculation unit and an output from the finish-width-shrinkage adaptive amount calculation unit.
13. The width control device according to claim 11 or 12, further comprising:
a width expansion model that estimates a width at the delivery-side of the rougher rolling mill from a reduction amount, a roll gap, and hardness information of a steel product at the edger mill and the rougher rolling mill;
an edger mill setup unit that determines a control instruction to the edger mill using the width expansion model;
a rougher width adaptive amount calculation unit that quantifies a model error of the width expansion model from a difference between a width estimated by the width expansion model and a width detected by the rougher delivery-side width gauge;
a second width target value correction unit that further corrects a target width corrected by the first width target value correction unit using an output from the rougher width adaptive amount calculation unit,
wherein the control instruction calculated based on the corrected target width is output to the edger mill.
14. The width control device according to claim 11 or 12, further comprising:
a finish-width-shrinkage-adaptive-amount effectiveness quantification unit which acquires a difference between a width measured by the rougher delivery-side width gauge and a width measured by the finish delivery-side width gauge for each of plural steel products rolled recently, and quantifies a correlation of differences for each steel product,
wherein the first width target value correction unit estimates a width shrinkage amount based on an output from the finish-width-shrinkage adaptive amount calculation unit when the correlation is large, estimates a width shrinkage amount based on an output from the finish-width-shrinkage estimation amount calculation unit when the correlation is small, and estimates a width shrinkage amount while changing a ratio between the output from the finish-width-shrinkage adaptive amount calculation unit and the output from the finish-width-shrinkage estimation amount calculation unit in accordance with a largeness of the correlation.
15. The width control device according to claim 11 or 12, further comprising:
a rougher-width-adaptive-amount effectiveness quantification unit which acquires a difference between a width measured by the rougher delivery-side width gauge and a width estimated by the width expansion model for each of plural steel products rolled recently, and quantifies a correlation of differences for each steel product,
wherein the second width target value correction unit determines a gain for correction for correcting the target width using an output from the rougher-width-adaptive-amount calculation unit in accordance with a largeness of the correlation.
16. A width control method for a hot rolling mill which rolls a heated steel product in a
width direction, rolls the rolled steel product in a thickness direction to produce a rougher
material, and further rolls the rougher material in a thickness direction to produce a hot coil
having a width matching a target value, the method comprising:
calculating a width shrinkage estimation amount from a difference between a measured width of the rougher material and a width of the hot coil; and
correcting the target value using the width shrinkage estimation amount and calculating a width target value for rolling a steel product in a width direction.
17. The width control method for a hot rolling mill according to claim 16, the hot rolling mill comprising an edger mill that rolls a heated steel product in a width direction; a rougher rolling mill that rolls the steel product rolled by the edger mill in a thickness direction; a finishing mill that further rolls the steel product rolled by the roughe/ rolling mill in a thickness direction to produce a hot coil; and a width expansion model that estimates a width at a delivery-side of the rougher rolling mill, the method comprising:
quantifying a model error of the width expansion model from a difference between a width estimated by the width expansion model and a measured width of the rougher material; and
correcting the target value using the width shrinkage estimation amount and a value of the model error, and calculates a width target value for rolling a steel product in the width direction.