CN102794310B - Rolling model optimization device - Google Patents
Rolling model optimization device Download PDFInfo
- Publication number
- CN102794310B CN102794310B CN201110199333.8A CN201110199333A CN102794310B CN 102794310 B CN102794310 B CN 102794310B CN 201110199333 A CN201110199333 A CN 201110199333A CN 102794310 B CN102794310 B CN 102794310B
- Authority
- CN
- China
- Prior art keywords
- model
- value
- boundary value
- formation
- rolling
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Control Of Metal Rolling (AREA)
Abstract
The present invention provides a rolling model optimization device capable of connecting predicted values of adjacent models without gradient, comprising: a boundary value table for storing boundary values representing boundaries of a plurality of models; a model formula connection boundary value change part for changing the boundary values stored in the boundary value table according to a rolled actual value; a model formula priority table for storing priority of the plurality of models; an evaluation part for determining the model suitable for a rolling requirement based on the boundary values obtained from the boundary value table and the priority obtained from the model formula priority table in the case of each input calculation condition; a model switching part for switching the model to the determined model; a learning calculation part for calculating a learning coefficient based on a difference between the calculated predicted value and the actual value for each model in the plurality of models; and a model calculation part for calculating the predicted value of the model by utilizing a model formula of the model switched by the model switching part and the learning coefficient calculated by the learning calculation part.
Description
Technical field
The present invention relates to a kind of rolling model optimization device, this rolling model optimization device is according to rolling condition, distinguishes and uses multiple models, the plurality of model to show the model for the rolling mill practice of rolled metal etc.
Background technology
In the situation that rolling mill practice is controlled, conventionally, generate the rolling model represented by the model formation that represents rolling phenomenon (following, sometimes be only called " model "), this model is applied to rolling condition, and rolling load, material temperature and deflection etc. are predicted, with the value that determines milling train to set, make scantling and temperature etc. after rolling become desirable value.Therefore,, in order to be rolled with higher precision, need correctly to represent the model formation of rolling phenomenon.
Whole region at the object that becomes prediction and calculation all has single physical characteristic, is suitable for a model, adjusted to consistent with actual rolling result, thereby can carry out stable prediction and calculation.But, there is following situation:, become in the region of object of prediction and calculation and mixing the part with different physical characteristics, thereby sometimes cannot show with a model.In this case, can, by each rolling condition preparation model, try hard to improve the precision of prediction in the region that physical characteristic is different.In the case of each region is suitable for different models, predetermine the boundary value on the border that represents each model, and model is adjusted, make each model on the border represented by this boundary value not have jump,, each model can be connected continuously.In addition, each model in multiple models has learning functionality, thereby can, by learning, try hard to improve precision of prediction in regional.
As such technology, for example, patent documentation 1 has disclosed a kind of plate width controlling method of continuous-rolling.In this technology, due to the outlet side plate of milling train is wide and entrance side tension force between there is nonlinear relation, therefore, prepare scope multiple and entrance side tension force corresponding to carry out the model of width prediction, under the tension force by the represented border of predefined boundary value, without jump connect the border of each model.Then, for minimizing system when the switching model becomes unsettled situation, the difference of and its desired value wide in order to the outlet side plate in each prediction moment square evaluate multiple models as parameter, single evaluation function, implement thus the wide control of plate.
Patent documentation 1: Japanese Patent Laid-Open 2006-150371 communique
Summary of the invention
In the plate width controlling method of the continuous-rolling disclosing at above-mentioned patent documentation 1, owing to showing non-linear stronger rolling mill practice with multiple models that carry out width prediction, therefore, although can improve precision of prediction, but in the time showing single rolling mill practice with multiple models, there is following problem.
That is, the more difficult boundary value predetermining for switching model, if boundary value is set in beyond the scope of application of a model, near this boundary value, precision of prediction may decline.Thereby, wish to have the function that can confirm actual rolling result while suitably change boundary value.
In addition, changing boundary value in the situation that, due to cannot be on the border being represented by boundary value link model automatically, therefore, there is following problem:, while changing boundary value, all need to revise according to analysis result the parameter of model each.
In addition, even all have the function of learning at multiple models, for by the represented border of boundary value, owing near the learning outcome border of a model cannot being reflected on another model, therefore, there is following problem:, exist predicted value on border, to become discontinuous situation.
The present invention completes in order to address the above problem and respond requirement, its object is, a kind of rolling model optimization device is provided, and this rolling model optimization device can be set the suitable application region of most suitable model, and can without jump connect the predicted value of adjacent model.
In order to address the above problem, the invention is characterized in, comprising: boundary value form, these boundary value form his-and-hers watches show that the boundary value on the border of multiple models stores; Model formation fillet dividing value changing unit, the actual value of this model formation fillet dividing value changing unit based on rolling, changes the boundary value being stored in boundary value form; Model formation priority form, this model formation priority form stores the priority of multiple models; Evaluation section, this evaluation section is under each inputted design conditions, and the boundary value based on getting from boundary value form and the priority getting from model formation priority form, decide the model that is applicable to rolling condition; Model switching part, this model switching part switches to model on the model being determined by evaluation section; Study calculating part, this study calculating part, to each model in multiple models, calculates learning coefficient according to the difference of calculated predicted value and actual value; And model calculating part, the model formation of the model that this model calculating part utilization is switched by model switching part and by the learning coefficient that calculates of study calculating part, calculates the predicted value of this model.
According to the present invention, owing to changing the border of multiple models of the single technique of performance, therefore, can set the suitable application region of most suitable model, and can be on the border of model without jump connect predicted value.Consequently, owing to setting the region that is suitable for each model, therefore, can try hard to improve the precision of prediction of model.
Brief description of the drawings
Fig. 1 is the block diagram that represents the structure of the related rolling model optimization device of embodiments of the invention 1.
Fig. 2 is the figure describing for the summary of computing related to embodiments of the invention 1, that carry out with the model formation connection correction coefficient operational part of rolling model optimization device.
Fig. 3 is the block diagram that represents the structure of the related rolling model optimization device of embodiments of the invention 2.
Fig. 4 is the flow chart that represents embodiments of the invention 2 calculation procedure related, that undertaken by the model formation fillet dividing value changing unit of rolling model optimization device.
Label declaration
1 design conditions input unit
2 model optimization devices
3 control device
11 evaluation sections
12 boundary value forms
13 model switching parts
14 model formation storage parts
15 study calculating parts
16 learning coefficient storage parts
17 model calculating parts
18 model calculated value storage units
21 model formation fillet dividing value changing units
22 model formation priority forms
23 model formations connect correction coefficient operational part
24 model formations connect correction coefficient storage part
Detailed description of the invention
Below, with reference to accompanying drawing, embodiments of the present invention are elaborated.In addition, below, describe as example taking the situation of prediction and calculation of the rolling load that applies the present invention to single milling train.
[embodiment 1]
Generally, for example, represent load model with formula (1), represent coefficient of friction model with formula (2), can predict rolling load thus.
P
i＝Z
Pi·km
i·L
i·W·Q
i(H，h，μ
i，R
di，r)...(1)
In formula,
I: the numbering of the rolling model of regional (i=1,2,3 ...)
P
i: loading prediction value [kN]
Z
pi: load learning coefficient
Km
i: deformation drag predicted value [MPa]
L: contact arc length [mm]
W: the plate wide [mm] of rolling stock
Q
i: roll-force function
H: the entrance side thickness of slab [mm] of milling train
H: the outlet side thickness of slab [mm] of milling train
μ
i: coefficient of friction predicted value
R
di: flat roller radius [mm]
R: rolling rate
μ
i oRI: coefficient of friction model predication value
Z
fi: coefficient of friction learning coefficient.
In addition, in the roll-force function of formula (1), comprise coefficient of friction, between load and coefficient of friction, have non-linear relation.
In the case of with a model to representing that the single technique of nonlinear characteristic shows, conventionally, in the region that shows nonlinear characteristic, precision of prediction can decline.Therefore, to other model of the different area applications of physical characteristic (rolling phenomenon), thereby show single technique with multiple models.Here for convenience of description, to across being described by two rolling models on the represented border of boundary value.
In the present embodiment 1, show as described as follows rolling model.
(model 1)
P
1＝Z
P1·km
1·L
1·W·Q
1(H，h，μ
1，L
1)...(3)
(model 2)
P
2＝Z
P2·km
2·L
2·W·Q
2(H，h，μ
2，R
d2，r)...(5)
In formula,
α
i: model formation connects correction coefficient
Model formation connects correction coefficient alpha
ibe on the border represented by boundary value, remove the jump of loading prediction value,, make the correction coefficient that equates across the loading prediction value of a model and the loading prediction value of another model on border.Originally, on the border being showed by boundary value, the loading prediction value of each load model must be identical, can think that the precision of prediction of coefficient of friction model causes that loading prediction value is different on border.Represent a coefficient of friction model by formula (4), connect correction coefficient with model formation it is revised, thereby loading prediction value is revised.Represent another load model by formula (5), and using this as borderline benchmark.
Although represent that the formula (6) of the coefficient of friction model of the rolling model that becomes benchmark shows identically with formula (4), in the rolling model that becomes benchmark, connects correction coefficient by model formation and is made as " 1 ".Because the rolling model that becomes benchmark can impact the precision of prediction of another rolling model across border, therefore, consider that operational rolling scope predetermines the rolling model that becomes benchmark.Below, centered by the variation of the boundary value as feature of the present invention and the method for solving of model formation connection correction coefficient and using method, describe.
Fig. 1 is the block diagram that represents the structure of the related rolling model optimization device of embodiments of the invention 1.Rolling model optimization device comprises design conditions input unit 1, model optimization device 2 and control device 3.
Design conditions input unit 1 such as, for inputting the rolling condition of the design conditions of lower a kind of rolling stock, thickness of slab etc.The design conditions of inputting from this design conditions input unit 1 are sent to model optimization device 2.Model optimization device 2 is carried out loading prediction according to the design conditions that send from design conditions input unit 1 and is calculated, and result of calculation is sent to control device 3.The details of this model optimization device 2 will be explained hereinafter.Control device 3, according to the result of calculation sending from model optimization device 2, is controlled not shown milling train.
Next, the details of model optimization device 2 is described.Model optimization device 2 comprises evaluation section 11, boundary value form 12, model switching part 13, model formation storage part 14, study calculating part 15, learning coefficient storage part 16, model calculating part 17, model calculated value storage unit 18, model formation fillet dividing value changing unit 21, model formation priority form 22, model formation connects correction coefficient operational part 23 and model formation connects correction coefficient storage part 24.
Evaluation section 11 is according to the design conditions that send from design conditions input unit 1, the priority of the boundary value of the model based on obtaining from boundary value form 12 and the rolling model that obtains from model formation priority form 22, decides and the matched model of rolling condition.Incidental the model being determined by evaluation section 11 numbering is sent to model switching part 13 and is connected correction coefficient operational part 23 with model formation.
The pattern number sending from evaluation section 11 is sent to model formation storage part 14, learning coefficient storage part 16, model formation connection correction coefficient storage part 24 and model calculating part 17 by model switching part 13, thereby the model as handling object is switched.
Model formation storage part 14 stores multiple rolling models each region, that show with model formation.In multiple rolling models, comprise load model, deformation drag and coefficient of friction model etc.In the situation that pattern number is sent from model switching part 13, the rolling model corresponding with this pattern number is sent to model calculating part 17 by model formation storage part 14.
Study calculating part 15 calculates learning coefficient, to make up the error between predicted value and the actual value of model according to the actual conditions of each rolling.The result of calculation being obtained by study calculating part 15 is sent to learning coefficient storage part 16.
The learning coefficient each model of learning coefficient storage part 16 to regional, that send from study calculating part 15 stores.In the situation that pattern number is sent from model switching part 13, the learning coefficient corresponding with this pattern number is sent to model calculating part 17 by learning coefficient storage part 16.
In the situation that receiving pattern number from model switching part 13, the rolling model of model calculating part 17 based on receiving from model formation storage part 14, the learning coefficient receiving from learning coefficient storage part 16 and connect from model formation the model formation that correction coefficient storage part 24 receives and connect correction coefficient, carry out loading prediction and calculate.The result of calculation being obtained by model calculating part 17 is sent to model calculated value storage unit 18 and preserves, and be sent to control device 3.Thus, each actuator of the milling train that be connected, not shown with control device 3 is set to initial value.
Model formation fillet dividing value changing unit 21, according to the operation of not shown input unit, changes the boundary value that is stored in boundary value form 12.In the case of single technique is suitable for multiple models, forethought rolling scope, thus can predetermine the boundary value of the approximate bounds that represents model.But, determine in advance comparatively difficulty of most suitable boundary value, when being set with outward boundary value in the scope of application of a model, near by the represented border of this boundary value, the precision of prediction of model likely can decline.Therefore, needs can be confirmed rolling result (actual value) while change boundary value according to this actual value.Owing to utilizing model formation fillet dividing value changing unit 21, can be on one side according to rolling result confirm multiple models precision of prediction, adjust boundary value on one side, therefore, can set most suitable boundary value, thereby can improve the precision of prediction of model.
In model formation priority form 22, store in advance within the scope of operational rolling the priority as the rolling model of benchmark.The priority of reading from model formation priority form 22 is sent to evaluation section 11.As mentioned above, evaluation section 11 is according to the design conditions that send from design conditions input unit 1, the priority of the boundary value of the model based on obtaining from boundary value form 12 and the rolling model that obtains from model formation priority form 13, decides the model as benchmark.Be the higher model of frequency of utilization dreaming up according to rolling scope as the model of benchmark, owing to it being adjusted and learning, therefore, the precision of other load models that connect on border also can improve.In addition, when existing in the situation of more than three models, utilize evaluation section 11, according to the priority of model and boundary value, determining becomes the rolling model of benchmark, and it is connected with the load model of regional successively.
Model formation connects the pattern number of the model as benchmark of correction coefficient operational part 23 based on receiving from evaluation section 11, the boundary value getting from boundary value form 12, the model formation getting from model formation storage part 14, the learning coefficient of each model getting from learning coefficient storage part 16, the predicted value that the last time getting from model calculated value storage unit 18 calculates, and the model formation that the last time receiving from model formation connection correction coefficient storage part 24 calculates connects correction coefficient, calculate the correction value that new model formation connects correction coefficient and learning coefficient.
With reference to Fig. 2, the details that model formation is connected to the function of correction coefficient operational part 23 describes.Fig. 2 is the figure that represents the summary of the computing of being undertaken by model formation connection correction coefficient operational part 23, and transverse axis is that rolling rate, the longitudinal axis are load.Across the scope of application of can be distinguished one from the other by the represented border of boundary value model 1 and model 2, solid line is the loading prediction value of each model before proofreading and correct.Fig. 2 represents the load model taking model 2 as benchmark.Contact between the boundary value of each model is the A point of model 1 and the B point of model 2.Owing to being model 2 as the load model of benchmark, therefore, for solid line is connected continuously on border, as long as be B point by the contact being connected with the borderline phase of model 1 from A point calibration, connect as shown by dashed lines by this correction.
Connect the operation method of correction coefficient and describe realizing the model formation of correction of above-mentioned model.The calculated value of the coefficient of friction at the A point place of model 1 is made as to μ
1A.Use a model on B point 2 LOAD FOR value, utilizes model 1 coefficient of friction of seizing back confiscated property, and is made as μ
1B.Then, utilize each friction co-efficient value, obtain model formation according to formula (7) and connect correction coefficient.
In formula,
μ
1A: the coefficient of friction at the boundary value place of model 1
μ
1B: the loading prediction value at 2 the boundary value place of using a model, utilize model 1 the obtained coefficient of friction of seizing back confiscated property.
Why revise coefficient of friction model according to formula (7), nothing more than be consider the precision of LOAD FOR value not high be bad caused by the precision of coefficient of friction model.
Then, the operation method of the correction value to learning coefficient describes.B point shown in Fig. 2 is not limited to fixed value conventionally, but changes according to the change of the adjustment of model 2, study or boundary value.Change because the variation of B point can cause model formation to connect correction coefficient, therefore, on utilizing, learning coefficient once carries out computing coefficient of friction, and the calculated value of coefficient of friction can produce error.Therefore the variable quantity that, the correction of the learning coefficient shown in through type (8) connects correction coefficient to model formation compensates.
In formula,
Z
f1 nEW: revised learning coefficient
α
1 oLD: the model formation that the last time calculates connects correction coefficient
α
1 oRIOLD: the predicted value of the coefficient of friction model that the last time calculates
Z
f1 oLD: the learning coefficient before correction
Utilize above function, even in the case of adjustment, study or the boundary value of model 2 change, also can be by the correction of learning coefficient, proofread and correct the error of the loading prediction causing because of the variation of model formation connection correction coefficient.Consequently, can guarantee the precision of prediction of model.
Model formation connects correction coefficient operational part 23 after calculating the correction value of above-mentioned model formation connection correction coefficient and learning coefficient, model formation is connected to correction coefficient and be stored in model formation connection correction coefficient storage part 24, and utilize revised learning coefficient, upgrade being stored in learning coefficient in learning coefficient storage part 16, corresponding.In addition, connect correction coefficient storage part 24 at model formation " 1 " is saved as to the model formation connection correction coefficient as the model of benchmark.In addition, the factor in the easy generation error of utilizing rolling model is carried out layering the learning coefficient of learning coefficient storage part 16, model formation is connected to correction coefficient storage part 24 and also need to carry out the layering identical with learning coefficient with the content of model calculated value storage unit 18.
If from model switching part 13 Acceptance Model numberings, the model formation connection correction coefficient corresponding with this pattern number is sent to model calculating part 17 by model formation connection correction coefficient storage part 24.As mentioned above, in the situation that receiving pattern number from model switching part 13, the model formation of model calculating part 17 based on from model formation storage part 14 rolling model corresponding with pattern number that receive, the learning coefficient receiving from learning coefficient storage part 16 and connect from model formation the model formation that correction coefficient storage part 24 receives and connect correction coefficient, carry out loading prediction and calculate.
As mentioned above, why revise coefficient of friction model, nothing more than be consider the precision of LOAD FOR value not high be to be caused by the predicated error of coefficient of friction model.Therefore, utilize formula (4) to carry out computing coefficient of friction, thereby can improve the precision of prediction of coefficient of friction, can also try hard to improve the precision of other rolling models that use this coefficient of friction.
Rolling model optimization device as discussed above, related according to embodiment 1, owing to changing the boundary value of multiple models, therefore, can be on the boundary value of model without jump connect predicted value.Consequently, owing to setting the region that is suitable for each model, therefore, can improve the precision of prediction of model.
[embodiment 2]
Fig. 3 is the block diagram that represents the structure of the related rolling model optimization device of embodiments of the invention 2.The structure of the related rolling model optimization device of the present embodiment 2 rolling model optimization device related with embodiment 1 is identical, but the related rolling model optimization device of the present embodiment 2 has following structure:, model formation fillet dividing value changing unit 21 is in obtaining boundary value from boundary value form 12, obtain learning coefficient from learning coefficient storage part 16, the rolling model optimization device related from embodiment 1 is different in this.Below, centered by the part different from embodiment 1, describe.
Model formation fillet dividing value changing unit 21 is in obtaining boundary value from boundary value form 12, obtain learning coefficient from learning coefficient storage part 16, the mean value of the learning coefficient in the region predetermining is carried out computing and compared, thereby the boundary value being stored in boundary value form 12 is changed.For example, model optimization device 2 carries out LOAD FOR and study with the adjust the distance region of boundary value-5% of two models conventionally., though beyond the region of applicable models 2, in appointed region, also first model 2 is learnt.
Fig. 4 is the flow chart that represents the step of the calculating of being undertaken by model formation fillet dividing value changing unit 21.First, model formation fillet dividing value changing unit 21 is in the region apart from boundary value-5%, and the mean value of the absolute value to load learning coefficient carries out computing (step S1).Then, whether the mean value of the absolute value of the learning coefficient of model formation fillet dividing value changing unit 21 to model 1 is that the mean value of the absolute value of the learning coefficient of model 2 is investigated below.
In step S2, below the mean value of absolute value that the mean value of the absolute value of the learning coefficient of model 1 is the learning coefficient of model 2 if be judged as, model formation fillet dividing value changing unit 21, in the region of the higher model of the mean value of the absolute value of learning coefficient, changes+1% (step S3) by boundary value.On the other hand, in step S2, if be judged as the mean value that the mean value of the absolute value of the learning coefficient of model 1 is greater than the absolute value of the learning coefficient of model 2, model formation fillet dividing value changing unit 21, in the region of the higher model of the mean value of the absolute value of learning coefficient, changes-1% (step S4) by boundary value.
That is, it is more stable that model formation fillet dividing value changing unit 21 is judged as the side's that mean value is lower model, thereby in the region of the higher model of the mean value of learning coefficient, boundary value is changed to 1%.Thus, owing to conventionally boundary value being adjusted into most suitable boundary value, therefore, can improve the precision of prediction of model.In addition, owing to automatically controlling boundary value, therefore, can improve rapidly the precision of prediction of model, adjust operation thereby can alleviate.
As discussed above, the rolling model optimization device related according to embodiment 2, owing to monitoring near the learning coefficient boundary value of multiple models while change boundary value, therefore, conventionally can the higher model of applied forecasting precision in regional, thus the precision of prediction of model can be improved.
In addition, in the above-described embodiment, represent that with the form of multiplication the model formation of coefficient of friction model connects correction coefficient, but be not limited to this.In addition, represent the learning coefficient of each load model with the form of multiplication, represent the learning coefficient of each coefficient of friction model with the form of addition, but be not limited to this.
In addition, model is made as to load model and coefficient of friction model and is illustrated, but can be also between other different models, such as deformation drag model and coefficient of friction model, torque model and coefficient of friction model etc.
In addition, the present invention can be applicable to all milling train such as hot-rolling mill, cold-rolling mill or tandem mill.And operation is made as rolling process by the present invention, but be not limited to this, as long as the operation of utilizing the models such as heating, refrigerating work procedure to control, just can be suitable for.
Claims (3)
1. a rolling model optimization device, is characterized in that, comprising:
Boundary value form, these boundary value form his-and-hers watches show that the boundary value on the border of multiple models stores;
Model formation fillet dividing value changing unit, this model formation fillet dividing value changing unit, according to the actual value of rolling, changes the boundary value being stored in described boundary value form;
Model formation priority form, this model formation priority form stores the priority of described multiple models;
Evaluation section, this evaluation section is under each inputted design conditions, and the boundary value based on getting from described boundary value form and the priority getting from described model formation priority form, decide the model that is applicable to rolling condition;
Model switching part, this model switching part switches to model on the model being determined by described evaluation section;
Study calculating part, this study calculating part, to each model in described multiple models, calculates learning coefficient according to the difference of calculated predicted value and described actual value;
Model calculating part, the model formation of the model that this model calculating part utilization is switched by described model switching part and the learning coefficient being calculated by described study calculating part, calculate the predicted value of this model; And
Model formation connects correction coefficient operational part, this model formation connects correction coefficient operational part model formation connection correction coefficient is calculated, described model formation connects correction coefficient model is proofreaied and correct, to make each predicted value of two adjacent models continuous on the represented border of the boundary value by obtaining from described boundary value form
The model formation of the model that the utilization of described model calculating part is switched by described model switching part, the learning coefficient being calculated by described study calculating part and the model formation connection correction coefficient calculating with described model formation connection correction coefficient operational part, calculate the predicted value of model.
2. rolling model optimization device as claimed in claim 1, is characterized in that,
In the case of calculated model formation connect correction coefficient with on the model formation that once calculates is connected correction coefficient and compares and occurred variation, described model formation connects correction coefficient operational part to be revised the learning coefficient being calculated by described study calculating part, compensates thereby model formation is connected to correction coefficient.
3. rolling model optimization device as claimed in claim 1, is characterized in that,
Described model formation fillet dividing value changing unit is calculated and compares near the mean value of the learning coefficient boundary value of multiple models, in determined region, thereby the boundary value being stored in described boundary value form is revised.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2011114673A JP5727865B2 (en) | 2011-05-23 | 2011-05-23 | Rolling model optimization device |
| JP2011-114673 | 2011-05-23 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102794310A CN102794310A (en) | 2012-11-28 |
| CN102794310B true CN102794310B (en) | 2014-10-22 |
Family
ID=47193757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110199333.8A Active CN102794310B (en) | 2011-05-23 | 2011-07-04 | Rolling model optimization device |
Country Status (2)
| Country | Link |
|---|---|
| JP (1) | JP5727865B2 (en) |
| CN (1) | CN102794310B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6484265B2 (en) | 2017-02-15 | 2019-03-13 | ファナック株式会社 | Robot system having learning control function and learning control method |
| JP7281958B2 (en) * | 2019-05-08 | 2023-05-26 | 株式会社Uacj | Feature prediction device, manufacturing condition optimization device, control method for feature prediction device, control program |
| KR20210154832A (en) * | 2020-05-13 | 2021-12-21 | 도시바 미쓰비시덴키 산교시스템 가부시키가이샤 | Physical Model Identification System |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1105296A (en) * | 1993-10-05 | 1995-07-19 | 日立有限公司 | Rolling mill control method and apparatus |
| CN101256401A (en) * | 2007-02-27 | 2008-09-03 | 国际商业机器公司 | System and method for determining processing order of steel plate |
| CN101389417A (en) * | 2007-01-22 | 2009-03-18 | 东芝三菱电机产业系统株式会社 | Method of heating control in steel sheet production line and apparatus therefor |
| CN101590489A (en) * | 2008-05-26 | 2009-12-02 | 株式会社日立制作所 | Wide control device of the plate of hot-rolling mill and control method |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS62214813A (en) * | 1986-03-17 | 1987-09-21 | Sumitomo Metal Ind Ltd | Control method for rolling mill |
| JPH09155420A (en) * | 1995-12-07 | 1997-06-17 | Kawasaki Steel Corp | Method for learning setup model of rolling mill |
| JP2006122980A (en) * | 2004-10-29 | 2006-05-18 | Jfe Steel Kk | Method for predicting deformation resistance of rolling material and frictional coefficient between roll and rolling material in tandem cold rolling mill, and tandem cold rolling method |
| JP4286211B2 (en) * | 2004-11-25 | 2009-06-24 | 株式会社神戸製鋼所 | Sheet width control method for continuous rolling mill |
-
2011
- 2011-05-23 JP JP2011114673A patent/JP5727865B2/en active Active
- 2011-07-04 CN CN201110199333.8A patent/CN102794310B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1105296A (en) * | 1993-10-05 | 1995-07-19 | 日立有限公司 | Rolling mill control method and apparatus |
| CN101389417A (en) * | 2007-01-22 | 2009-03-18 | 东芝三菱电机产业系统株式会社 | Method of heating control in steel sheet production line and apparatus therefor |
| CN101256401A (en) * | 2007-02-27 | 2008-09-03 | 国际商业机器公司 | System and method for determining processing order of steel plate |
| CN101590489A (en) * | 2008-05-26 | 2009-12-02 | 株式会社日立制作所 | Wide control device of the plate of hot-rolling mill and control method |
Non-Patent Citations (2)
| Title |
|---|
| JP特开2006-150371A 2006.06.15 |
| JP特开平9-155420A 1997.06.17 |
Also Published As
| Publication number | Publication date |
|---|---|
| JP5727865B2 (en) | 2015-06-03 |
| CN102794310A (en) | 2012-11-28 |
| JP2012240103A (en) | 2012-12-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101522325B (en) | Method of monitoring the physical state of a hot-rolled sheet or hot-rolled strip while controlling a plate rolling train for working a hot-rolled sheet or hot-rolled strip | |
| KR100237506B1 (en) | Strip crown measuring method and control method for continuous rolling machines | |
| KR101889668B1 (en) | Rolling simulation device | |
| Bemporad et al. | Optimization-based automatic flatness control in cold tandem rolling | |
| JP5759206B2 (en) | Learning coefficient controller | |
| KR101617375B1 (en) | Feedforward thickness control method for performance of cold rolling mill | |
| US9180505B2 (en) | Water injection controller, water injection control method, and water injection control program for rolling lines | |
| CN102794310B (en) | Rolling model optimization device | |
| CN103108706B (en) | Real-time determination method for temperature and geometry of a hot metal hot in a finishing train | |
| CN111352403A (en) | Process design method for rolling plates | |
| US8935945B2 (en) | Control system | |
| JP6641849B2 (en) | Power prediction method | |
| KR101749018B1 (en) | Flatness control device | |
| JP2018010521A (en) | Product state prediction device, product state control device, product state prediction method and program | |
| JP2009072807A (en) | Method and device for controlling planar shape of rolled stock and method of manufacturing thick steel plate | |
| JP4983589B2 (en) | Control device for cold continuous rolling equipment | |
| KR102486545B1 (en) | Method for predicting demand of requirement man power and material according to new ship contract | |
| JP2010115692A (en) | Device and method for predicting quality in manufacturing process, program and computer-readable recording medium | |
| CN111132773B (en) | Flatness control using optimizer | |
| JP6652095B2 (en) | Method of rolling steel sheet and method of manufacturing steel sheet | |
| JPH0760320A (en) | Method for rolling thick plate | |
| JP6031344B2 (en) | Rolling control device, rolling control method, and rolling control program | |
| JP2006277093A (en) | Energy demand variation risk prediction system | |
| JP3354053B2 (en) | Feed forward control system | |
| JP2021010914A (en) | Rolled-shape control device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |