JPH10263640A - Method for learning-controlling rolling load in rolling mill - Google Patents

Method for learning-controlling rolling load in rolling mill

Info

Publication number
JPH10263640A
JPH10263640A JP9071404A JP7140497A JPH10263640A JP H10263640 A JPH10263640 A JP H10263640A JP 9071404 A JP9071404 A JP 9071404A JP 7140497 A JP7140497 A JP 7140497A JP H10263640 A JPH10263640 A JP H10263640A
Authority
JP
Japan
Prior art keywords
rolling
learning
rolling load
value
load
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.)
Granted
Application number
JP9071404A
Other languages
Japanese (ja)
Other versions
JP3467677B2 (en
Inventor
Hiroshi Kurakake
浩 鞍掛
Hisafumi Tsuchida
尚史 土田
Shunji Goto
俊二 後藤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
JFE Steel Corp
Original Assignee
Kawasaki Steel Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Kawasaki Steel Corp filed Critical Kawasaki Steel Corp
Priority to JP07140497A priority Critical patent/JP3467677B2/en
Publication of JPH10263640A publication Critical patent/JPH10263640A/en
Application granted granted Critical
Publication of JP3467677B2 publication Critical patent/JP3467677B2/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Landscapes

  • Control Of Metal Rolling (AREA)

Abstract

PROBLEM TO BE SOLVED: To obtain a rolling load learning-control technique by separating a learning factor decided with a rolling load actual value and a rolling load model calculated value into the components learning an intrinsic error of the rolling material and an error with change in the lapse of time of a rolling mill, reflecting the results individu ally learning both components on the learning factor and predicting the rolling load in the next time. SOLUTION: In order to learn by separating the error of the rolling load model into the intrinsic component of the rolling material corresponding to deforming resistance and the changing component in the lapse of time corresponding to friction coefficient at the rolling mill side, two of the first and the second learning factors are used. The first learning factor Zpk uses the different value in prefixed each kind group of steel because the deforming resistance is different in the kind of steel, and uses the different value in each stand, too. The second learning factor Zpm uses the different value in each rolling pattern dividing rolling material into the thick product and the thin product, and in each stand because the friction coefficient is affected by the roll characteristic and the rolling oil characteristic. The setting of the rolling load P is predicted with P=Zpk .Zpm .Po in the case of using Po for rolling load model calculating value.

Description

【発明の詳細な説明】DETAILED DESCRIPTION OF THE INVENTION

【0001】[0001]

【発明の属する技術分野】本発明は、特に冷間圧延機に
圧延荷重をセットアップする際に適用して好適な、圧延
機における圧延荷重の学習制御方法に関する。
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rolling load learning control method for a rolling mill, which is particularly suitable for setting up a rolling load in a cold rolling mill.

【0002】[0002]

【従来の技術】従来より、冷間圧延機の圧延制御におい
ては、圧延数式モデルを用いて次回の圧延材に対する圧
延荷重を予測してその設定を行っており、その際の設定
精度を向上するためには圧延荷重をより正確に予測する
ことが不可欠である。
2. Description of the Related Art Conventionally, in rolling control of a cold rolling mill, a rolling load on a next rolled material is predicted and set by using a rolling mathematical model, and the setting accuracy at that time is improved. Therefore, it is essential to more accurately predict the rolling load.

【0003】一般に、上記のような数式モデル(圧延荷
重モデル)による予測精度を向上させるためには、直近
の実績値を用いて該モデルを補正する学習を行う方法が
採られている。このような圧延荷重モデルの精度向上の
ために行われる最も一般的な学習方法としては、例えば
特開昭50−108150に開示されているように、圧
延荷重実績値と、圧延実績を用いて算出した圧延荷重モ
デル計算値との比(又は差)を取り、これを次の(1)
式の予測式で次回の圧延荷重を予測計算するときの補正
係数(学習係数)Zp とする方法が知られている。な
お、学習係数Zpとしては、実績荷重とモデル計算値の
“差”を用いることもできるが、以下では“比”を用い
る場合を中心に説明する。
[0003] In general, in order to improve the prediction accuracy by the above mathematical model (rolling load model), a method of performing learning to correct the model using the latest actual value has been adopted. As the most common learning method performed to improve the accuracy of such a rolling load model, for example, as disclosed in Japanese Patent Application Laid-Open No. 50-108150, a calculation using a rolling load actual value and a rolling actual result is performed. The ratio (or difference) with the calculated value of the rolling load model is calculated, and is calculated by the following (1).
There is known a method of using a correction coefficient (learning coefficient) Zp for predicting and calculating the next rolling load using a prediction equation. As the learning coefficient Zp, the "difference" between the actual load and the model calculation value can be used, but the following description will focus on the use of the "ratio".

【0004】P=Zp ・P0 …(1) ここで、P:圧延荷重予測値 Zp :荷重比学習係数 P0 :圧延荷重モデル計算値P = Zp · P0 (1) where P: predicted rolling load Zp: load ratio learning coefficient P0: calculated rolling load model

【0005】但し、この(1)式で次回の圧延荷重を予
測計算する場合は、通常、この荷重比学習係数Zp とし
ては、次の(2)式のように、前記比の今回の値と、前
回の荷重比学習係数を用いて平滑化処理した値を使用し
ている。
However, when predicting and calculating the next rolling load by the equation (1), usually, the load ratio learning coefficient Zp is calculated by subtracting the current value of the ratio from the present value of the ratio as shown in the following equation (2). , A value obtained by performing a smoothing process using the previous load ratio learning coefficient.

【0006】 Zp =(1−α)Zp-1 +α・Pact /Pcal …(2) ここで、Zp-1 :荷重比学習係数の前回値 Pact :荷重実績値 Pcal :圧延荷重モデル計算値 α:0<α<1の定数Zp = (1−α) Zp−1 + α · Pact / Pcal (2) where Zp−1: the previous value of the load ratio learning coefficient Pact: the actual load value Pcal: the calculated value of the rolling load model α: 0 <α <1 constant

【0007】前記(1)式を用いて次回の圧延荷重を予
測計算する際、その予測誤差を生じさせる要因として
は、圧延材固有の誤差と圧延機固有の誤差の2つの要因
に分けることができる。
[0007] When predicting and calculating the next rolling load using the above equation (1), factors that cause the prediction error can be divided into two factors: an error specific to the rolled material and an error specific to the rolling mill. it can.

【0008】前者の圧延材固有の誤差は、圧延材の変形
抵抗の予測誤差に現われるものであり、圧延材の成分や
熱間圧延工程における製造条件等により、鋼種毎に(同
一の鋼種の中でもある程度は)ランダムな誤差をもって
いると考えられる。
The former error inherent in the rolled material appears in the prediction error of the deformation resistance of the rolled material, and depends on the type of the rolled material, the manufacturing conditions in the hot rolling process, etc., for each steel type (of the same steel type). It is thought to have (to some extent) random errors.

【0009】一方、後者の圧延機固有の誤差は、摩擦係
数の予測誤差に現われるものであり、ロール条件や圧延
油条件により、経時的にある傾向をもって変化すると考
えられる。
On the other hand, the latter error peculiar to the rolling mill appears in the prediction error of the coefficient of friction, and is considered to change with a certain tendency over time depending on the rolling conditions and rolling oil conditions.

【0010】[0010]

【発明が解決しようとする課題】しかしながら、従来の
圧延荷重モデルの学習方法では、上記のような2つの要
因によると考えられる誤差成分を、前記(1)式に示し
たように、1つの補正係数(学習係数)Zp のみで学習
していたため、以下のような問題があった。
However, in the conventional rolling load model learning method, the error component considered to be due to the above two factors is corrected by one correction as shown in the above equation (1). Since the learning is performed only by the coefficient (learning coefficient) Zp, the following problem occurs.

【0011】例えば、補正係数を圧延材の鋼種毎の学習
区分に分けていたとしても、圧延機の経時的な変化に起
因する誤差を学習できない。又、上記補正係数の学習区
分として、圧延が進むに連れて生じるロール表面粗度の
低下や、使用する圧延油の濃度変化等に起因すると考え
られる圧延機側の経時変化を学習することができる区分
を持っていたとしても、圧延材の鋼種の変化に対して
は、それまで使用してきた補正係数をそのまま使用する
ことはできない。
For example, even if the correction coefficient is divided into learning sections for each type of steel of a rolled material, it is not possible to learn an error caused by a change over time in a rolling mill. Further, as the learning section of the correction coefficient, it is possible to learn a change with time on the rolling mill side, which is considered to be caused by a decrease in roll surface roughness caused as the rolling progresses or a change in the concentration of the used rolling oil. Even if there is a classification, the correction coefficient used so far cannot be used as it is for a change in the steel type of the rolled material.

【0012】本発明は、前記従来の問題点を解決するべ
くなされたもので、次回圧延時に設定する圧延荷重を、
圧延実績を用いて正確に予測計算できるようにすること
により、最初から高精度に圧延制御することができる圧
延荷重の学習制御に関する技術を提供することを課題と
する。
The present invention has been made to solve the above-mentioned conventional problems.
An object of the present invention is to provide a technique related to learning control of a rolling load capable of performing rolling control with high precision from the beginning by enabling accurate prediction calculation using rolling results.

【0013】[0013]

【課題を解決するための手段】本発明は、圧延荷重モデ
ルに次回圧延条件を代入して求めた圧延荷重モデル計算
値に学習係数を適用して得られた圧延荷重予測値を、次
回圧延時に設定して圧延制御を行う圧延機における圧延
荷重の学習制御方法において、圧延荷重実績値と、前記
圧延荷重モデルに圧延条件実績値を代入して求めた圧延
荷重モデル計算値とに基づいて決定される学習係数を、
圧延材固有の誤差を学習する成分と、圧延機の経時変化
による誤差を学習する成分とに分離すると共に、これら
両成分をそれぞれ個別に学習させ、その学習結果を前記
学習係数に反映させ、次回圧延時に設定する圧延荷重を
予測することにより、前記課題を解決したものである。
According to the present invention, a rolling load prediction value obtained by applying a learning coefficient to a rolling load model calculation value obtained by substituting the next rolling condition into a rolling load model is used in the next rolling. In the rolling load learning control method in a rolling mill that performs rolling control by setting, the rolling load actual value is determined based on a rolling load model calculated value obtained by substituting the rolling condition actual value into the rolling load model. Learning coefficient
The component for learning the error inherent to the rolled material and the component for learning the error due to the aging of the rolling mill are separated, and both these components are individually learned, and the learning result is reflected in the learning coefficient. This problem has been solved by predicting a rolling load to be set at the time of rolling.

【0014】即ち、本発明においては、圧延荷重モデル
計算値に適用する学習係数を、圧延材固有の誤差を学習
する成分と、圧延機の経時的変化による誤差を学習する
成分とに分離して、それぞれ個別に学習し、両成分の学
習結果を次回圧延に使用する学習係数に、例えば両者の
乗算値として反映させるようにしたので、圧延荷重予測
値の精度向上を図ることが可能となった。
That is, in the present invention, the learning coefficient applied to the calculated value of the rolling load model is divided into a component for learning an error specific to a rolled material and a component for learning an error due to a change over time of a rolling mill. Since the learning results of both components are individually learned, and the learning coefficients used for the next rolling are reflected as, for example, a multiplication value of the two components, the accuracy of the rolling load prediction value can be improved. .

【0015】[0015]

【発明の実施の形態】以下、図面を参照して、本発明の
実施の形態について詳細に説明する。
Embodiments of the present invention will be described below in detail with reference to the drawings.

【0016】まず、本発明に係る一実施形態である圧延
荷重の学習制御方法の原理について説明する。
First, the principle of a rolling load learning control method according to an embodiment of the present invention will be described.

【0017】本実施形態においては、圧延荷重モデルの
誤差を、圧延材固有の成分(変形抵抗相当分)と、圧延
機側の経時的変化成分(摩擦係数相当分)に分離し、以
下の方法により学習を行う。
In this embodiment, the error of the rolling load model is separated into a component unique to the rolled material (corresponding to deformation resistance) and a time-varying component on the rolling mill side (corresponding to friction coefficient). Learning is performed by

【0018】本実施形態では、上記のように誤差成分を
2つに分けて学習するため、圧延荷重モデルの学習係数
(荷重比学習係数)として、それぞれの成分に対応する
次の第1、第2の2つの係数を使用する。
In the present embodiment, since the error component is divided into two and learned as described above, the learning coefficients (load ratio learning coefficients) of the rolling load model are set as the following first and second coefficients corresponding to the respective components. Two coefficients of 2 are used.

【0019】 Zpk:第1学習係数(変形抵抗相当分の荷重比に当る) Zpm:第2学習係数(摩擦係数相当分の荷重比に当る)Zpk: a first learning coefficient (corresponding to a load ratio corresponding to deformation resistance) Zpm: a second learning coefficient (corresponding to a load ratio corresponding to a friction coefficient)

【0020】上記第1学習係数Zpkは、変形抵抗が圧延
材の鋼種(成分)により異なるため、圧延材について予
め適切に決めてある鋼種グループ毎に異なる値を使用す
る。又、このZpkは、スタンド毎でも異なる値を使用
し、各スタンドでのモデルの誤差を吸収するようにして
いる。
Since the first learning coefficient Zpk has a different deformation resistance depending on the steel type (component) of the rolled material, a different value is used for each steel type group which is appropriately determined in advance for the rolled material. In addition, this Zpk uses a different value for each stand so as to absorb an error of the model in each stand.

【0021】又、上記第2学習係数Zpmは、摩擦係数が
ロール性状や圧延油性状に大きく影響を受けるため、圧
延材の厚さの違いにより厚物グループや薄物グループ等
のように分類した圧延パターン毎に、且つ、スタンド毎
にそれぞれ異なる値を使用し、圧延材の鋼種に依存する
ことなく学習を行う。
In addition, since the friction coefficient is greatly affected by the roll properties and rolling oil properties, the second learning coefficient Zpm is classified into a thick group or a thin group according to the difference in the thickness of the rolled material. Using different values for each pattern and for each stand, learning is performed without depending on the steel type of the rolled material.

【0022】本実施形態では、次回圧延時に設定する圧
延荷重Pは、上記第1、第2の学習係数Zpk、Zpmを、
圧延荷重モデル計算値P0 に乗じた次の(3)式で予測
する。
In the present embodiment, the rolling load P set at the time of the next rolling is determined by calculating the first and second learning coefficients Zpk and Zpm as follows:
Predicted by the following equation (3) multiplied by the calculated rolling load model value P0.

【0023】P=Zpk・Zpm・P0 …(3)P = Zpk.Zpm.P0 (3)

【0024】この(3)式に適用する第1、第2学習係
数Zpk、Zpmについて、圧延実績に基づいて次の(4)
式、(5)式の順に計算する学習を行い、その後両者を
圧延荷重モデル計算値に掛け合わせることにより、上記
(3)式により次回圧延荷重を予測する。
With respect to the first and second learning coefficients Zpk and Zpm applied to the equation (3), the following (4)
Learning to calculate in the order of the equation and the equation (5) is performed, and then the two are multiplied by the calculated value of the rolling load model, thereby predicting the next rolling load by the above equation (3).

【0025】 Zpk=(1−αk )Zpk-1+αk ・Pact /(Zpm-1・Pcal ) …(4) Zpm=(1−αm )Zpm-1+αm ・Pact /(Zpk・Pcal ) …(5) ここで、Zpk-1:第1学習係数の前回値 Zpm-1:第2学習係数の前回値 αk 、αm :0<αk ,αm <1の定数Zpk = (1−αk) Zpk−1 + αk · Pact / (Zpm−1 · Pcal) (4) Zpm = (1−αm) Zpm−1 + αm · Pact / (Zpk · Pcal) (5) And Zpk-1: the previous value of the first learning coefficient Zpm-1: the previous value of the second learning coefficient αk, αm: 0 <αk, a constant of αm <1

【0026】上記(4)、(5)式による学習の考え方
は、圧延機の経時変化の傾きは小さいとして、最初に鋼
種グループ毎に分けた変形抵抗相当分の荷重誤差に当る
第1学習係数Zpkを、第1、第2学習係数として、いず
れも前回値を用いる上記(4)式により学習する。
The concept of learning based on the above equations (4) and (5) is that the first learning coefficient corresponding to the load error corresponding to the deformation resistance first divided for each steel type group is considered assuming that the gradient of the change with time of the rolling mill is small. Zpk is used as the first and second learning coefficients, and learning is performed by the above equation (4) using both previous values.

【0027】一方、残った誤差を、圧延機の経時変化に
より現われた荷重誤差と考え、これを第2学習係数Zpm
に学習させるため、これを鋼種グループに関係なく圧延
順で、第1学習係数として上記(4)式で算出した今回
値を含む上記(5)式により学習を行う。
On the other hand, the remaining error is considered as a load error caused by a change over time of the rolling mill, and is considered as a second learning coefficient Zpm.
In this case, the learning is performed by the above-described equation (5) including the current value calculated by the above-described equation (4) as the first learning coefficient in the rolling order regardless of the steel type group.

【0028】なお、第1学習係数Zpkと第2学習係数Z
pmの学習順序は、次の(6)式、(7)式に示すよう
に、逆にしてもよい。
The first learning coefficient Zpk and the second learning coefficient Zpk
The learning order of pm may be reversed as shown in the following equations (6) and (7).

【0029】 Zpm=(1−αm )Zpm-1+αm ・Pact /(Zpk-1・Pcal ) …(6) Zpk=(1−αk )Zpk-1+αk ・Pact /(Zpm・Pcal ) …(7)Zpm = (1−αm) Zpm−1 + αm · Pact / (Zpk−1 · Pcal) (6) Zpk = (1−αk) Zpk−1 + αk · Pact / (Zpm · Pcal) (7)

【0030】前記(4)式、(5)式による学習と、こ
の(6)式、(7)式による学習方法のいずれを選択す
るかは、平滑化係数αk 、αm の関係で決めればよい。
The selection between the learning based on the equations (4) and (5) and the learning method based on the equations (6) and (7) may be determined by the relationship between the smoothing coefficients αk and αm. .

【0031】以上詳述した本実施形態の学習方法によれ
ば、圧延材に起因する荷重誤差成分と、圧延機の経時変
化による荷重誤差成分を別々の学習係数で学習させるこ
とができるため、より高精度な圧延荷重の予測計算が可
能となった。
According to the learning method of the present embodiment described in detail above, the load error component caused by the rolled material and the load error component caused by the aging of the rolling mill can be learned with different learning coefficients. High-precision rolling load prediction calculation became possible.

【0032】次に、本実施形態の具体例である実施例に
ついて説明する。
Next, an example which is a specific example of the present embodiment will be described.

【0033】[0033]

【実施例】図1は、本実施形態の学習制御方法を実機に
適用した場合の圧延荷重の予測計算の手順を示すフロー
チャートである。但し、ここでは、前記(6)式、
(7)式による学習方法を採用している。
FIG. 1 is a flow chart showing a procedure for predicting and calculating a rolling load when the learning control method of the present embodiment is applied to an actual machine. However, here, the equation (6) is used.
The learning method using the equation (7) is adopted.

【0034】まず、ステップ1で、図示しない制御装置
に、圧延荷重、入側板厚、出側板厚、入側張力、出側張
力等の圧延実績を取り込み、ステップ2で、これらの圧
延実績を圧延荷重モデルに代入して圧延荷重モデル計算
値Pcal を算出する。ここで使用する圧延荷重モデルと
しては、例えば公知のHill の式や、Brond&Fordの
式等の任意のモデル式を利用することができる。
First, in step 1, rolling results such as rolling load, incoming plate thickness, outgoing plate thickness, incoming tension, and outgoing tension are loaded into a controller (not shown). In step 2, these rolling results are rolled. The rolling model calculation value Pcal is calculated by substituting into the load model. As the rolling load model used here, for example, a known Hill equation or an arbitrary model equation such as a Brond & Ford equation can be used.

【0035】次いで、ステップ3で、前記(6)式によ
り、第2学習係数Zpmを学習し、この学習値を用いて、
前記(7)式により第1学習係数Zpkを学習して決定す
ると共に、ステップ4で、これら学習値をそれぞれの格
納テーブルに書き込む。
Next, in step 3, the second learning coefficient Zpm is learned from the above equation (6), and this learning value is used to calculate the second learning coefficient Zpm.
The first learning coefficient Zpk is learned and determined according to the above equation (7), and in step 4, these learning values are written in respective storage tables.

【0036】次いで、前記ステップ2で使用したものと
同一の圧延荷重モデル式に、次回圧延時に設定する、前
記ステップ1で示した圧延実績と同種の圧延条件を代入
して、圧延荷重モデル計算値P0 を算出すると共に、学
習した上記第1、第2の各学習係数を適用した前記
(3)式に、該計算値P0 を代入することにより、次回
の圧延荷重を予測した。
Next, the same rolling conditions as those used in step 2 and the same rolling results as those set in step 1 and set in the next rolling are substituted into the same rolling load model formula, and the calculated rolling load model values are calculated. The next rolling load was predicted by calculating P0 and substituting the calculated value P0 into the above equation (3) to which the learned first and second learning coefficients were applied.

【0037】図2は、上述した本実施形態の学習制御方
法を、実機に適用した場合の圧延荷重予測値(計算値)
と、圧延荷重実績値の経時的な推移を示すと共に、併せ
て、学習区分は鋼種グループ毎に設定した従来法で学習
を行った場合の荷重予測値を示したものである。
FIG. 2 shows a predicted rolling load (calculated value) when the learning control method of this embodiment described above is applied to an actual machine.
And the rolling load actual value over time, and the learning category shows the predicted load value when learning is performed by the conventional method set for each steel type group.

【0038】この図2より、本実施形態の学習制御方法
では、特にロール交換直後のロール性状が経時的に大き
く変化する場合でも、その変化を従来法の鋼種グループ
毎の学習係数から分離して学習させるようにしたため、
良好な荷重予測結果が得られていることが分かる。
FIG. 2 shows that the learning control method of the present embodiment separates the change from the learning coefficient for each steel type group of the conventional method, even if the roll property immediately after the roll change greatly changes over time. Because we let you learn,
It can be seen that good load prediction results have been obtained.

【0039】以上詳述した如く、本実施形態によれば、
冷間圧延機のセットアップにおける圧延荷重予測精度を
向上できるため、次回圧延材に対する設定替えの精度を
向上することができる。その結果、任意の圧延材に対し
てコイル先端部における板厚のオフゲージ長さを削減で
きる上に、目標とした圧延パススケジュールの設定誤差
に起因する圧延不安定現象(ロールと圧延材の焼き付き
現象や、スリップ等)の発生を防止できる。
As described in detail above, according to the present embodiment,
Since the rolling load prediction accuracy in setting up the cold rolling mill can be improved, the accuracy of setting change for the next rolled material can be improved. As a result, the off-gauge length of the thickness of the coil at the tip of the coil can be reduced for any rolled material, and the rolling instability phenomenon (seizure of roll and rolled material) caused by the setting error of the target rolling pass schedule Or slips) can be prevented.

【0040】又、設定する圧延荷重の予測精度が向上す
ることから、圧延パワーの予測精度も向上するため、該
パワーの予測外れに起因するモータの過負荷状態の発生
を防止できるという効果もある。
Further, since the prediction accuracy of the set rolling load is improved, the prediction accuracy of the rolling power is also improved, so that an effect of preventing the motor from being overloaded due to the power prediction error can be prevented. .

【0041】以上、本発明について具体的に説明した
が、本発明は、前記実施形態に示したものに限られるも
のでなく、その要旨を逸脱しない範囲で種々変更可能で
ある。
Although the present invention has been specifically described above, the present invention is not limited to the above-described embodiment, and can be variously modified without departing from the gist thereof.

【0042】例えば、前記実施形態では学習係数として
荷重“比”を用いる場合を中心に説明したが、荷重
“差”を用いるようにしてもよい。
For example, in the above embodiment, the case where the load “ratio” is used as the learning coefficient has been mainly described, but the load “difference” may be used.

【0043】[0043]

【発明の効果】以上説明したとおり、本発明によれば、
次回圧延時に設定する圧延荷重を、圧延実績を用いて正
確に予測計算することができることから、最初から高精
度に圧延制御することができる。
As described above, according to the present invention,
Since the rolling load to be set at the next rolling can be accurately predicted and calculated using the rolling results, the rolling control can be performed with high accuracy from the beginning.

【図面の簡単な説明】[Brief description of the drawings]

【図1】本発明に係る一実施形態による学習の手順を示
すフローチャート
FIG. 1 is a flowchart showing a learning procedure according to an embodiment of the present invention.

【図2】発明の効果を示す線図FIG. 2 is a diagram showing the effect of the present invention.

Claims (2)

【特許請求の範囲】[Claims] 【請求項1】圧延荷重モデルに次回圧延条件を代入して
求めた圧延荷重モデル計算値に学習係数を適用して得ら
れた圧延荷重予測値を、次回圧延時に設定して圧延制御
を行う圧延機における圧延荷重の学習制御方法におい
て、 圧延荷重実績値と、前記圧延荷重モデルに圧延条件実績
値を代入して求めた圧延荷重モデル計算値とに基づいて
決定される学習係数を、圧延材固有の誤差を学習する成
分と、圧延機の経時変化による誤差を学習する成分とに
分離すると共に、 これら両成分をそれぞれ個別に学習させ、その学習結果
を前記学習係数に反映させ、次回圧延時に設定する圧延
荷重を予測することを特徴とする圧延機における圧延荷
重の学習制御方法。
1. A rolling control in which a rolling load prediction value obtained by applying a learning coefficient to a rolling load model calculated value obtained by substituting the next rolling condition into a rolling load model is set at the next rolling to perform rolling control. A learning coefficient determined based on a rolling load actual value and a calculated value of a rolling load model obtained by substituting the actual rolling condition value into the rolling load model. And a component that learns the error due to the aging of the rolling mill, and learns both of these components individually, reflects the learning result in the learning coefficient, and sets it at the next rolling. A learning control method of a rolling load in a rolling mill, which predicts a rolling load to be performed.
【請求項2】請求項1において、 圧延材固有の誤差を学習する成分の第1学習係数:Zpk
と、圧延機の経時変化による誤差を学習する成分の第2
学習係数:Zpmとを、それぞれ Zpk=(1−αk )Zpk-1+αk ・Pact /(Zpm-1・
Pcal ) Zpm=(1−αm )Zpm-1+αm ・Pact /(Zpk・P
cal ) により学習し、次回圧延荷重Pを P=Zpk・Zpm・P0 (Zpk-1:第1学習係数の前回値 Zpm-1:第2学習係数の前回値 Pact :荷重実績値 Pcal :圧延荷重モデル計算値 αk 、αm :0<αk ,αm <1の定数 P0 :圧延荷重モデル計算値) により予測することを特徴とする圧延機における圧延荷
重の学習制御方法。
2. The method according to claim 1, wherein a first learning coefficient of a component for learning an error specific to the rolled material is Zpk.
And the second of the components for learning the error due to the aging of the rolling mill
Learning coefficient: Zpm, Zpk = (1−αk) Zpk−1 + αk · Pact / (Zpm−1 ·
Pcal) Zpm = (1-αm) Zpm-1 + αm · Pact / (Zpk · P
cal)) and the next rolling load P is calculated as P = Zpk · Zpm · P0 (Zpk-1: previous value of the first learning coefficient Zpm-1: previous value of the second learning coefficient Pact: actual load value Pcal: rolling load A learning control method for a rolling load in a rolling mill, wherein the prediction is performed based on model calculation values αk, αm: 0 <αk, a constant P0 of αm <1 (rolling load model calculation value).
JP07140497A 1997-03-25 1997-03-25 Learning control method of rolling load in rolling mill Expired - Fee Related JP3467677B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP07140497A JP3467677B2 (en) 1997-03-25 1997-03-25 Learning control method of rolling load in rolling mill

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP07140497A JP3467677B2 (en) 1997-03-25 1997-03-25 Learning control method of rolling load in rolling mill

Publications (2)

Publication Number Publication Date
JPH10263640A true JPH10263640A (en) 1998-10-06
JP3467677B2 JP3467677B2 (en) 2003-11-17

Family

ID=13459553

Family Applications (1)

Application Number Title Priority Date Filing Date
JP07140497A Expired - Fee Related JP3467677B2 (en) 1997-03-25 1997-03-25 Learning control method of rolling load in rolling mill

Country Status (1)

Country Link
JP (1) JP3467677B2 (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001025807A (en) * 1999-07-15 2001-01-30 Toshiba Corp Learning control device of reverse rolling mill
JP2003340508A (en) * 2002-05-27 2003-12-02 Toshiba Ge Automation Systems Corp Learning control apparatus for device of calculating setting of rolling mill
JP2005202803A (en) * 2004-01-16 2005-07-28 Sumitomo Metal Ind Ltd Learning control method
WO2009113719A1 (en) 2008-03-14 2009-09-17 新日本製鐵株式会社 Rolling load prediction learning method for hot plate rolling
EP3006126A1 (en) 2014-10-07 2016-04-13 Hitachi, Ltd. Control device of tandem rolling mill and control method
KR20200032821A (en) * 2018-09-19 2020-03-27 주식회사 포스코 Rolling apparatus and method using predictive model
JP2021133415A (en) * 2020-02-28 2021-09-13 Jfeスチール株式会社 Model learning method, flying plate thickness changing method, steel plate manufacturing method, model learning device, flying plate thickness changing device and steel plate manufacturing device

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001025807A (en) * 1999-07-15 2001-01-30 Toshiba Corp Learning control device of reverse rolling mill
JP2003340508A (en) * 2002-05-27 2003-12-02 Toshiba Ge Automation Systems Corp Learning control apparatus for device of calculating setting of rolling mill
JP2005202803A (en) * 2004-01-16 2005-07-28 Sumitomo Metal Ind Ltd Learning control method
JP4543684B2 (en) * 2004-01-16 2010-09-15 住友金属工業株式会社 Learning control method
WO2009113719A1 (en) 2008-03-14 2009-09-17 新日本製鐵株式会社 Rolling load prediction learning method for hot plate rolling
US8185232B2 (en) 2008-03-14 2012-05-22 Nippon Steel Corporation Learning method of rolling load prediction for hot rolling
EP3006126A1 (en) 2014-10-07 2016-04-13 Hitachi, Ltd. Control device of tandem rolling mill and control method
KR20200032821A (en) * 2018-09-19 2020-03-27 주식회사 포스코 Rolling apparatus and method using predictive model
JP2021133415A (en) * 2020-02-28 2021-09-13 Jfeスチール株式会社 Model learning method, flying plate thickness changing method, steel plate manufacturing method, model learning device, flying plate thickness changing device and steel plate manufacturing device

Also Published As

Publication number Publication date
JP3467677B2 (en) 2003-11-17

Similar Documents

Publication Publication Date Title
AU603309B1 (en) Wet skin-pass rolling method
JPH10263640A (en) Method for learning-controlling rolling load in rolling mill
JPH11707A (en) Method for controlling rolling mill and controller for rolling mill
US5873277A (en) Control process for a roll stand for rolling a strip
JPS59101212A (en) Controlling method of rolling mill
JP3742966B2 (en) Control device
JP3300208B2 (en) Learning control method in process line
JP2005111546A (en) Method for controlling plate thickness and crown
JP2696058B2 (en) Control method for constant length section of variable thickness part
JPH06259107A (en) Learning control method for process line
JPS61269912A (en) Method for setting roll gap in cold rolling of steel sheet
JPH10180321A (en) Learning control method in rolling mill
KR100815690B1 (en) Method for Predicting Roll Force in Hot Rolling
JP2000126809A (en) Set up device of rolling mill
JPH10249422A (en) Method for automatic control of plate thickness in rolling
JP3498126B2 (en) Rolling mill dynamic setup control apparatus and method
JPS61269913A (en) Method for controlling learning of roll gap in cold rolling
JPH09155420A (en) Method for learning setup model of rolling mill
JPH08332506A (en) Method for controlling thickness of taper plate
JPS59183918A (en) Gage meter type automatic controlling method of plate thickness
JPS60162514A (en) Method for learning control of rolling load in cold tandem rolling
JPS61189810A (en) Shape-controlling method in finish rolling
JP2878060B2 (en) Automatic hot thickness control method and apparatus
JPS58212806A (en) Method for controlling thickness with rolling mill
JPH0472602B2 (en)

Legal Events

Date Code Title Description
FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20080905

Year of fee payment: 5

LAPS Cancellation because of no payment of annual fees