JP4158254B2 - Raw tube rolling method by mandrel mill and mandrel mill - Google Patents

Description

【００２０】
前記寸法偏差には、経時的な変化及び他の外乱が含まれるので、それらを考慮して、（１）式を、１次遅れのオフセットγを有する次の（２）式に書き換える。
【００２１】
【数２】

【００２２】
ｎ回学習後のロールギャップ誤差学習量Ｚ_C （ｉ）、及びｎ回学習後のマンドレルバー寸法誤差学習量Ｚ_Mj（ｉ）（マンドレルバー誤差）は次の（３）式及び（４）式によって算出する。
Ｚ_C （ｉ）＝Ｚ_C （ｉ−１）＋Ｋ_C ．Δｗｔ …（３）
但し、Ｋ_C ：ロールギャップ誤差学習ゲイン（第１ゲイン）
Ｋ_C ≦１
Δｗｔ：圧延された素管の実績寸法と目標寸法との寸法偏差
Ｚ_Mj（ｉ）＝Ｚ_Mj（ｉ−１）−Ｋ_M ・（Δｗｔ／Ｒ_Mj） …（４）
但し、Ｋ_M ：マンドレルバー寸法誤差学習ゲイン（第２ゲイン）
Ｋ_C ＋Ｋ_M ≦１
Ｒ_Mj：マンドレルバーの径方向の公称寸法（半径）
【００２３】
ｎ回学習後のロールギャップ誤差学習量Ｚ_C （ｉ）及びその学習回数（算出回数）を記憶しておく。また、ｎ回学習後のマンドレルバー寸法誤差学習量Ｚ_Mj（ｉ）及びその学習回数（算出回数）を、複数のマンドレルバー別に記憶しておく。一方、マンドレルバー寸法誤差初期学習量及びロールギャップ誤差初期学習量を予め設定しておく。これらの数値を（２）式に代入することによって、前述した第１差分及び第２差分をそれぞれ算出する。そして、次の（５）式及び（６）式によって、マンドレルバー寸法誤差学習ゲインＫ_M 及びロールギャップ誤差学習ゲインＫ_C を求める。
Ｋ_M ：Ｋ_C ＝（第１差分）：（第２差分） …（５）
Ｋ_M ＋Ｋ_C ＝一定 …（６）
【００２４】
このようにして求めたマンドレルバー寸法誤差学習ゲインＫ_M 及びロールギャップ誤差学習ゲインＫ_C 、圧延対象の素管より１本前に圧延した素管に係る寸法偏差を（３）式及び（４）式に代入して、ロールギャップ誤差学習量Ｚ_C （ｉ）及びマンドレルバー寸法誤差学習量Ｚ_Mj（ｉ）を算出し、それらを次の（７）式に代入することによって、修正ロールギャップＣを求める。
Ｃ＝Ｃ₀ −Ｚ_C （ｉ）×２＋Ｚ_Mj（ｉ）・Ｒ_Mj×２ …（７）
但し、Ｃ₀ ：予め定めたロールギャップ
【００２５】
マンドレルバーを交換した場合、交換したマンドレルバーに係るマンドレルバー寸法誤差学習回数を零にする。また、圧延スタンドのロールを交換した場合、又は、ロールの基準位置を変更した場合、ロールギャップ誤差学習回数を零にする。
【００２６】
いま、ロールギャップの修正を複数回行った後に、新たなものに交換した直後のマンドレルバーを素管に挿入して、該素管を圧延した場合を考える。圧延された素管に係る寸法偏差においてマンドレルバーの径方向の寸法誤差に起因する第１寸法偏差とマンドレルバーの寸法誤差との第１差分が、前記寸法偏差においてロールギャップ誤差に起因する第２寸法偏差とロールギャップ誤差との第２差分より大きい。そのため、（５）式及び（６）式から明らかな如く、マンドレルバー寸法誤差学習ゲインＫ_M がロールギャップ誤差学習ゲインＫ_C より大きな数値に定められる。
【００２７】
従って、（７）式において、ロールギャップ誤差学習量Ｚ_C （ｉ）が殆ど変更されることなく、マンドレルバー寸法誤差学習量Ｚ_Mj（ｉ）が大きく変更されて、修正ロールギャップが算出される。このように、マンドレルバー若しくは各スタンドのロールを交換した場合、又は、各スタンドの圧下基準位置を修正した場合であっても、それらに対応して修正ロールギャップを算出することができるため、圧延された素管の寸法偏差を十分低減することができる。
【００２８】
【発明の実施の形態】
以下、本発明の実施の形態を図面に基づいて具体的に説明する。
図１は、本発明に係るマンドレルミルの構成を示すブロック図であり、図中、１は中空の素管である。円形柱状の材料をピアサで穿孔してなる素管１内に、尖先円柱状のマンドレルバーＢ₁ が挿入してあり、素管１及びマンドレルバーＢ₁ は矢符方向へ移送される。
【００２９】
素管１及びマンドレルバーＢ₁ の移送域に臨ませて、一対の縦ロール及び横ロールが交互に設けてある複数の圧延スタンド♯１，♯２，…が適宜の間隔で配置してあり、各圧延スタンド♯１，♯２，…は、対をなすロール11，11，…を素管１の外周面に転接させると共に、圧下装置２，２，…によって素管１に圧下力を加える。各圧延スタンド♯１，♯２，…には後述する如く算定したロールギャップが設定してあり、素管１及びマンドレルバーＢ₁ が各圧延スタンド♯１，♯２，…を通過する間に、素管１は圧延スタンド♯１，♯２，…に設定してあるロールギャップ、及びマンドレルバーＢ₁ の直径によって定まる厚さに圧延される。
【００３０】
素管１の圧延が終了した場合、圧延後の素管からマンドレルバーＢ₁ が、図示しないストリッパによって引き抜かれ、引き抜かれたマンドレルバーＢ₁ は、複数のマンドレルバーＢ_n ，…，Ｂ₃ ，Ｂ₂ を末尾から先頭へ循環させるサーキュレータ７の末尾に移送され、そこで次の圧延に備えて冷却される。次の素管１がピアサからマンドレルミルに供給されると、サーキュレータ７の先頭に位置するマンドレルバーＢ_n がその素管１内に挿入される。
【００３１】
なお、マンドレルバーは、複数の直径寸法別に複数のバーＢ₁ ，Ｂ₂ ，…，Ｂ_n がそれぞれ準備してあり、注文に応じた直径寸法に属する複数のマンドレルバーＢ₁ ，Ｂ₂ ，…，Ｂ_n がサーキュレータ７に搭載してある。
【００３２】
各圧下装置２，２，…に対応して、ロールの圧下位置を制御する圧下位置制御装置３，３，…が設けてあり、各圧下位置制御装置３，３，…には、ロールギャップの修正演算を行う演算装置５から、修正ロールギャップがそれぞれ与えられるようになっている。
【００３３】
サーキュレータ７に搭載されたマンドレルバーＢ₁ ，Ｂ₂ ，…，Ｂ_n には相異なる識別番号がそれぞれ付してあり、サーキュレータ７によって循環されるマンドレルバーＢ₁ ，Ｂ₂ ，…，Ｂ_n 及びその識別番号は、トラッキング装置６によってトラッキングされる。そして、トラッキング装置６は、サーキュレータ７から素管１に挿入されたマンドレルバーＢの識別番号を演算装置５に与える。
【００３４】
また、最終圧延スタンドの出側には圧延された素管１の厚さを測定する厚さ計８が配置してあり、該厚さ計８は最終圧延スタンドから圧延された素管１が送出された場合、その素管１の厚さを測定し、その結果を演算装置５に与える。演算装置５は、マンドレルバーＢが挿入された素管１の圧延に先立って、修正ロールギャップを次のように算出し、それを圧下位置制御装置３，３，…に与える。
【００３５】
演算装置５から各圧下位置制御装置３，３，…に修正ロールギャップが与えられると、圧下位置制御装置３，３，…は圧下装置２，２，…をして、各圧延スタンド♯１，♯２，…のロール11，11，…の圧下位置を修正ロールギャップに対応する位置に調整せしめる。
【００３６】
図２は、図１に示した演算装置５の構成を示すブロック図であり、図３及び図４は、図２に示した演算装置５による修正ロールギャップの算出手順を示すフローチャートである。演算装置５は、素管に係る目標厚さ及びロールギャップ等のデータ、並びに複数の識別番号別及びマンドレルバーＢ₁ 〜Ｂ_n の半径等のデータが登録してあるデータ登録部53、ロールギャップの誤差を学習した回数（ロールギャップ誤差学習回数ｎ_R ）をカウントする第１カウンタ51、マンドレルバーＢ₁ 〜Ｂ_n の径方向の寸法誤差を学習した回数（マンドレルバー寸法誤差学習回数ｎ_M ）を複数の識別番号別にカウントする第２カウンタ52、並びに後述する種々の演算を行う演算部55を備えている。
【００３７】
演算部55は、第１カウンタ51及び／又は第２カウンタ52のクリア指令が与えられたか否かを判断し、それが与えられた場合、第１カウンタ51及び／又は第２カウンタ52の対応する学習回数を零にする（ステップＳ１．Ｓ２）。
【００３８】
演算部55は、素管１が圧延される都度、ロールギャップ誤差学習量Ｚ_C 及びマンドレルバー寸法誤差学習量Ｚ_Mjを後述する如く算出し、得られたロールギャップ誤差学習量Ｚ_C 及び複数の識別番号別にマンドレルバー寸法誤差学習量Ｚ_Mjを学習量記憶部56に与え、そこに記憶更新させると共に、第１カウンタ51及び第２カウンタ52を更新させるようになしてある。
【００３９】
演算部55は、厚さ計８から圧延された素管１の実績厚さが与えられるまで待機する（ステップＳ３）。演算部55は、厚さ計８から実績厚さが与えられた場合、それを読み込む（ステップＳ４）と共に、データ登録部53から、対応する素管１の目標厚さを読み出し、読み出した目標厚さと実績厚さとの差である厚さ偏差Δｗｔを算出する（ステップＳ５，Ｓ６）。演算部55は、第１カウンタ51及び第２カウンタ52からロールギャップ誤差学習回数ｎ_R 及びマンドレルバー寸法誤差学習回数ｎ_M を読み出し、また、学習量記憶部56からロールギャップ誤差学習量Ｚ_C 及びマンドレルバー寸法誤差学習量Ｚ_Mjを読み出す（ステップＳ７）。
【００４０】
演算部55には、次の（２）式が予め設定してあり、演算部55は、（２）式のｙ_n にマンドレルバー寸法誤差学習量Ｚ_Mj又はロールギャップ誤差学習量Ｚ_C を代入し、ｎにロールギャップ誤差学習回数ｎ_R 又はマンドレルバー寸法誤差学習回数ｎ_M を代入することによって、前述した寸法偏差Δｗｔにおいてマンドレルバーの径方向の寸法誤差に起因する第１寸法偏差（ｙ）及び寸法偏差Δｗｔにおいてロールギャップ誤差に起因する第２寸法偏差（ｙ）を求め（ステップＳ８）、第１寸法偏差とマンドレルバー寸法誤差学習量Ｚ_Mjとの第１差分、及び第２寸法偏差とロールギャップ誤差学習量Ｚ_C との第２差分を算出する（ステップＳ９）。なお、（２）式中のｙ₀ は予め設定してある。
【００４１】
【数３】

【００４２】
演算部55には、次の（５）式及び（６）式が予め設定してあり、演算部55は、（５）式及び（６）式によって、マンドレルバー寸法誤差学習ゲインＫ_M 及びロールギャップ誤差学習ゲインＫ_C を求める（ステップＳ10，Ｓ11）。
Ｋ_M ：Ｋ_C ＝（第１差分）：（第２差分） …（５）
Ｋ_M ＋Ｋ_C ＝一定 …（６）
【００４３】
演算部55には次の（３）式及び（４）式が設定してあり、演算部55は、両式に、前述した如く求めたマンドレルバー寸法誤差学習ゲインＫ_M 及びロールギャップ誤差学習ゲインＫ_C 、学習量記憶部56から読み出したロールギャップ誤差学習量Ｚ_C （ｉ−１）及びマンドレルバー寸法誤差学習量Ｚ_Mj（ｉ−１）、並びに厚さ偏差Δｗｔを代入して、ロールギャップ誤差学習量Ｚ_C （ｉ）、及びマンドレルバー寸法誤差学習量Ｚ_Mj（ｉ）を算出する（ステップＳ12）。
Ｚ_C （ｉ）＝Ｚ_C （ｉ−１）＋Ｋ_C ．Δｗｔ …（３）
Ｚ_Mj（ｉ）＝Ｚ_Mj（ｉ−１）−Ｋ_M ・（Δｗｔ／Ｒ_Mj） …（４）
但し、Ｒ_Mj：マンドレルバーの径方向の公称寸法
【００４４】
演算部55は、学習量記憶部56に記憶してあるロールギャップ誤差学習量及びマンドレルバー寸法誤差学習量を、前述した如く算出して得られたロールギャップ誤差学習量Ｚ_C （ｉ）、及びマンドレルバー寸法誤差学習量Ｚ_Mj（ｉ）に更新すると共に、第１カウンタ51のロールギャップ誤差学習回数ｎ_R 、及び第２カウンタ52の対応するマンドレルバー寸法誤差学習回数ｎ_M をそれぞれ１つ増大させる（ステップＳ13，Ｓ14）。
【００４５】
演算部55には次の（７）式が予め設定してある。演算部55は、データ登録部53から、素管１を圧延すべく予め定めたロールギャップＣ₀ を読み出し（ステップＳ15）、読み出したロールギャップＣ₀ 、並びに算出したロールギャップ誤差学習量Ｚ_C （ｉ）及び識別番号別のマンドレルバー寸法誤差学習量Ｚ_Mj（ｉ）、マンドレルバーの径方向の公称寸法Ｒ_Mjを（７）式に代入することによって、修正ロールギャップＣを求める（ステップＳ16）。
Ｃ＝Ｃ₀ −Ｚ_C （ｉ）×２＋Ｚ_Mj（ｉ）・Ｒ_Mj×２ …（７）
【００４６】
なお、本実施の形態では、圧延された素管１の厚さを厚さ計８によって測定しているが、本発明はこれに限らず、マンドレルミルの出側に圧延された素管の長さを測定する長さ計を配置しておくと共に、マンドレルミルの入側に圧延される素管の重量を測定する重量計を配置しておいてもよい。この場合、長さ計によって測定された素管の長さＬ及び重量計によって測定された素管の重量Ｗを、次の（８）式に代入することによって、圧延された素管の厚さＴを算出する。
Ｔ＝（Ｄ／２）−√（Ｄ／２）² −Ｗ（１−ΔＷ／１００）／（πρＬ）…（８）
但し、Ｄ ：圧延された素管の目標外径、又は圧延された素管の実績外径
ΔＷ：素管のスケールロス率
π：円周率
ρ：圧延時の素管の比重
【００４７】
また、本実施の形態では、圧延された素管の厚さ偏差を低減するようになしてあるが、本発明はこれに限らず、圧延された素管の長さ偏差を低減するようになしてもよいことはいうまでもない。
【００４８】
ところで、マンドレルバー寸法誤差学習ゲインＫ_M 及びロールギャップ誤差学習ゲインＫ_C を求めるステップＳ８〜ステップＳ11にあっては、次の操作を実施してもよい。即ち、（２）式の右辺における、厚さ偏差においてロールギャップ誤差に起因する寸法偏差又は厚さ偏差においてマンドレルバー寸法誤差に起因する寸法偏差ｙの概算値を予め想定し、ｎ回学習後の第１差分及び第２差分（ｙ−ｙ_n ）を、ロールギャップ学習回数ｎ_R 又はマンドレルバー寸法誤差学習回数ｎ_M の関数として表すと、次の（９）式及び（10）式のようになる。
第１差分（ｙ−ｙ₀ ）＝ｆ_M （ｎ_M ） …（９）
第２差分（ｙ−ｙ₀ ）＝ｆ_M （ｎ_R ） …（10）
【００４９】
この（９）式及び（10）式、並びに（５）式及び（６）式から、マンドレルバー寸法誤差学習ゲインＫ_M 及びロールギャップ誤差学習ゲインＫ_C を算出するための次の（11）式及び（12）式が導かれる。
Ｋ_M ＝ｇ_M （ｎ_M ，ｎ_R ） …（11）
Ｋ_C ＝ｇ_R （ｎ_M ，ｎ_R ） …（12）
【００５０】
このようにして求めた（11）式及び（12）式を、演算部55に予め設定しておく。そして、演算部55は、（11）式及び（12）式によって、マンドレルバー寸法誤差学習ゲインＫ_M 及びロールギャップ誤差学習ゲインＫ_C を求める。
【００５１】
更に、前述したステップＳ16で用いる（７）式に代えて、例えば、最終圧延スタンドの出側に配置してあるサイザ（図示せず）によって素管１に加える張力を修正するようになした式を用いてもよい。また、ステップＳ16で用いるロールギャップＣ₀ は、データ登録部53に予め登録しておいてもよいし、予め定めた数式に基づいて圧延条件に応じて演算装置５で算出するようになしてもよい。
【００５２】
【実施例】
次に比較試験を行った結果について説明する。
図５は、比較試験を行った結果を示すヒストグラムであり、図中、ハッチングを付したバーは本発明方法による結果を、ハッチングを付していないバーは特開昭61−269909号公報に開示された方法による結果を示している。また、縦軸は頻度を、横軸は実績厚さと目標厚さとの偏差の目標厚さに対する割合をそれぞれ示している。本発明方法では、素管を圧延する都度、図６に示した如く、ロールギャップ誤差学習ゲイン（◇印）及びマンドレルバー寸法誤差学習ゲイン（〇印）を変更して圧延を行った。
【００５３】
図５から明らかな如く、従来方法では、圧延した素管の実績厚さと目標厚さとの偏差の目標厚さに対する割合が±０．５％以内であった割合は６８．２％であった。これに対し、本発明方法では、圧延した素管の実績厚さと目標厚さとの偏差の目標厚さに対する割合が±０．５％以内であった割合は８３．４％であり、前述した偏差を更に低減することができた。
【００５４】
【発明の効果】
以上詳述した如く、本発明にあっては、マンドレルバー若しくは各スタンドのロールを交換した場合、又は、各スタンドの圧下基準位置を修正した場合であっても、圧延された素管の寸法偏差を十分低減することができる等、本発明は優れた効果を奏する。
【図面の簡単な説明】
【図１】本発明に係るマンドレルミルの構成を示すブロック図である。
【図２】図１に示した演算装置の構成を示すブロック図である。
【図３】図２に示した演算装置による修正ロールギャップの算出手順を示すフローチャートである。
【図４】図２に示した演算装置による修正ロールギャップの算出手順を示すフローチャートである。
【図５】比較試験を行った結果を示すヒストグラムである。
【図６】本発明方法において、ロールギャップ誤差学習ゲイン及びマンドレルバー寸法誤差学習ゲインを変更した結果を示すグラフである。
【図７】マンドレルミルによって素管を圧延している状態を示す模式図である。
【符号の説明】
１ 素管
２ 圧下装置
３ 圧下位置制御装置
５ 演算装置
６ トラッキング装置
７ サーキュレータ
８ 厚さ計
Ｂ マンドレルバー
♯１ 圧延スタンド[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method of rolling a raw tube by a mandrel mill in which a rolling roll is rolled on the outer peripheral surface of a hollow raw tube in which a mandrel bar is inserted, and a mandrel mill used for the method.
[0002]
[Prior art]
FIG. 7 is a schematic view showing a state in which the raw tube is rolled by a mandrel mill, and in the figure, 1 is a hollow raw tube. A pointed columnar mandrel bar B is inserted into a tube 1 formed by drilling a circular columnar material with a piercer, and the tube 1 and the mandrel bar B are moved in the direction of the arrow. As the mandrel bar, a plurality of bars are prepared for each of a plurality of diameters, and a plurality of mandrel bars belonging to a diameter corresponding to an order are circulated and used.
[0003]
A plurality of rolling stands # 1, # 2,... Alternately having a pair of vertical rolls and horizontal rolls are arranged in tandem so as to face the transfer area of the raw tube 1 and the mandrel bar B, and each rolling stand # 1. , # 2,... Make a pair of rolls 11, 11,... Roll into contact with the outer peripheral surface of the tube 1 and apply a rolling force to the tube 1 by a rolling device (not shown). A predetermined roll gap is set in each of the rolling stands # 1, # 2,... While the raw tube 1 and the mandrel bar B pass through each of the rolling stands # 1, # 2,. It is rolled to a thickness determined by the roll gap set in the rolling stands # 1, # 2,... And the diameter of the mandrel bar B.
[0004]
When the rolling of the blank tube 1 is completed, the mandrel bar B is pulled out of the rolled blank tube 1 by a stripper (not shown), and the pulled mandrel bar B is cooled in preparation for the next rolling.
[0005]
When rolling the blank 1 with such a mandrel mill, a deviation between the actual thickness of the blank rolled immediately before and a predetermined target thickness is obtained, and each rolling stand # 1, # 1 is set so that the thickness deviation is reduced. The roll gap of # 2, ... was adjusted.
[0006]
By the way, as a cause of occurrence of the thickness deviation, a roll gap setting error and a dimensional error of the mandrel bar B due to a contact wear with the raw tube 1 and a manufacturing error can be considered. Although the influence of the setting error of the roll gap on the thickness deviation and the influence of the dimensional error of the mandrel bar B on the thickness deviation are different from each other in the conventional method described above, the thickness cannot be distinguished. There was a problem that the deviation could not be reduced sufficiently.
[0007]
Therefore, the following method is disclosed in Japanese Patent Laid-Open No. 61-269909. A plurality of mandrel bars are respectively rolled by a plurality of mandrel bars, and thickness deviations between the actual thickness and the target thickness of the rolled raw tubes are obtained. The actual thickness of the rolled raw tube is measured by measuring the length of the rolled raw tube on the exit side of the final stand while placing a weigh scale that measures the weight of the raw tube on the entry side of the first stand. It is possible to obtain a length meter to be calculated, based on the weight of the raw tube measured with the weight meter and the length of the raw tube measured with the length meter, or rolled to the exit side of the final stand. Alternatively, a thickness meter for measuring the thickness of the raw pipe may be arranged, and the thickness of the rolled raw pipe may be determined by measuring with the thickness gauge.
[0008]
An average of a plurality of thickness deviations obtained as described above is calculated, and this average thickness deviation is defined as a roll gap-induced thickness deviation resulting from a roll gap setting error. Further, for each mandrel bar, a partial average thickness deviation, which is an average of thickness deviations of a plurality of raw pipes obtained by rolling using the mandrel bar, is calculated. Thus, the difference between the partial average thickness deviation obtained for each mandrel bar and the above-described thickness deviation is obtained, and these are set as the mandrel bar-induced thickness deviation resulting from the dimensional error of the mandrel bar. And based on the mandrel bar-derived thickness deviation and roll gap-derived thickness deviation related to the mandrel bar used for the next rolling, the roll gap adjustment amount is calculated, and each stand is adjusted so that the obtained adjustment amount is obtained. Adjust the roll gap.
[0009]
[Problems to be solved by the invention]
However, the method disclosed in Japanese Patent Laid-Open No. 61-269909 has the following problems. When the mandrel bar or roll of each stand is replaced with a new one, or when the reduction reference position of each stand is corrected, the mandrel bar-derived thickness deviation and roll gap-derived thickness deviation obtained as described above are discarded. , Using the mandrel bar or roll after replacement, or the reduced reference position after correction, as before, again based on the thickness deviation obtained by rolling a plurality of blank tubes with a plurality of mandrel bars, Thickness deviation and roll gap-induced thickness deviation must be calculated respectively. Therefore, the thickness accuracy of the raw tube rolled during this period is low.
[0010]
Also, to calculate the mandrel bar-derived thickness deviation and roll gap-derived thickness deviation from the calculated thickness deviation or the average thickness deviation value obtained by measurement, the calculation is performed when the average parameter is small. The effect of error or measurement error is large, and the accuracy of mandrel bar-induced thickness deviation and roll gap-induced thickness deviation is low. Therefore, there is a limit to reducing the thickness deviation.
[0011]
The present invention has been made in view of such circumstances, and the purpose thereof is when the mandrel bar or the roll of each stand is replaced, or when the reduction reference position of each stand is corrected. Another object of the present invention is to provide a mandrel mill rolling method using a mandrel mill that can sufficiently reduce the dimensional deviation of the rolled element tube, and a mandrel mill used for the method.
[0012]
[Means for Solving the Problems]
The raw tube rolling method using the mandrel mill according to the first aspect of the present invention is a method in which a single mandrel bar selected from a plurality of mandrel bars is inserted into the raw tube, and the raw tube and the mandrel bar are placed on a rolling stand provided with a pair of rolls. When rolling the raw pipe at the rolling stand, a dimensional deviation between the actual dimension of the previously rolled raw pipe and a predetermined target dimension is obtained, and a predetermined rolling is performed based on the obtained dimensional deviation. In the method of correcting the setting value and rolling the raw tube with the corrected rolling setting value, each time the raw tube is rolled, the rolling setting value required for rolling the raw tube to the target thickness and the corrected rolling setting value The rolling set value error is calculated using the first gain determined corresponding to the dimensional deviation and the rolling set value error, and the mandrel bar error, which is an error between the reference size of the mandrel bar and the actual size, Said When calculating using the second gain determined corresponding to the legal deviation and the mandrel bar error, the first set of times is calculated using the number of calculations of the rolling set value and the number of calculations of the mandrel bar error related to the mandrel bar. A gain and a second gain are determined, and the rolling setting value is corrected using the calculated rolling setting value error and mandrel bar error.
[0013]
According to the second aspect of the present invention, there is provided a method of rolling a tube using a mandrel mill, wherein a mandrel bar selected from a plurality of mandrel bars is inserted into the tube, and the tube and the mandrel bar are placed on a rolling stand provided with a pair of rolls. When rolling the raw pipe at the rolling stand, a dimensional deviation between the actual dimension of the previously rolled raw pipe and a predetermined target dimension is obtained, and a predetermined rolling is performed based on the obtained dimensional deviation. In the method of correcting the setting value and rolling the raw tube with the corrected rolling setting value, each time the raw tube is rolled, the rolling setting value required for rolling the raw tube to the target thickness and the corrected rolling setting value The rolling set value error is calculated using a first gain determined corresponding to the dimensional deviation and the rolling set value error, the obtained rolling set value error is stored, and the number of calculation of the rolling set value error is calculated. Counting and also The mandrel bar error, which is the error between the standard and actual dimensions of the dodrel bar, is calculated using the second gain determined corresponding to the dimensional deviation and mandrel bar error, and the obtained mandrel bar error is calculated for each mandrel bar. While storing, the number of calculation of mandrel bar error is counted for each mandrel bar, and the first difference between the partial dimensional deviation caused by the rolling set value error and the rolling set value error in the dimensional deviation is stored. A mandrel bar that is calculated using the number of calculation of the error and the rolling set value error, and stores a second difference between the partial dimensional deviation due to the mandrel bar error and the mandrel bar error in the dimensional deviation according to the mandrel bar. Calculated using the error and the number of mandrel bar errors calculated according to the mandrel bar. Based on the ratio between the first difference and the second difference, the first gain and the second gain are obtained, and the rolling set value error and the mandrel bar error calculated using the obtained first gain and second gain are used. The rolling set value is corrected.
[0014]
When the roll is replaced in the first or second invention, or when the rolling reference position of the roll is changed in the first or second invention, the number of calculation of the rolling set value error is reduced to zero. It is characterized by doing.
[0015]
According to a fourth aspect of the present invention, there is provided a method of rolling a tube using a mandrel mill, wherein, in any of the first to third inventions, when any of the mandrel bars is replaced, the corresponding mandrel bar error is calculated to be zero. And
[0016]
A mandrel mill according to a fifth aspect of the invention inserts one mandrel bar selected from a plurality of mandrel bars into the raw pipe, feeds the raw pipe and the mandrel bar to a rolling stand provided with a pair of rolls, When rolling the pipe on the rolling stand, obtain the dimensional deviation between the actual dimension of the previously rolled pipe and the predetermined target dimension, and correct the predetermined rolling setting value based on the obtained dimensional deviation. In a mandrel mill that rolls the raw tube with a corrected rolling setting value, a rolling setting value error between the rolling setting value required for rolling the raw tube to a target thickness and the corrected rolling setting value is represented by the dimensional deviation and Means for calculating using a first gain determined corresponding to the rolling set value error, means for storing the obtained rolling set value error, means for counting the number of times of calculation of the rolling set value error, and a mandrel bar Base Means for calculating a mandrel bar error, which is an error between a size and an actual size, using a second gain determined corresponding to the dimensional deviation and the mandrel bar error, and storing the obtained mandrel bar error for each mandrel bar A means for counting the number of mandrel bar errors calculated for each mandrel bar, and the first difference between the partial dimensional deviation caused by the rolling set value error and the rolling set value error in the dimensional deviation According to the mandrel bar, the second difference between the partial dimensional deviation and the mandrel bar error due to the mandrel bar error in the dimensional deviation is calculated using the calculation value of the set value error and the rolling set value error. Calculate using the stored mandrel bar error and the number of mandrel bar errors calculated according to the mandrel bar. And means for determining the ratio between the obtained first difference and the second difference; means for determining the first gain and the second gain based on the ratio between the first difference and the second difference; And a means for correcting the rolling set value using the rolling set value error and the mandrel bar error calculated using the first gain and the second gain.
[0017]
Hereinafter, the present invention will be described using a roll gap which is one of rolling setting values. Assuming that the radial dimension of the mandrel bar is the reference dimension, the roll gap error, which is the error between the roll gap required to roll the blank tube to the target dimension and the corrected roll gap, and the radial direction of the mandrel bar A mandrel bar error that is an error between the reference dimension and the actual dimension is learned.
[0018]
When exponential smoothing learning represented by Zi + 1 = K · E + (1−K) · Zi (where Zi is the learning amount, K is the learning gain, and E is the actual deviation) is applied The second difference between the partial dimensional deviation due to the radial dimension error of the mandrel bar in the dimensional deviation and the dimensional error of the mandrel bar, and the partial dimensional deviation due to the roll gap error in the dimensional deviation and the roll gap error. The first difference can be expressed by the following equation (1).
[0019]
[Expression 1]

[0020]
Since the dimensional deviation includes a change with time and other disturbances, the equation (1) is rewritten into the following equation (2) having a first-order lag offset γ in consideration of these.
[0021]
[Expression 2]

[0022]
Roll gap error learning amount Z after learning n times _C (I) and the mandrel bar dimension error learning amount Z after learning n times _Mj (I) (Mandrel bar error) is calculated by the following equations (3) and (4).
Z _C (I) = Z _C (I-1) + K _C . Δwt (3)
However, K _C : Roll gap error learning gain (first gain)
K _C ≦ 1
Δwt: Dimensional deviation between the actual dimension of the rolled blank and the target dimension
Z _Mj (I) = Z _Mj (I-1) -K _M ・ (Δwt / R _Mj (4)
However, K _M : Mandrel bar dimensional error learning gain (2nd gain)
K _C + K _M ≦ 1
R _Mj : Nominal dimension (radius) of the mandrel bar in the radial direction
[0023]
Roll gap error learning amount Z after learning n times _C (I) and the learning count (calculation count) are stored. In addition, the mandrel bar dimension error learning amount Z after learning n times _Mj (I) and the number of times of learning (the number of calculations) are stored for each of a plurality of mandrel bars. On the other hand, a mandrel bar dimension error initial learning amount and a roll gap error initial learning amount are set in advance. By substituting these numerical values into equation (2), the above-described first difference and second difference are calculated. Then, the mandrel bar dimension error learning gain K is obtained by the following equations (5) and (6). _M And roll gap error learning gain K _C Ask for.
K _M : K _C = (First difference): (second difference) (5)
K _M + K _C = Constant ... (6)
[0024]
Mandrel bar size error learning gain K thus obtained _M And roll gap error learning gain K _C , Substituting the dimensional deviation of the raw pipe rolled one before the raw pipe to be rolled into the equations (3) and (4), the roll gap error learning amount Z _C (I) and mandrel bar dimensional error learning amount Z _Mj The corrected roll gap C is obtained by calculating (i) and substituting them into the following equation (7).
C = C ₀ -Z _C (I) × 2 + Z _Mj (I) ・ R _Mj × 2 (7)
However, C ₀ : Predetermined roll gap
[0025]
When the mandrel bar is replaced, the number of times of learning the mandrel bar dimension error related to the replaced mandrel bar is set to zero. In addition, when the roll of the rolling stand is replaced or when the reference position of the roll is changed, the number of roll gap error learning is set to zero.
[0026]
Now, consider the case where the roll gap is corrected a plurality of times, the mandrel bar immediately after replacement with a new one is inserted into the blank, and the blank is rolled. The first difference between the first dimensional deviation due to the radial dimension error of the mandrel bar and the dimensional error of the mandrel bar in the dimensional deviation related to the rolled raw pipe is the second due to the roll gap error in the dimensional deviation. It is larger than the second difference between the dimensional deviation and the roll gap error. Therefore, as apparent from the equations (5) and (6), the mandrel bar size error learning gain K _M Is roll gap error learning gain K _C Set to a larger number.
[0027]
Therefore, in equation (7), the roll gap error learning amount Z _C Mandrel bar dimensional error learning amount Z with almost no change in (i) _Mj (I) is significantly changed, and the corrected roll gap is calculated. Thus, even when the mandrel bar or the roll of each stand is replaced, or even when the reduction reference position of each stand is corrected, the corrected roll gap can be calculated correspondingly, so rolling Therefore, the dimensional deviation of the formed raw tube can be sufficiently reduced.
[0028]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a block diagram showing a configuration of a mandrel mill according to the present invention, in which 1 is a hollow shell. A circular columnar material is drilled with a piercer, and a pointed columnar mandrel bar B is formed in the tube 1. ₁ Is inserted, the tube 1 and the mandrel bar B ₁ Is moved in the direction of the arrow.
[0029]
Base tube 1 and mandrel bar B ₁ , A plurality of rolling stands # 1, # 2,... In which a pair of vertical rolls and horizontal rolls are alternately provided are arranged at appropriate intervals, and each rolling stand # 1, # 2 is arranged. ,... Make pairs of rolls 11, 11,. Each rolling stand # 1, # 2,... Has a roll gap calculated as described later, and the raw tube 1 and the mandrel bar B ₁ Passes through each of the rolling stands # 1, # 2,..., The raw tube 1 has a roll gap set in the rolling stands # 1, # 2,. ₁ Rolled to a thickness determined by the diameter.
[0030]
When the rolling of the blank 1 is completed, the mandrel bar B is rolled from the blank after rolling. ₁ Is pulled out by a stripper (not shown) and pulled out. ₁ The multiple mandrel bars B _n , ..., B _Three , B ₂ Is transferred to the end of the circulator 7 that circulates from the end to the start, where it is cooled in preparation for the next rolling. When the next tube 1 is supplied from the piercer to the mandrel mill, the mandrel bar B located at the head of the circulator 7 _n Is inserted into the tube 1.
[0031]
In addition, the mandrel bar has a plurality of bars B according to a plurality of diameter dimensions. ₁ , B ₂ , ..., B _n Are prepared, and a plurality of mandrel bars B belonging to the diameter size according to the order ₁ , B ₂ , ..., B _n Is mounted on the circulator 7.
[0032]
Corresponding to each of the reduction devices 2, 2,..., Reduction position control devices 3, 3,... For controlling the reduction position of the roll are provided, and each of the reduction position control devices 3, 3,. A correction roll gap is given from each of the arithmetic units 5 that perform the correction calculation.
[0033]
Mandrel bar B mounted on circulator 7 ₁ , B ₂ , ..., B _n Each has a different identification number, and a mandrel bar B circulated by the circulator 7. ₁ , B ₂ , ..., B _n The identification number is tracked by the tracking device 6. Then, the tracking device 6 gives the identification number of the mandrel bar B inserted into the raw tube 1 from the circulator 7 to the arithmetic device 5.
[0034]
In addition, a thickness meter 8 for measuring the thickness of the rolled raw tube 1 is disposed on the exit side of the final rolling stand, and the thickness meter 8 is fed by the raw tube 1 rolled from the final rolling stand. If so, the thickness of the tube 1 is measured and the result is given to the arithmetic unit 5. Prior to the rolling of the tube 1 in which the mandrel bar B is inserted, the arithmetic device 5 calculates a corrected roll gap as follows and gives it to the reduction position control devices 3, 3,.
[0035]
When the correction roll gap is given from the arithmetic unit 5 to the respective rolling position control devices 3, 3,..., The rolling position control devices 3, 3,. The roll-down positions of the rolls 11, 11, ... of # 2, ... are adjusted to positions corresponding to the corrected roll gap.
[0036]
FIG. 2 is a block diagram showing a configuration of the arithmetic device 5 shown in FIG. 1, and FIGS. 3 and 4 are flowcharts showing a procedure for calculating a corrected roll gap by the arithmetic device 5 shown in FIG. The arithmetic unit 5 includes data such as a target thickness and a roll gap related to the raw tube, and a plurality of identification numbers and mandrel bars B ₁ ~ B _n The data registration unit 53 in which data such as the radius is registered, the number of times of learning the error of the roll gap (the number of times of roll gap error learning n _R ) Counting first counter 51, mandrel bar B ₁ ~ B _n Number of times of learning the dimensional error in the radial direction of the mandrel bar (number of mandrel bar dimensional error learning times _M ) For each of a plurality of identification numbers, and a calculation unit 55 for performing various calculations described later.
[0037]
The calculation unit 55 determines whether or not a clear command for the first counter 51 and / or the second counter 52 is given, and when it is given, the first counter 51 and / or the second counter 52 corresponds. The number of learning is set to zero (steps S1 and S2).
[0038]
The calculation unit 55 calculates the roll gap error learning amount Z every time the raw tube 1 is rolled. _C And mandrel bar dimensional error learning amount Z _Mj Is calculated as described later, and the obtained roll gap error learning amount Z _C Mandrel bar dimensional error learning amount Z for each of a plurality of identification numbers _Mj Is supplied to the learning amount storage unit 56, and the first counter 51 and the second counter 52 are updated at the same time.
[0039]
The calculation unit 55 waits until the actual thickness of the rolled raw tube 1 is given from the thickness gauge 8 (step S3). When the actual thickness is given from the thickness meter 8, the calculation unit 55 reads the actual thickness (step S 4), reads the target thickness of the corresponding raw tube 1 from the data registration unit 53, and reads the read target thickness Thickness deviation Δwt, which is the difference between the actual thickness and the actual thickness, is calculated (steps S5 and S6). The calculation unit 55 calculates the number n of roll gap error learning times from the first counter 51 and the second counter 52. _R And mandrel bar dimensional error learning count n _M And a roll gap error learning amount Z from the learning amount storage unit 56. _C And mandrel bar dimensional error learning amount Z _Mj Is read (step S7).
[0040]
The following equation (2) is set in advance in the calculation unit 55, and the calculation unit 55 stores the y in the equation (2). _n Mandrel bar dimension error learning amount Z _Mj Or roll gap error learning amount Z _C And n is the number of roll gap error learning times n _R Or mandrel bar size error learning count n _M Is substituted for the first dimensional deviation (y) due to the dimensional error in the radial direction of the mandrel bar at the dimensional deviation Δwt and the second dimensional deviation (y) due to the roll gap error at the dimensional deviation Δwt. (Step S8), first dimensional deviation and mandrel bar dimensional error learning amount Z _Mj First difference, second dimensional deviation and roll gap error learning amount Z _C Is calculated (step S9). In the formula (2), y ₀ Are preset.
[0041]
[Equation 3]

[0042]
The following equation (5) and equation (6) are preset in the calculation unit 55, and the calculation unit 55 calculates the mandrel bar size error learning gain K according to the equations (5) and (6). _M And roll gap error learning gain K _C Is obtained (steps S10 and S11).
K _M : K _C = (First difference): (second difference) (5)
K _M + K _C = Constant ... (6)
[0043]
The following equation (3) and equation (4) are set in the calculation unit 55, and the calculation unit 55 sets the mandrel bar size error learning gain K obtained as described above in both equations. _M And roll gap error learning gain K _C The roll gap error learning amount Z read from the learning amount storage unit 56 _C (I-1) and mandrel bar dimensional error learning amount Z _Mj Substituting (i-1) and thickness deviation Δwt, roll gap error learning amount Z _C (I) and mandrel bar dimensional error learning amount Z _Mj (I) is calculated (step S12).
Z _C (I) = Z _C (I-1) + K _C . Δwt (3)
Z _Mj (I) = Z _Mj (I-1) -K _M ・ (Δwt / R _Mj (4)
However, R _Mj : Nominal dimension of the mandrel bar in the radial direction
[0044]
The calculation unit 55 calculates the roll gap error learning amount Z obtained by calculating the roll gap error learning amount and the mandrel bar dimension error learning amount stored in the learning amount storage unit 56 as described above. _C (I) and mandrel bar dimensional error learning amount Z _Mj In addition to updating to (i), the number n of roll gap error learning times of the first counter 51 _R , And the corresponding mandrel bar size error learning count n of the second counter 52 _M Is increased by one (steps S13 and S14).
[0045]
The following equation (7) is preset in the calculation unit 55. The calculation unit 55 receives a roll gap C determined in advance from the data registration unit 53 so as to roll the blank 1. ₀ (Step S15), and the read roll gap C ₀ And the calculated roll gap error learning amount Z _C (I) and mandrel bar dimensional error learning amount Z for each identification number _Mj (I) Nominal dimension R in the radial direction of the mandrel bar _Mj Is substituted into equation (7) to obtain a corrected roll gap C (step S16).
C = C ₀ -Z _C (I) × 2 + Z _Mj (I) ・ R _Mj × 2 (7)
[0046]
In addition, in this Embodiment, although the thickness of the rolled raw tube 1 is measured by the thickness meter 8, this invention is not restricted to this, The length of the raw tube rolled on the exit side of the mandrel mill A length meter for measuring the length may be disposed, and a weight meter for measuring the weight of the raw tube to be rolled may be disposed on the entry side of the mandrel mill. In this case, the thickness L of the rolled tube is substituted by substituting the length L of the tube measured by the length meter and the weight W of the tube measured by the weight meter into the following equation (8). T is calculated.
T = (D / 2) -√ (D / 2) ² -W (1-ΔW / 100) / (πρL) (8)
However, D: The target outer diameter of the rolled raw pipe, or the actual outer diameter of the rolled raw pipe
ΔW: Scale loss rate of the tube
π: Pi ratio
ρ: Specific gravity of the raw tube during rolling
[0047]
In the present embodiment, the thickness deviation of the rolled raw tube is reduced. However, the present invention is not limited to this, and the length deviation of the rolled raw tube is reduced. Needless to say, it may be.
[0048]
By the way, mandrel bar dimension error learning gain K _M And roll gap error learning gain K _C In steps S8 to S11 for obtaining the following, the following operation may be performed. That is, in the right side of the equation (2), an estimated value of a dimensional deviation y caused by a roll gap error in a thickness deviation or a dimensional deviation y caused by a mandrel bar dimensional error in a thickness deviation is assumed in advance, and after learning n times First difference and second difference (y−y _n ) Is the roll gap learning count n _R Or mandrel bar size error learning count n _M When expressed as a function, the following equations (9) and (10) are obtained.
First difference (y−y ₀ ) = F _M (N _M (9)
Second difference (y−y ₀ ) = F _M (N _R ) …(Ten)
[0049]
From these equations (9) and (10), and equations (5) and (6), the mandrel bar dimensional error learning gain K _M And roll gap error learning gain K _C The following equations (11) and (12) for calculating
K _M = G _M (N _M , N _R (11)
K _C = G _R (N _M , N _R ... (12)
[0050]
The expressions (11) and (12) thus determined are set in advance in the calculation unit 55. Then, the calculation unit 55 calculates the mandrel bar dimensional error learning gain K by the equations (11) and (12). _M And roll gap error learning gain K _C Ask for.
[0051]
Furthermore, instead of the above-described equation (7) used in step S16, for example, an equation for correcting the tension applied to the raw tube 1 by a sizer (not shown) arranged on the exit side of the final rolling stand, for example. May be used. The roll gap C used in step S16 ₀ May be registered in advance in the data registration unit 53, or may be calculated by the arithmetic unit 5 in accordance with rolling conditions based on a predetermined mathematical formula.
[0052]
【Example】
Next, the results of comparative tests will be described.
FIG. 5 is a histogram showing the results of comparative tests. In the figure, hatched bars are disclosed by the method of the present invention, and unhatched bars are disclosed in JP-A-61-269909. Shows the results of the proposed method. The vertical axis represents frequency, and the horizontal axis represents the ratio of the deviation between the actual thickness and the target thickness to the target thickness. In the method of the present invention, each time the raw pipe was rolled, the roll gap error learning gain (ロ ー ル mark) and the mandrel bar dimension error learning gain (◯ mark) were changed as shown in FIG.
[0053]
As is apparent from FIG. 5, in the conventional method, the ratio of the deviation between the actual thickness of the rolled raw pipe and the target thickness to the target thickness was within ± 0.5%, which was 68.2%. On the other hand, in the method of the present invention, the ratio of the deviation between the actual thickness of the rolled raw tube and the target thickness to the target thickness was within ± 0.5%, which was 83.4%. Can be further reduced.
[0054]
【The invention's effect】
As described above in detail, in the present invention, even when the mandrel bar or the roll of each stand is replaced, or when the reduction reference position of each stand is corrected, the dimensional deviation of the rolled raw tube The present invention has an excellent effect, such as being able to sufficiently reduce.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a mandrel mill according to the present invention.
FIG. 2 is a block diagram showing a configuration of the arithmetic device shown in FIG.
FIG. 3 is a flowchart showing a procedure for calculating a corrected roll gap by the arithmetic unit shown in FIG. 2;
4 is a flowchart showing a procedure for calculating a corrected roll gap by the arithmetic device shown in FIG. 2;
FIG. 5 is a histogram showing the results of a comparative test.
FIG. 6 is a graph showing a result of changing a roll gap error learning gain and a mandrel bar size error learning gain in the method of the present invention.
FIG. 7 is a schematic view showing a state in which a raw pipe is rolled by a mandrel mill.
[Explanation of symbols]
1 Elementary tube
2 Reduction device
3 Reduction position control device
5 Arithmetic unit
6 Tracking device
7 Circulator
8 Thickness gauge
B Mandrel bar
# 1 Rolling stand

Claims (5)

When one mandrel bar selected from a plurality of mandrel bars is inserted into the raw pipe, the raw pipe and the mandrel bar are fed to a rolling stand provided with a pair of rolls, and the raw pipe is rolled by the rolling stand. Determining a dimensional deviation between the actual dimension of the previously rolled raw tube and a predetermined target dimension, correcting a predetermined rolling setting value based on the obtained dimensional deviation, and using the corrected rolling setting value to In a method of rolling a tube,
Each time the raw pipe is rolled, a rolling set value error between the rolling set value required to roll the raw pipe to the target thickness and the corrected rolling set value is determined corresponding to the dimensional deviation and the rolling set value error. Calculation is performed using the first gain, and a mandrel bar error that is an error between the reference dimension of the mandrel bar and the actual dimension is calculated using the second gain determined corresponding to the dimensional deviation and the mandrel bar error. In this case, the first gain and the second gain are determined using the number of calculations of the rolling set value and the number of calculations of the mandrel bar error related to the mandrel bar,
A rolling method using a mandrel mill, wherein the rolling setting value is corrected using the calculated rolling setting value error and the mandrel bar error. When one mandrel bar selected from a plurality of mandrel bars is inserted into the raw pipe, the raw pipe and the mandrel bar are fed to a rolling stand provided with a pair of rolls, and the raw pipe is rolled by the rolling stand. Determining a dimensional deviation between the actual dimension of the previously rolled raw tube and a predetermined target dimension, correcting a predetermined rolling setting value based on the obtained dimensional deviation, and using the corrected rolling setting value to In a method of rolling a tube,
Each time the raw tube is rolled, a rolling set value error between the rolling set value required to roll the raw tube to the target thickness and the corrected rolling set value is determined in accordance with the dimensional deviation and the rolling set value error. Calculation using the first gain, storing the obtained rolling set value error, counting the number of calculation of the rolling set value error, and mandrel bar error, which is an error between the reference and actual dimensions of the mandrel bar Is calculated using the second gain determined corresponding to the dimensional deviation and the mandrel bar error, and the obtained mandrel bar error is stored for each mandrel bar, and the number of mandrel bar errors calculated for each mandrel bar Count,
A first difference between a partial dimensional deviation and a rolling set value error due to the rolling set value error in the dimensional deviation is calculated using the stored rolling set value error and the rolling set value error count, and the dimensional deviation The second difference between the partial dimensional deviation caused by the mandrel bar error and the mandrel bar error is stored using the mandrel bar error stored according to the mandrel bar and the number of mandrel bar errors calculated according to the mandrel bar. And calculating the first gain and the second gain based on the ratio between the obtained first difference and the second difference,
A blank rolling method using a mandrel mill, wherein a rolling set value is corrected using a rolling set value error and a mandrel bar error calculated using the obtained first gain and second gain. 3. The blank tube rolling method using a mandrel mill according to claim 1 or 2, wherein the number of calculation of the rolling set value error is set to zero when the roll is replaced or when the roll reduction reference position is changed. 4. A method of rolling a tube by a mandrel mill according to claim 1, wherein when any of the mandrel bars is replaced, the corresponding mandrel bar error is calculated to be zero. When one mandrel bar selected from a plurality of mandrel bars is inserted into a blank tube, the blank tube and the mandrel bar are fed to a rolling stand provided with a pair of rolls, and the blank tube is rolled by the rolling stand. Determining a dimensional deviation between the actual dimension of the previously rolled raw pipe and a predetermined target dimension, correcting a predetermined rolling setting value based on the obtained dimensional deviation, and correcting the predetermined rolling setting value with the corrected rolling setting value; In a mandrel mill that rolls tubes,
A rolling setting value error between a rolling setting value required for rolling the blank tube to a target thickness and a corrected rolling setting value is calculated using a first gain determined corresponding to the dimensional deviation and the rolling setting value error. Means for storing the obtained rolling set value error, means for counting the number of times of calculation of the rolling set value error, and mandrel bar error, which is an error between the reference and actual dimensions of the mandrel bar, Means for calculating using the second gain determined corresponding to the deviation and mandrel bar error, means for storing the obtained mandrel bar error for each mandrel bar, and counting the number of mandrel bar error calculations for each mandrel bar And means for storing the first difference between the partial dimensional deviation and the rolling set value error caused by the rolling set value error in the dimensional deviation, and storing the rolling set value error and the rolling set value error. Means for calculating using the number of times of calculation, and a mandrel bar error stored in accordance with the mandrel bar and a second difference between the partial dimensional deviation caused by the mandrel bar error and the mandrel bar error in the dimensional deviation, and the mandrel bar Based on the means for calculating the mandrel bar error count counted according to the above, means for determining the ratio between the obtained first difference and the second difference, and the ratio between the first difference and the second difference , Means for obtaining the first gain and the second gain, and means for correcting the rolling set value using the rolling set value error and the mandrel bar error calculated using the obtained first gain and the second gain. A mandrel mill characterized by

Images