Deprecated: The each() function is deprecated. This message will be suppressed on further calls in /home/zhenxiangba/zhenxiangba.com/public_html/phproxy-improved-master/index.php on line 456
JP3845295B2 - Operation analysis apparatus and operation analysis method in steel plate manufacturing process - Google Patents
[go: Go Back, main page]

JP3845295B2 - Operation analysis apparatus and operation analysis method in steel plate manufacturing process - Google Patents

Operation analysis apparatus and operation analysis method in steel plate manufacturing process Download PDF

Info

Publication number
JP3845295B2
JP3845295B2 JP2001358067A JP2001358067A JP3845295B2 JP 3845295 B2 JP3845295 B2 JP 3845295B2 JP 2001358067 A JP2001358067 A JP 2001358067A JP 2001358067 A JP2001358067 A JP 2001358067A JP 3845295 B2 JP3845295 B2 JP 3845295B2
Authority
JP
Japan
Prior art keywords
component
manufacturing process
quality
quality distribution
correlation
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.)
Expired - Fee Related
Application number
JP2001358067A
Other languages
Japanese (ja)
Other versions
JP2003154409A (en
Inventor
俊夫 赤木
淳治 伊勢
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.)
Nippon Steel Corp
Original Assignee
Nippon 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 Nippon Steel Corp filed Critical Nippon Steel Corp
Priority to JP2001358067A priority Critical patent/JP3845295B2/en
Publication of JP2003154409A publication Critical patent/JP2003154409A/en
Application granted granted Critical
Publication of JP3845295B2 publication Critical patent/JP3845295B2/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Landscapes

  • Metal Rolling (AREA)
  • Length Measuring Devices With Unspecified Measuring Means (AREA)

Description

【0001】
【発明の属する技術分野】
本発明は、鋼板製造プロセスにおける操業解析装置及びその操業解析方法に関し、特に、鉄鋼プロセスなどの製造プロセスにおいて、プロセス操業データと製品の品質データとから、製品の品質に影響を与えているプロセス操業データを見出すために用いて好適なものである。
【0002】
【従来の技術】
鋼板製品の品質管理などを目的として、走行中の帯状鋼板の表面疵を全面的に検査をできる自動疵検査装置が各種開発されており、製造現場への導入が進んでいる。また、自動疵検査装置には、疵の有無を検出する機能のみを持つものの他に、検出した疵の種類を判別する機能も併せて持つものも多い。これらの疵検査装置は、品質管理上のみならず、設備・操業異常の早期発見や、操業方法の改善に役立つからである。
【0003】
このような疵の種類を特定する疵判別機能を有する自動疵検査装置及びその検査方法としては、例えば、特開昭54−74793の如く、通常、疵の形状の特徴量などを用いて判別が行われることが多い。
【0004】
一方、圧延ロールに異物が付着して、圧延中の鋼板に繰り返し転写されてできるような圧延ロール疵は、鋼板長手方向に、前記ロールの周囲長に相当する、特有の周期性をもった疵が現れる。この圧延ロール疵に対しては、例えば、特開平6−324005の如く、圧延ロールの回転周期と一致するような位置に繰り返し発生するような疵を監視することで検出が行われている。
【0005】
【発明が解決しようとする課題】
しかし、鋼板は、鋳造、圧延、焼鈍、表面処理など、複雑な工程を経て製造されるため、最終製品の表面にて検出される疵には、様々な種類のものが混合して含まれている。そのため、疵の個々の形状や、発生位置のロール周囲長に相当する周期性を評価するだけでは、それぞれの疵の発生要因を特定することは容易でない。
【0006】
最終製品の表面にて検出される疵には、形状や発生位置のロール周囲長に相当する周期性の他にも、例えば、鋼板の先端付近に多く発生したり、特定のプロセス操業データが変化するときに多く発生するなど、疵の発生要因により、鋼板長手方向で特有の偏りを持って疵が存在する場合も多いが、前記のように、様々な種類の疵が混合して含まれている状況下で、それらの疵を識別することは困難である。
【0007】
本発明は前述の問題点にかんがみてなされたもので、製品に、様々な発生要因の疵が含まれている状況下において、その製品の疵発生に影響を及ぼしていると考えられる製造プロセス操業データを見出すことができるようにすることを目的とする。
【0008】
【課題を解決するための手段】
本発明の鋼板製造プロセスにおける操業解析装置は、鋼板の製造プロセスにて、前記製造プロセス操業データと製品品質との相関を評価する鋼板製造プロセスにおける操業解析装置において、前記鋼板が走行中にその表面疵を連続して検出し、当該表面疵の位置情報を出力する品質測定手段と、前記品質測定手段が出力する表面疵の鋼板長手方向の位置情報から、鋼板全長を、予め定めた一定個数Nで、かつ等しい長さの区間に分けて、各区間での合計疵個数を単位幅当たりの疵個数に換算して算出するか、あるいは所定の幅の範囲内に存在する疵個数を算出する品質分布算出手段と、前記品質分布算出手段が出力するN個の成分からなる品質分布情報をM回蓄積し、前記蓄積したM行N列の品質分布情報行列を独立成分分析により統計的に互いに独立な成分に分解して成分抽出を行い、品質分布のM行N列の独立成分行列を求める成分抽出手段と、前記成分抽出手段で抽出された成分である前記独立成分行列の行ベクトルを用いて、前記品質分布情報である品質分布情報行ベクトルに含まれる成分の大きさを両行ベクトルの内積を求めることによって算出する成分スコア算出手段と、前記品質測定手段にて測定した鋼板が製造された際の製造プロセス操業データを入力する操業データ入力手段と、前記操業データ入力手段によって入力されたM個のデータからなる製造プロセス操業データと、前記成分スコア算出手段によって算出された各品質分布情報に対しM個のデータからなる成分スコアとの相関係数を算出することによって、前記製造プロセス操業データと前記成分スコアとの相関を解析して前記製造プロセス操業データと独立成分との相関の強さを求める相関解析手段とを備えることを特徴とする。
【0009】
本発明の鋼板製造プロセスにおける操業解析方法は、鋼板の製造プロセスにて、前記製造プロセス操業データと製品品質との相関を評価する鋼板製造プロセスにおける操業解析方法において、前記鋼板が走行中にその表面疵を連続して検出し、当該表面疵の位置情報を出力する品質測定工程と、前記品質測定工程が出力する表面疵の鋼板長手方向の位置情報から、鋼板全長を、予め定めた一定個数Nで、かつ等しい長さの区間に分けて、各区間での合計疵個数を単位幅当たりの疵個数に換算して算出するか、あるいは所定の幅の範囲内に存在する疵個数を算出する品質分布算出工程と、前記品質分布算出工程により出力されるN個の成分からなる品質分布情報をM回蓄積し、前記蓄積したM行N列の品質分布情報行列を独立成分分析により統計的に互いに独立な成分に分解して成分抽出を行い、品質分布のM行N列の独立成分行列を求める成分抽出工程と、前記成分抽出工程で抽出された成分である前記独立成分行列の行ベクトルを用いて、前記品質分布情報である品質分布情報行ベクトルに含まれる成分の大きさを両行ベクトルの内積を求めることによって算出する成分スコア算出工程と、前記品質測定工程にて測定した鋼板が製造された際の製造プロセス操業データを入力する操業データ入力工程と、前記操業データ入力工程によって入力されたM個のデータからなる製造プロセス操業データと、前記成分スコア算出工程によって算出された各品質分布情報に対しM個のデータからなる成分スコアとの相関係数を算出することによって、前記製造プロセス操業データと前記成分スコアとの相関を解析して前記製造プロセス操業データと独立成分との相関の強さを求める相関解析工程とを備えることを特徴とする。
【0018】
【発明の実施の形態】
以下、図面を参照して、本発明の鋼板の製造プロセスにおける操業解析装置の実施の形態について説明する。
【0019】
図1は、本実施の形態の操業解析装置の構成を示した図である。
図1において、101は品質測定部であり、走行中の鋼板の品質を連続して測定できる各種の自動疵検査装置、内部欠陥検出装置、あるいは、材料特性測定装置を用いることができる。
【0020】
本実施の形態では、表面疵検査装置を用いた場合について記述する。表面疵検査装置を用いた場合は、同装置によって検出された鋼板上の表面疵の、鋼板上における長手方向位置情報が出力される。
【0021】
102は品質分布算出部であり、前記表面疵の鋼板長手方向の位置情報から、鋼板全長を、予め定めた一定個数NIC個で、かつ等しい長さの区間に分けて、各区間での合計疵個数xを算出する。このとき、xはNIC個の要素を持つベクトルとなる。
【0022】
通常、鋼板の幅寸法は、製品毎に異なるため、前記、各区間での合計疵個数を幅寸法で割り、単位幅あたりの疵個数を算出する。あるいは、鋼板の特定の位置で、一定の大きさの幅の範囲、例えば鋼板のエッジから一定距離以内の疵のみを算出対象としても良い。
【0023】
103は成分抽出部であり、品質分布算出部が出力する品質分布情報xを、M個分蓄積し、蓄積したM個の品質分布情報xm(m=1,2,…,M)の成分抽出を行う。成分の抽出手法としては、主成分分析、独立分析などを用いることができるが、ここでは、独立成分分析の手法を用いて、鋼板の長手方向の疵個数分布に含まれる互いに独立な成分を抽出する実施形態を記述する。独立成分分析法とは、時系列的に連続したデータを統計的に互いに独立な成分に分解するための信号処理方法である。
【0024】
次に、図2を用いて、成分抽出部103における、独立成分分析の手法を用いた、鋼板の長手方向の疵個数分布に含まれる互いに独立な成分を抽出する手法について記述する。
【0025】
独立成分分析に用いるM個の品質分布情報xm(m=1,2,…,M)は、xを行ベクトルとしてM行並べることにより、M行NIC列の行列Xとみなすことができる(ステップS201)。
【0026】
次に、成分の計算に必要な変換行列Wを求めるために、各要素をランダムな値に設定したM行M列の行列を初期行列W=W0として与える(ステップS202)。品質分布データXを、式(1)に従って、変換行列Wを用いて変換し、独立成分を行要素に持つ行列である独立成分行列ICを求める(ステップS203)。
【0027】
次にステップS204では、式(2)に示すような、独立成分行列ICの統計的独立性指標φ(W)が最大であるか否かを判定する。この判定の結果、最大でない場合には、ステップ205に進む。ステップS205では、変換行列Wtを、式(3),式(4)に従って修正を行い、その後、ステップS203に戻り前述した処理を繰り返し行い、最大となるまで繰り返し更新する。また、この判定の結果、最大である場合には、ステップS206に進む。
【0028】
【数1】

Figure 0003845295
【0029】
ただし、E(x)はxの期待値、tは繰り返しの回数をそれぞれ示している。
【0030】
以上の手順によって変換行列Wが求まる。また、同時に、変換行列Wを式(1)に代入して得られる独立成分行列ICの行要素として、M個の独立成分icn(n=1,2,…,M)が求まる(ステップS206)。
【0031】
再び図1に戻り、104は、成分スコア算出部であり、蓄積されたM個の品質分布情報に基づいて抽出された、品質分布の独立成分行列ICを用いて、各品質分布情報xm(m=1,2,…,M)に含まれる各成分icn(n=1,2,…,M) の大きさ(以下、スコアと呼ぶ)iamnを、式(5)に示すように、両者の内積を算出することによって求める。このとき、独立成分のスコアiamnは、スカラー量となる。
【0032】
【数2】
Figure 0003845295
【0033】
ただし、jは、ベクトルxm、icnのj番目の要素を示す添え字である。
【0034】
図1の105は操業データ入力部であり、前記自動疵検査装置にて検査した鋼板が製造された際のプロセス操業データを入力する。プロセス操業データは、例えば、1本の鋼板を製造する際の平均通板速度や、最高板温度と最低板温度との差など、1本の鋼板全長あたりの代表値となるプロセス操業パラメータの値である。
【0035】
プロセス操業パラメータは数多く存在するが、ここでは、2種類のプロセス操業パラメータP1,P2について、成分抽出部103に蓄積したM個の品質分布情報xm(m=1,2,…,M)に対応するデータ、P1m,P2m(m=1,2,…,M)が入力された場合について、以下記述する。
【0036】
図1の106は相関解析部であり、各独立成分icn(n=1,2,…,M)とプロセス操業データP1,P2との相関の強さSP1n,SP2n(n=1,2,…,M)を、以下のようにして求める。
【0037】
まず、各独立成分icn(n=1,2,…,M)とP1との相関の強さを求めるためには、n=1,2,…,Mの各々に対して、独立成分スコアiamn(m=1,2,…,M)と、プロセス操業データP1m(m=1,2,…,M)との相関係数を算出して、相関評価値SP1n(n=1,2,…,M)とする。
【0038】
次に、各独立成分icn(n=1,2,…,M)とP2との相関の強さについても、P1と同様に、n=1,2,…,Mの各々に対して、独立成分スコアiamn(m=1,2,…,M)と、プロセス操業データP2m(m=1,2,…,M)との相関係数を算出して、相関評価値SP2n(n=1,2,…,M)とする。
【0039】
ここでは、相関係数を算出することによって、相関の強さを求めているので、相関評価値SP1n,SP2nは、−1から1までの値をとることになる。このとき、評価値の絶対値が1に近いほど、成分icnに対する、プロセス操業パラメータの影響が大きく、評価値の絶対値が0に近いほど、成分icnに対する、プロセス操業パラメータの影響が小さいと判断することができる。
【0040】
帯状鋼板5本分のデータに対して、SP1n,SP2n(n=1,2,…,5)の値、すなわち独立成分ic1〜ic5に対する、プロセス操業パラメータP1、P2との相関の強さを求めたところ、図3に示すような結果を得る。
【0041】
ここでは、図3の結果から、相関の評価値の絶対値が0.6以上である、SP13と、SP21が高い相関を示していると判断する。すなわち、独立成分ic3はプロセス操業パラメータP1と相関が高く、独立成分ic1はプロセス操業パラメータP2と相関が高いと判断することができる。
【0042】
ここで、成分スコア算出部104で算出したiamn(m=1,2,…,5、n=1,2,…,5)、すなわち、ここで用いた帯状鋼板5本の品質分布情報に含まれる、成分ic1〜ic5に対する成分スコアを参照することにより、成分ic3に対するスコアが相対的に高い鋼板は、その品質がプロセス操業パラメータP1の影響を受けていることを、また、成分ic1に対するスコアが相対的に高い鋼板は、その品質がプロセス操業パラメータP2の影響を受けていることを、知ることができる。
【0043】
さらに、ここで用いた帯状鋼板5本以外の、新たな鋼板に対しても、品質測定部101にて疵を測定し、品質分布算出部102にて品質分布情報xを算出し、成分スコア算出部104にて、成分ic1〜ic5に対する成分スコアを算出することによって、プロセス操業パラメータが品質に与えている影響を知ることができる。
【0044】
すなわち、ここでは、成分ic3に対するスコアが相対的に高ければ、その品質がプロセス操業パラメータP1の影響を受けていることを、また、成分ic1に対するスコアが相対的に高ければ、その品質がプロセス操業パラメータP2の影響を受けていることを、知ることができる。
【0045】
これに加え、プロセス操業パラメータP1,P2など、各々のプロセス操業パラメータの影響を受けて発生した表面疵が、鋼板上に混在している状況下においても、例えば、いくつか異なるP1の値で操業したときの鋼板における疵個数分布情報に含まれる成分ic3のスコアを評価することによって、P1を要因とする表面疵を減らすための、望ましいプロセス操業パラメータP1の値を求めることができる。
【0046】
ところで、疵の発生要因は、同一工程を経て製造された製品群毎に異なることが多いため、成分抽出部103で蓄積するM個の品質分布情報は、同一工程を経て製造された製品群を対象に蓄積すると、より適切な成分の抽出が可能となる。
【0047】
以上述べた本実施の形態による操業解析装置によれば、品質分布情報に対して、成分分析を行い、各成分のスコアを計算して定量化するため、製品に、様々な発生要因の疵が含まれている状況下においても、その製品の疵発生に影響を及ぼしていると考えられるプロセス操業データを見出すことが容易にできるようになる。
【0048】
本実施の形態では、本発明を鋼板製造プロセスに適用した例を説明したが、他の鉄鋼プロセス、例えば、形鋼、鋼管、線材などの製造プロセスにおいても適用可能である。また、鉄鋼以外の製造プロセス、例えば、アルミ板、銅板などの製造プロセスにおける解析にも本発明を適用することも可能である。
【0049】
なお、本発明の製造プロセスにおける操業解析装置は、複数の機器から構成されるものであっても、1つの機器から構成されるものであってもよい。
【0050】
また、前述した実施の形態は、コンピュータのCPU或いはMPU、RAM、ROM等で構成されるものであり、RAMやROMに記録されたプログラムが動作することで実現される。したがって、前記実施の形態の機能を実現するためのソフトウェアのプログラムコードをコンピュータに供給するための手段、例えばかかるプログラムコードを格納した記憶媒体は本発明の範疇に含まれる。
【0051】
【発明の効果】
以上説明してきたように、本発明によれば、走行中の鋼板の品質を連続して測定する品質測定手段を用いて、鋼板長手方向の品質分布を求めて蓄積し、蓄積した品質分布情報の成分分析を行い、各成分のスコアを計算して定量化し、その成分スコアと製造プロセス操業データとの相関を解析するようにしたので、製品に、様々な発生要因の疵が含まれている状況下においても、その製品の疵発生に影響を及ぼしていると考えられる製造プロセス操業データを見出すことが容易になり、製品品質の向上を図るようにすることができる。
【図面の簡単な説明】
【図1】本発明の実施の形態における操業解析装置の構成を示す図である。
【図2】本実施の形態における操業解析装置の処理動作を説明するためのフローチャートである。
【図3】本実施の形態における操業解析装置の品質分布の独立成分と製造プロセス操業パラメータとの相関の強さの例を示す図である。
【符号の説明】
101 品質測定部
102 品質分布算出部
103 成分抽出部
104 成分スコア算出部
105 操業データ入力部
106 相関解析部[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an operation analysis apparatus and an operation analysis method thereof in a steel plate manufacturing process, and in particular, in a manufacturing process such as a steel process, the process operation affecting the quality of a product from the process operation data and the product quality data. It is suitable for use in finding data.
[0002]
[Prior art]
For the purpose of quality control of steel plate products, various automatic scissors inspection devices that can inspect the surface flaws of a running strip steel plate are being developed and are being introduced to manufacturing sites. Further, many automatic wrinkle inspection devices have not only a function of detecting the presence or absence of wrinkles but also a function of determining the type of detected wrinkles. This is because these soot inspection devices are useful not only for quality control but also for early detection of abnormalities in equipment and operation and for improvement of operation methods.
[0003]
As such an automatic wrinkle inspection apparatus having a wrinkle discrimination function for specifying the type of wrinkle and its inspection method, for example, as disclosed in Japanese Patent Laid-Open No. 54-74793, discrimination is usually performed using a feature amount of a wrinkle shape. Often done.
[0004]
On the other hand, a rolling roll with a foreign material attached to the rolling roll and repeatedly transferred to the steel sheet being rolled has a specific periodicity corresponding to the circumferential length of the roll in the longitudinal direction of the steel sheet. Appears. For example, as disclosed in Japanese Patent Laid-Open No. 6-324005, this rolling roll wrinkle is detected by monitoring a wrinkle that repeatedly occurs at a position that coincides with the rotation period of the rolling roll.
[0005]
[Problems to be solved by the invention]
However, since steel plates are manufactured through complex processes such as casting, rolling, annealing, and surface treatment, the types of flaws detected on the surface of the final product include a mixture of various types. Yes. Therefore, it is not easy to specify the cause of each wrinkle simply by evaluating the individual shape of the wrinkles and the periodicity corresponding to the roll perimeter of the occurrence position.
[0006]
In addition to the periodicity corresponding to the roll perimeter of the shape and location of occurrence, the wrinkles detected on the surface of the final product, for example, are often generated near the tip of the steel plate or the specific process operation data changes There are many cases where wrinkles exist with a specific bias in the longitudinal direction of the steel sheet due to the generation factors of wrinkles, such as when they occur frequently, but as mentioned above, various types of wrinkles are mixed and included Under certain circumstances, it is difficult to identify those traps.
[0007]
The present invention has been made in view of the above-mentioned problems, and in a situation in which a product contains defects caused by various factors, a manufacturing process operation that is considered to affect the occurrence of defects in the product. The purpose is to be able to find data.
[0008]
[Means for Solving the Problems]
Operation analysis system of steel sheet production process of the present invention, in the manufacturing process of the steel sheet, the operation analysis system of steel sheet production process of evaluating the correlation between the manufacturing processes operational data and product quality, its surface the steel sheet during travel From the quality measuring means for continuously detecting wrinkles and outputting the position information of the surface wrinkles, and the position information in the longitudinal direction of the steel flaws outputted by the quality measuring means , the total length of the steel plates is determined by a predetermined number N. In addition, it is divided into sections of equal length and calculated by converting the total number of wrinkles in each section into the number of wrinkles per unit width, or calculating the number of wrinkles that exist within a predetermined width range a distribution calculating means, the quality distribution information of N components the quality distribution calculating means outputs accumulates M times, statistically by ICA quality distribution information matrix of the accumulated M rows and N columns The component extraction have rows decomposed into independent components to have a row of the independent component matrix with M rows and N and component extracting means for obtaining a sequence of independent component matrix, the component extracted by the component extracting means quality distribution using vector, a component score calculation means for calculating by the size of the components included in the quality distribution data row vector obtains the dot product of the two banks vector the a quality distribution data, the steel sheet is measured at said quality measuring means Operation data input means for inputting manufacturing process operation data at the time of manufacture, manufacturing process operation data composed of M pieces of data input by the operation data input means, and each quality calculated by the component score calculation means by calculating the correlation coefficient between component scores of M data to the distribution information, the manufacturing process operational data and the component score By analyzing the correlation, characterized in that it comprises a correlation analysis unit asking you to strength of correlation between the manufacturing process operational data and independent component.
[0009]
The operation analysis method in the steel plate manufacturing process of the present invention is the operation analysis method in the steel plate manufacturing process for evaluating the correlation between the manufacturing process operation data and product quality in the steel plate manufacturing process. From the quality measurement step of detecting the wrinkles continuously and outputting the position information of the surface wrinkles, and the position information of the surface wrinkles in the longitudinal direction of the steel plates output from the quality measurement step, the total length of the steel plates is determined by a predetermined number N. In addition, it is divided into sections of equal length and calculated by converting the total number of wrinkles in each section into the number of wrinkles per unit width, or calculating the number of wrinkles that exist within a predetermined width range Quality distribution information composed of N components output by the distribution calculation step and the quality distribution calculation step is accumulated M times, and the accumulated quality distribution information matrix of M rows and N columns is integrated by independent component analysis. The component extraction step is performed by extracting the components into components independent of each other and extracting the independent component matrix of M rows and N columns of the quality distribution, and the row of the independent component matrix that is the component extracted in the component extraction step Using a vector, a component score calculating step for calculating the size of a component contained in the quality distribution information row vector, which is the quality distribution information, by calculating an inner product of both row vectors, and a steel plate measured in the quality measuring step Operation data input step for inputting manufacturing process operation data when manufactured, manufacturing process operation data composed of M pieces of data input in the operation data input step, and each quality calculated by the component score calculation step The manufacturing process operation data and the component are calculated by calculating a correlation coefficient between the distribution information and a component score composed of M pieces of data. By analyzing the correlation between the core characterized in that it comprises a correlation analyzing step of obtaining the magnitude of correlation between the manufacturing process operational data and independent component.
[0018]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, with reference to the drawings, an embodiment of an operation analysis apparatus in a steel sheet manufacturing process of the present invention will be described.
[0019]
FIG. 1 is a diagram showing the configuration of the operation analysis apparatus of the present embodiment.
In FIG. 1, reference numeral 101 denotes a quality measuring unit, which can use various automatic flaw inspection devices, internal defect detection devices, or material property measuring devices that can continuously measure the quality of a running steel plate.
[0020]
In this embodiment, a case where a surface defect inspection apparatus is used will be described. When the surface flaw inspection apparatus is used, longitudinal position information on the steel sheet of the surface flaw on the steel sheet detected by the apparatus is output.
[0021]
102 is a quality distribution calculating section, from the steel plate longitudinal position information of the surface defects, the steel sheet full length, a constant number N IC number was predetermined and equal divided to the length of the interval, the total in each section Calculate the number of cocoons x. At this time, x is a vector having N IC elements.
[0022]
Usually, since the width dimension of the steel sheet is different for each product, the total number of wrinkles in each section is divided by the width dimension to calculate the number of wrinkles per unit width. Or it is good also considering only the wrinkle within a fixed distance from the range of the width | variety of a fixed magnitude | size, for example, the edge of a steel plate, in the specific position of a steel plate.
[0023]
A component extraction unit 103 accumulates M pieces of quality distribution information x output from the quality distribution calculation unit, and the components of the accumulated M pieces of quality distribution information x m (m = 1, 2,..., M). Perform extraction. As the component extraction method, principal component analysis, independent analysis, etc. can be used, but here, independent component analysis is used to extract independent components included in the number distribution of wrinkles in the longitudinal direction of the steel sheet. Embodiments to be described are described. The independent component analysis method is a signal processing method for decomposing time series continuous data into statistically independent components.
[0024]
Next, with reference to FIG. 2, a method for extracting components independent of each other contained in the number distribution in the longitudinal direction of the steel sheet using the independent component analysis method in the component extraction unit 103 will be described.
[0025]
M quality distribution information x m (m = 1, 2,..., M) used for independent component analysis can be regarded as a matrix X of M rows and N IC columns by arranging M rows with x as a row vector. (Step S201).
[0026]
Next, in order to obtain a transformation matrix W necessary for component calculation, an M-row M-column matrix in which each element is set to a random value is given as an initial matrix W = W 0 (step S202). The quality distribution data X is converted using the conversion matrix W according to the equation (1) to obtain an independent component matrix IC which is a matrix having independent components as row elements (step S203).
[0027]
Next, in step S204, it is determined whether or not the statistical independence index φ (W) of the independent component matrix IC as shown in Expression (2) is the maximum. If the result of this determination is not the maximum, the process proceeds to step 205. In step S205, a transformation matrix W t, make the corrections according to equation (3), Equation (4), then repeats the process described above returns to step S203, repeatedly updated until the maximum. If the result of this determination is the maximum, the process proceeds to step S206.
[0028]
[Expression 1]
Figure 0003845295
[0029]
Here, E (x) represents the expected value of x, and t represents the number of repetitions.
[0030]
The transformation matrix W is obtained by the above procedure. At the same time, M independent components ic n (n = 1, 2,..., M) are obtained as row elements of the independent component matrix IC obtained by substituting the transformation matrix W into equation (1) (step S206). ).
[0031]
Returning to FIG. 1 again, reference numeral 104 denotes a component score calculation unit, which uses the quality distribution independent component matrix IC extracted based on the accumulated M pieces of quality distribution information, and outputs each quality distribution information x m ( m = 1, 2,..., M), the size (hereinafter referred to as score) ia mn of each component ic n (n = 1, 2,..., M) is expressed as shown in Equation (5). It is obtained by calculating the inner product of both. At this time, the score ia mn of the independent component is a scalar quantity.
[0032]
[Expression 2]
Figure 0003845295
[0033]
Here, j is a subscript indicating the j-th element of the vectors x m and ic n .
[0034]
Reference numeral 105 in FIG. 1 denotes an operation data input unit for inputting process operation data when a steel plate inspected by the automatic scissor inspection device is manufactured. The process operation data is, for example, the values of process operation parameters that are representative values for the entire length of one steel sheet, such as the average plate passing speed when manufacturing one steel sheet, and the difference between the maximum and minimum plate temperatures. It is.
[0035]
There are many process operation parameters, but here, for the two types of process operation parameters P1 and P2, M pieces of quality distribution information x m (m = 1, 2,..., M) stored in the component extraction unit 103 are used. The case where the corresponding data, P1 m , P2 m (m = 1, 2,..., M) are input will be described below.
[0036]
Reference numeral 106 in FIG. 1 denotes a correlation analysis unit, and correlation strengths SP1 n and SP2 n (n = 1, n) between each independent component ic n (n = 1, 2,..., M) and the process operation data P1, P2. 2, ..., M) is obtained as follows.
[0037]
First, in order to obtain the strength of correlation between each independent component ic n (n = 1, 2,..., M) and P1, an independent component score is obtained for each of n = 1, 2,. A correlation coefficient between ia mn (m = 1, 2,..., M) and the process operation data P1 m (m = 1, 2,..., M) is calculated, and a correlation evaluation value SP1 n (n = 1) is calculated. , 2, ..., M).
[0038]
Next, regarding the strength of the correlation between each independent component ic n (n = 1, 2,..., M) and P2, similarly to P1, for each of n = 1, 2,. A correlation coefficient between the independent component score ia mn (m = 1, 2,..., M) and the process operation data P2 m (m = 1, 2,..., M) is calculated, and a correlation evaluation value SP2 n ( n = 1, 2,..., M).
[0039]
Here, since the strength of correlation is obtained by calculating the correlation coefficient, the correlation evaluation values SP1 n and SP2 n take values from −1 to 1. At this time, as the absolute value of the evaluation value is closer to 1, the influence of the process operation parameter on the component ic n is larger, and as the absolute value of the evaluation value is closer to 0, the influence of the process operation parameter on the component ic n is smaller. It can be judged.
[0040]
With respect to the data of five strip steel plates, the values of SP1 n , SP2 n (n = 1, 2,..., 5), that is, the correlation with the process operation parameters P1 and P2 for the independent components ic 1 to ic 5 When the strength is obtained, a result as shown in FIG. 3 is obtained.
[0041]
Here, based on the result of FIG. 3, it is determined that SP1 3 and SP2 1 whose absolute value of the correlation evaluation value is 0.6 or more show high correlation. That is, it can be determined that the independent component ic 3 has a high correlation with the process operation parameter P1, and the independent component ic 1 has a high correlation with the process operation parameter P2.
[0042]
Here, ia mn (m = 1, 2,..., 5, n = 1, 2,..., 5) calculated by the component score calculation unit 104, that is, the quality distribution information of the five strip steel plates used here. By referring to the component scores for the components ic 1 to ic 5 included, the steel sheet having a relatively high score for the component ic 3 has its quality influenced by the process operation parameter P1, and the component It can be seen that the steel sheet having a relatively high score for ic 1 has its quality influenced by the process operation parameter P2.
[0043]
Further, for the new steel plates other than the five strip steel plates used here, the quality measuring unit 101 measures wrinkles, the quality distribution calculating unit 102 calculates the quality distribution information x, and calculates the component score. By calculating the component scores for the components ic 1 to ic 5 in the unit 104, it is possible to know the influence of the process operation parameters on the quality.
[0044]
That is, here, if the score for the component ic 3 is relatively high, the quality is affected by the process operation parameter P1, and if the score for the component ic 1 is relatively high, the quality is high. It can be known that the process is affected by the process operation parameter P2.
[0045]
In addition to this, even when surface flaws generated by the influence of each process operation parameter such as process operation parameters P1 and P2 are mixed on the steel plate, for example, operation is performed with some different values of P1. by evaluating the score component ics 3 contained flaws number distribution information in to steel when the can be obtained for reducing the surface defects to cause the P1, the value of the desired process operational parameters P1.
[0046]
By the way, since the occurrence factor of wrinkles is often different for each product group manufactured through the same process, the M quality distribution information accumulated in the component extraction unit 103 is the product group manufactured through the same process. When accumulated in the target, more appropriate components can be extracted.
[0047]
According to the operation analysis apparatus according to the present embodiment described above, the component analysis is performed on the quality distribution information, and the score of each component is calculated and quantified. Even under such circumstances, it becomes easy to find process operation data that is thought to have an effect on the occurrence of defects in the product.
[0048]
In the present embodiment, an example in which the present invention is applied to a steel plate manufacturing process has been described. However, the present invention can also be applied to manufacturing processes of other steel processes, for example, a shape steel, a steel pipe, and a wire rod. Further, the present invention can also be applied to analysis in a manufacturing process other than steel, for example, a manufacturing process of an aluminum plate, a copper plate, or the like.
[0049]
Note that the operation analysis apparatus in the manufacturing process of the present invention may be composed of a plurality of devices or a single device.
[0050]
Further, the above-described embodiment is configured by a CPU or MPU of a computer, a RAM, a ROM, and the like, and is realized by operating a program recorded in the RAM or ROM. Therefore, means for supplying software program codes for realizing the functions of the above-described embodiments to a computer, for example, a storage medium storing such program codes is included in the scope of the present invention.
[0051]
【The invention's effect】
As described above, according to the present invention, the quality measurement means for continuously measuring the quality of the running steel sheet is used to obtain and accumulate the quality distribution in the longitudinal direction of the steel sheet, and the accumulated quality distribution information Since component analysis was performed, the score of each component was calculated and quantified, and the correlation between the component score and manufacturing process operation data was analyzed, so that the product contains a flaw of various occurrence factors Even below, it becomes easy to find manufacturing process operation data that is considered to have an influence on the occurrence of wrinkles of the product, and it is possible to improve the product quality.
[Brief description of the drawings]
FIG. 1 is a diagram showing a configuration of an operation analysis apparatus according to an embodiment of the present invention.
FIG. 2 is a flowchart for explaining a processing operation of the operation analysis apparatus according to the present embodiment.
FIG. 3 is a diagram showing an example of the strength of correlation between an independent component of the quality distribution and the manufacturing process operation parameter of the operation analysis apparatus according to the present embodiment.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 101 Quality measurement part 102 Quality distribution calculation part 103 Component extraction part 104 Component score calculation part 105 Operation data input part 106 Correlation analysis part

Claims (2)

鋼板の製造プロセスにて、前記製造プロセス操業データと製品品質との相関を評価する鋼板製造プロセスにおける操業解析装置において、
前記鋼板が走行中にその表面疵を連続して検出し、当該表面疵の位置情報を出力する品質測定手段と、
前記品質測定手段が出力する表面疵の鋼板長手方向の位置情報から、鋼板全長を、予め定めた一定個数Nで、かつ等しい長さの区間に分けて、各区間での合計疵個数を単位幅当たりの疵個数に換算して算出するか、あるいは所定の幅の範囲内に存在する疵個数を算出する品質分布算出手段と、
前記品質分布算出手段が出力するN個の成分からなる品質分布情報をM回蓄積し、前記蓄積したM行N列の品質分布情報行列を独立成分分析により統計的に互いに独立な成分に分解して成分抽出を行い、品質分布のM行N列の独立成分行列を求める成分抽出手段と、
前記成分抽出手段で抽出された成分である前記独立成分行列の行ベクトルを用いて、前記品質分布情報である品質分布情報行ベクトルに含まれる成分の大きさを両行ベクトルの内積を求めることによって算出する成分スコア算出手段と、
前記品質測定手段にて測定した鋼板が製造された際の製造プロセス操業データを入力する操業データ入力手段と、
前記操業データ入力手段によって入力されたM個のデータからなる製造プロセス操業データと、前記成分スコア算出手段によって算出された各品質分布情報に対しM個のデータからなる成分スコアとの相関係数を算出することによって、前記製造プロセス操業データと前記成分スコアとの相関を解析して前記製造プロセス操業データと独立成分との相関の強さを求める相関解析手段とを備えることを特徴とする鋼板製造プロセスにおける操業解析装置。
In the steel sheet manufacturing process, in the steel sheet manufacturing process operation analysis device for evaluating the correlation between the manufacturing process operation data and product quality,
Quality measuring means for continuously detecting the surface flaw while the steel plate is running and outputting position information of the surface flaw ,
From the position information in the longitudinal direction of the steel sheet surface outputted by the quality measuring means , the total length of the steel sheet is divided into a predetermined number N and sections of equal length, and the total number of steel sheets in each section is determined as a unit width. A quality distribution calculating means for calculating the number of wrinkles calculated by converting to the number of wrinkles per hit, or calculating the number of wrinkles existing within a predetermined width range ;
The quality distribution information composed of N components output from the quality distribution calculating means is accumulated M times, and the accumulated quality distribution information matrix of M rows and N columns is statistically decomposed into independent components by independent component analysis. the component extraction have rows Te, a component extracting means for obtaining an independent component matrix of M rows and N columns of quality distribution,
Using the row vector of the independent component matrix that is the component extracted by the component extraction unit, the size of the component included in the quality distribution information row vector that is the quality distribution information is calculated by obtaining the inner product of both row vectors. Component score calculating means for
Operation data input means for inputting manufacturing process operation data when the steel sheet measured by the quality measuring means is manufactured;
A correlation coefficient between the manufacturing process operation data composed of M pieces of data input by the operation data input means and the component score composed of M pieces of data for each quality distribution information calculated by the component score calculation means. by calculating, the steel sheet characterized by comprising a correlation analysis unit asking you to strength of correlation between the manufacturing process operational data and independent component by analyzing the correlation between the components score and the manufacturing process operational data Operation analysis device in the manufacturing process.
鋼板の製造プロセスにて、前記製造プロセス操業データと製品品質との相関を評価する鋼板製造プロセスにおける操業解析方法において、
前記鋼板が走行中にその表面疵を連続して検出し、当該表面疵の位置情報を出力する品質測定工程と、
前記品質測定工程が出力する表面疵の鋼板長手方向の位置情報から、鋼板全長を、予め定めた一定個数Nで、かつ等しい長さの区間に分けて、各区間での合計疵個数を単位幅当たりの疵個数に換算して算出するか、あるいは所定の幅の範囲内に存在する疵個数を算出する品質分布算出工程と、
前記品質分布算出工程により出力されるN個の成分からなる品質分布情報をM回蓄積し、前記蓄積したM行N列の品質分布情報行列を独立成分分析により統計的に互いに独立な成分に分解して成分抽出を行い、品質分布のM行N列の独立成分行列を求める成分抽出工程と、
前記成分抽出工程で抽出された成分である前記独立成分行列の行ベクトルを用いて、前記品質分布情報である品質分布情報行ベクトルに含まれる成分の大きさを両行ベクトルの内積を求めることによって算出する成分スコア算出工程と、
前記品質測定工程にて測定した鋼板が製造された際の製造プロセス操業データを入力する操業データ入力工程と、
前記操業データ入力工程によって入力されたM個のデータからなる製造プロセス操業データと、前記成分スコア算出工程によって算出された各品質分布情報に対しM個のデータからなる成分スコアとの相関係数を算出することによって、前記製造プロセス操業データと前記成分スコアとの相関を解析して前記製造プロセス操業データと独立成分との相関の強さを求める相関解析工程とを備えることを特徴とする鋼板製造プロセスにおける操業解析方法。
In the steel sheet manufacturing process, in the operation analysis method in the steel sheet manufacturing process for evaluating the correlation between the manufacturing process operation data and product quality,
A quality measuring step of continuously detecting the surface flaw while the steel plate is traveling and outputting position information of the surface flaw ,
From the position information in the longitudinal direction of the steel plate output by the quality measuring step , the total length of the steel plate is divided into a predetermined constant number N and sections of equal length, and the total number of steel sheets in each section is determined as a unit width. A quality distribution calculating step for calculating the number of wrinkles in a range of a predetermined width ,
The quality distribution information composed of N components output in the quality distribution calculation step is accumulated M times, and the accumulated quality distribution information matrix of M rows and N columns is statistically decomposed into independent components by independent component analysis. a component extraction step to have the line component extraction, obtaining an independent component matrix of M rows and N columns of quality distribution,
Using the row vector of the independent component matrix that is the component extracted in the component extraction step, the size of the component included in the quality distribution information row vector that is the quality distribution information is calculated by obtaining the inner product of both row vectors. A component score calculating step,
An operation data input step for inputting manufacturing process operation data when the steel sheet measured in the quality measurement step is manufactured,
A correlation coefficient between manufacturing process operation data consisting of M pieces of data input in the operation data input step and component scores consisting of M pieces of data for each quality distribution information calculated in the component score calculation step. by calculating, the steel sheet characterized by comprising a correlation analyzing step asking you to strength of correlation between the manufacturing process operational data and independent component by analyzing the correlation between the components score and the manufacturing process operational data Operation analysis method in the manufacturing process.
JP2001358067A 2001-11-22 2001-11-22 Operation analysis apparatus and operation analysis method in steel plate manufacturing process Expired - Fee Related JP3845295B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2001358067A JP3845295B2 (en) 2001-11-22 2001-11-22 Operation analysis apparatus and operation analysis method in steel plate manufacturing process

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2001358067A JP3845295B2 (en) 2001-11-22 2001-11-22 Operation analysis apparatus and operation analysis method in steel plate manufacturing process

Publications (2)

Publication Number Publication Date
JP2003154409A JP2003154409A (en) 2003-05-27
JP3845295B2 true JP3845295B2 (en) 2006-11-15

Family

ID=19169311

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2001358067A Expired - Fee Related JP3845295B2 (en) 2001-11-22 2001-11-22 Operation analysis apparatus and operation analysis method in steel plate manufacturing process

Country Status (1)

Country Link
JP (1) JP3845295B2 (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4653526B2 (en) * 2005-03-11 2011-03-16 新日本製鐵株式会社 Quality analysis method, quality analysis apparatus, computer program, and computer-readable storage medium

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3255562B2 (en) * 1994-08-12 2002-02-12 株式会社東芝 Surface inspection equipment
JPH0954611A (en) * 1995-08-18 1997-02-25 Hitachi Ltd Process control equipment
JP2000321141A (en) * 1999-05-17 2000-11-24 Toyo Ink Mfg Co Ltd Method and apparatus for resynthesizing spectral reflectance

Also Published As

Publication number Publication date
JP2003154409A (en) 2003-05-27

Similar Documents

Publication Publication Date Title
JP5471818B2 (en) Method and apparatus for inspecting periodic defects in strip material
JP4862708B2 (en) Deterioration degree diagnosis method, deterioration degree diagnosis device, and deterioration diagnosis program
JP2001153865A (en) Evaluating method and device for damage of metallic material
JP3886865B2 (en) Metal material damage evaluation method and apparatus
CN106156503A (en) A kind of multi-scale entropy characterizing method of anchor system internal flaw distribution
JP2019007944A (en) Segregation detection method and segregation detection device
Nie et al. Baseline-free bridge damage identification using proper orthogonal decomposition and continuous wavelet transform with limited sensors
JP3845295B2 (en) Operation analysis apparatus and operation analysis method in steel plate manufacturing process
JP5083821B2 (en) Corrosion state inspection method for conduit using inspection device with ultrasonic sensor and conduit structure suitable for application of corrosion state inspection method
Zuo et al. Feature-informed machine learning for detecting material deformation and failure in aluminum pipes under bending load using acoustic emission sensors
JP2013111614A (en) Method of detecting chattering of cold rolling mill and device for detecting chattering
JP3892614B2 (en) Equipment and product process abnormality diagnosis method and apparatus
CN118585756B (en) Air quality monitoring method and system based on Internet of things
CN117592149B (en) Steel bridge deck fatigue assessment method, system, equipment and medium based on real bridge monitoring
JP4612585B2 (en) Method for evaluating deformation structure of ferritic steel sheet
JP2003211209A (en) Diagnosis method for rolling mill abnormalities
JP2000266613A (en) Material fracture surface analysis apparatus and method
EP4370897B1 (en) Fatigue evaluation in fibre sample
CN115062679B (en) Intelligent evaluation method and device for pipeline integrity based on detection signals
Haapamäki et al. Data Mining Methods in Hot Steel Rolling for Scale Defect Prediction.
JP2007178292A (en) Destruction risk evaluation device, its evaluation method, recording medium, and program
JP3446007B2 (en) Structure deterioration diagnosis method and structure deterioration diagnosis apparatus
CN113450324A (en) Method and system for analyzing length of internal defect of steel rail
KESSLER et al. Covariance of limit defining pairs (CLDP): A novel approach to establishing detection sensitivity for structural health monitoring data
CN117554762B (en) Transformer insulation part aging model building method, medium and system

Legal Events

Date Code Title Description
A131 Notification of reasons for refusal

Free format text: JAPANESE INTERMEDIATE CODE: A131

Effective date: 20060404

A521 Request for written amendment filed

Free format text: JAPANESE INTERMEDIATE CODE: A523

Effective date: 20060605

TRDD Decision of grant or rejection written
A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

Effective date: 20060808

A61 First payment of annual fees (during grant procedure)

Free format text: JAPANESE INTERMEDIATE CODE: A61

Effective date: 20060818

R151 Written notification of patent or utility model registration

Ref document number: 3845295

Country of ref document: JP

Free format text: JAPANESE INTERMEDIATE CODE: R151

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20090825

Year of fee payment: 3

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20100825

Year of fee payment: 4

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20100825

Year of fee payment: 4

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20110825

Year of fee payment: 5

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20120825

Year of fee payment: 6

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20130825

Year of fee payment: 7

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20130825

Year of fee payment: 7

S531 Written request for registration of change of domicile

Free format text: JAPANESE INTERMEDIATE CODE: R313531

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20130825

Year of fee payment: 7

R350 Written notification of registration of transfer

Free format text: JAPANESE INTERMEDIATE CODE: R350

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20130825

Year of fee payment: 7

S533 Written request for registration of change of name

Free format text: JAPANESE INTERMEDIATE CODE: R313533

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20130825

Year of fee payment: 7

R350 Written notification of registration of transfer

Free format text: JAPANESE INTERMEDIATE CODE: R350

S533 Written request for registration of change of name

Free format text: JAPANESE INTERMEDIATE CODE: R313533

R350 Written notification of registration of transfer

Free format text: JAPANESE INTERMEDIATE CODE: R350

LAPS Cancellation because of no payment of annual fees