US3631697A

US3631697A - Rolling mill workpiece delivery thickness control

Info

Publication number: US3631697A
Application number: US852627A
Authority: US
Inventors: Anthony D Deramo; Andrew W Smith Jr; Frank E Wallace; Robert J Goldbach
Original assignee: Westinghouse Electric Corp
Current assignee: CBS Corp
Priority date: 1969-08-25
Filing date: 1969-08-25
Publication date: 1972-01-04
Anticipated expiration: 1989-01-04
Also published as: FR2059673A1; FR2083299B1; GB1318096A; FR2059673B1; DE2102495C2; GB1330952A; US3631677A; FR2083299A1; DE2041896A1; BE762506A; DE2102495A1; BE755269A

Abstract

A workpiece delivery thickness control is provided for use with a programmed digital process control computer for controlling each operating stand of a rolling mill to improve mill setup relative to at least one selected operational variable, such as stand roll force, for a workpiece of known gauge and grade. For this purpose predetermined ratio comparisons are made between the measured value of the selected variable and the predicted value of that variable, for each stand operation with a workpiece. These ratio comparisons, determined in a predetermined manner in relation to previous rolling experience for each stand relative to the same gauge and grade category of workpiece, are stored in a classified memory location to improve subsequent operation with the same category of gauge and grade workpieces. The ratio comparison after each workpiece is rolled is weighted together in a predetermined manner for each stand operation with the ratio determined from previous rolling experience with a similar workpiece to provide operation control information for improving the stand operation with a subsequent workpiece to provide a desired delivery gauge for that workpiece.

Description

United States Patent m1 3,631,697

[72] Inventors sAntholg' D. Deramo FOREIGN PATENTS wlssv e; V Andrew w. Smith Jr. Pittsburgh; Frank 1,991,484 1 1/1967 Great Britain 72/8 E. Wallace, Irwin; Robert J. Goldbach, Pnmary ExaminerMilton S. Mehr McKggspofl, ll f P Attorneys-F. H. Henson and R. G. Brodahl [211 App]. No. 852,627

'I d A 25 1969 ga 1 24 372 ABSTRACT: A workpiece delivery thickness control is pro- 73 Assignee w fl h Electric Corporation vided for use with a programmed digital process control compm p puter for controlling each operating stand of a rolling mill to improve mill setup relative to at least one selected operational variable, such as stand roll force, for a workpiece of known [54] ROLLING MILL WORKPIECE DELIVERY gauge and grade. For this purpose predetermined ratio com- THICKNESS CONTROL parisons are made between the measured value of the selected 22 Claims,7Drawing Figs. variable and the predicted value of that variable, for each 52 us. Cl 72/8 with wmkpiece These ratio [51] Int Cl Bub 37/00 determined in a predetermined manner in relation to previous rolling experience for each stand relative to the same gauge [50] Field of Search 72/6-12, and grade g y of workpiece are stored in a classified memory locatlon to improve subsequent operation with the [56] References Cited same casgory ofhgaug; and gradilwgrkpiecels. 'Lhe rati? comparison ter eac wor pieceis ro e IS weig te toget er ma UNITED STATES PATENTS predetermined manner for each stand operation with the ratio 1186,20 6/1965 Lufibmokw 72/9 determined from previous rolling experience with a similar workpiece to provide operation control information for improving the stand operation with a subsequent workpiece to provide a desired delivery gauge for that workpiece.

'ROUGHING MILL FINISHING MILL scaswoovm EIQEY' QSX POSITION oerscroa ocrscron 3O Y IO 35 1 I H STRIP CHARACTERISTICS PROCESS CORE MEMORY CONTROL COMPUTER DRUM MEMORY IAIENIED JAN 4 I972 LESS THAN THIS LAST STAND DELIVERY LESS THAN THIS LAST STAND DELIVERY GAUGE OR HEIGHT IN MI LS GAUGE OR HEIGHT IN MI LS SHEEI 2 [IF 5 WORKPIECE cmc (mm AL sumo N OPERATION CORRECTION FACTORS) FIG. 2

WORKPIECE GRADE (UPDATED STAND N OPERATION CORRECTION FACTORS) FIG. 3

ROLLING MILL WORKPIECE DELIVERY THICKNESS CONTROL BACKGROUND OF THE INVENTION The present invention relates to the improved control of a rolling mill, and more particularly to the provision of a scheduled, mill stand setup based upon empirical model equation information.

In the operation of particularly a metal rolling mill having at least one stand, the unloaded roll opening and the speed, for each mill stand, as well as other variables, are predictably set up by a process control computer operative with predetermined model equations to provide a desired workpiece reduction resulting in an on gauge delivery work product from each stand. It may be assumed that the loaded roll opening at a stand equals the stand delivery gauge since there is substantially no elastic workpiece recovery. The predictive set up assumptions may be in error, and certain other mill-operating parameters effect the stand loaded roll operation after setup conditions have been established, such that a stand gauge control system is employed to closely control the stand delivery gauge work product. Recent experience with metal-rolling mills, such as a tandem hot strip mill, has demonstrated that a roll force gauge control system is particularly effective for this purpose. Such a roll force gauge control system employs Hookes Law in controlling the screw down position at a given rolling stand, with the loaded roll opening being substantially the delivery workpiece height H and, under normal rolling conditions, equal to the unloaded roll opening or screwdown position SD plus the determined offset and the mill spring stretch, which is obtained by the dividing of the measured stand roll separating force F by the predetermined mill spring constant M. To embody this rolling principle in a roll force gauge control system, a load cell or other stand roll force detector measures the roll-separating force F. The screwdown position is then controlled to minimize the roll force changes from a reference or setpoint value to thereby hold the loaded roll opening at a substantially constant and desired value. Once the unloaded roll opening for each stand and additionally the stand speed setup are determined by the process control computer for a particular workpiece stand pass, the rolling operation is begun. The respective screwdowns are then continuously controlled to regulate the workpiece delivery gauge from each mill stand.

A more detailed discussion of the theory behind a roll force workpiece gauge control operation can be found in US. Pat. No. 2,726,541 of R. B. Sims. In addition, reference is made to a background information providing article entitled, Automatic Gauge Control for Modern Hot Strip Mills, by .l. W. Wallace which appeared in the Dec. 1967 Iron and Steel Engineer at pages 75 to 86.

It is commercially desirable to provide predictive mill stand setup values, which in addition to providing a better on gauge rolling of particularly the head end of the workpiece strip, also establishes mill-operating conditions which are compatible with the subsequent take over relative to the remainder of the workpiece strip by the attendant and more conventional automatic roll force gauge control system once all of the mill stands become full.

Previously mill operational setup parameters have been set up by a human operator. As the measured rolling mill variables have increased both in number and complexity, a process control computer has been applied to take over the dominant role in determining mill stand setup, with the operator serving as a backup. The process control computer has operated to establish certain mill settings according to a predetermined mathematical equation model. As each workpiece strip or coil is rolled, information is gathered from the various mill operation sensors to improve the setup relative to the rolling of the next workpiece. Such a system has proved satisfactory in that the original predictive setup values based upon the model equations can be adapted to a better mill setup by off line data manipulation determined from the rolling of the workpieces.

In rolling mills operated under control of a digital process control computer, in an effort to provide substantially on gauge delivery strip product from each stand during the rolling of individual workpieces, a force feed forward control system has been provided whereby, for as long as the workpiece grade is the same, the actual roll force for each of the respective stands for the rolling of at least one previous workpiece is utilized to determine whether the general roll force level established by the model equations should be higher or lower as compared to at least this one previous similar alloy grade workpiece strip.

In general, a process control computer includes a memory which contains a stepped sequential logic instruction program for controlling the rolling mill operation, and in addition receives input information regarding the known characteristics of each workpiece strip that is roller, and then monitors the respective stand operational results for the rolling of each workpiece for improving the stored information within its memory. The following is illustrative of the information which enters into the operation of such a control system: The desired delivery gauge and temperature from the last stand is supplied to the computer as known input data, the entry temperature to the first finishing stand is estimated or is deter mined by an entry pyrometer; the entry gauge to each of the finishing stands is known since this is the delivery gauge from the last preceding finishing stand; the entry width to the finishing stands is supplied as input information or can be measured by a suitable width gauge.

After the head end of a given workpiece has passed through each of the finishing stands, such gauge effecting variables relative to the remainder of the workpiece as changes in workpiece temperature, hardness variations caused by hard spots, roll wear and the like are controlled by the conventional roll force gauge control system operative with individual stand roll force sensing load cells as well known to persons skilled in this particular art. These load cells measure and provide the stand actual roll force signals to the screwdown or rollopening regulators which are operative with a reference roll force to determine the adjustments made to the respective stand roll openings as required to deliver a desired workpiece gauge out of each stand of the rolling mill.

In general, a programmed digital process control computer can include a central integrated process control or setup processor, with associated input and output equipment, such as generally described in an article entitled Understanding Digital Computer Process Control," by B. H. Murphy, which appeared in Automation Magazine for Jan. 1965, pages 71 to 76.

A background description of process control computer application for a dynamic operation such as the control of a hot strip tandem rolling mill can be found in an article entitled Programming For Process Control by Paul E. Lego in the Jan. 1965 Westinghouse Engineer at pages 13 to 19, and in another article entitled Computer Program Organization For An Automatically Controlled Rolling Mill" by John S. Deliyannides and Arthur H. Green in the 1966 Iron and Steel Engineer Year Book at pages 328 to 334. An additional article of descriptive interest here is entitled On-line Computer Controls Giant Rolling Mill" by Alonzo F. Kenyon appeared in the Nov. 1965 Westinghouse Engineer at pages 182 to 187.

It is known and understood by persons skilled in this particular art of applying process computer control systems that a combined hardware and software process control system, or an extended special purpose control computer apparatus which is produced when a general purpose digital computer is operated under the control of a predetennined software instruction program, such as illustrated by the functional program flow chart shown in the attached drawings, can also be built using hardware or wired logic programming in view of the recognized general equivalence of a software pro gramming embodiment and a hardware programming embodiment of substantially the same control system. However, when an involved industrial application such as here-described becomes somewhat complex, the economics tend to favor the software approach due to the otherwise greater expense and lack of flexibility when logic circuits, such as the well-known NOR logic circuits, are wired together to provide the desired hardware programming circuit arrangement buildup of such logic circuits to perform the sequential program steps.

The use of a conventional roll force automatic work strip gauge control system for providing a substantially constant workstrip gauge or thickness for the remainder length of the workstrip out of one or more stands of the rolling mill after the head end of the workstrip has threaded all the stands is well known to a person skilled in this particular art. For example, a published article of interest to give a background understanding of the involved concept, can be found in the 1964 Iron and Steel Engineer Yearbook at pages 753 to 762 by John W. Wallace and is entitled Fundamentals of Strip Mill Automatic Gauge Control Systems." Another article of interest appeared in the Mar. 1964 Westinghouse Engineer at pages 34 to 40 by J. W. Wallace and was entitled Strip Mill Automatic Gauge Control Systems.

The use of an on-line digital computer control system requires that one or more model equations relating to the controlled process be stored in the memory unit of the computer to enable predictive operation and control of the process and adaptive control of the process relative to updating information obtained from actual operation of the process. For the example of a rolling mill, to permit a prediction of each stand roll force, relative to a given workpiece having a known grade, a suitable model equation is used to predict the roll force for each stand, and in relation to the desired reduction to be made in each stand the unloaded roll opening is predicted for each stand. This general information is already known by persons skilled in this art and is covered by several publications; for example, in the Iron and Steel Engineering Yearbook for 1962 at pages 587 to 592 is an article dealing with this subject matter, and two more articles can be found in the Iron and Steel Engineering Yearbook for 1965 at pages 461 to 467 and pages 468 to 475. A further publication of interest here to illustrate the rolling mill computer control environment in which the teachings of the present invention could be utilized can be found in the Westinghouse Engineer for Jan. 1969, pages 2 through 8 by John W. Wallace and is entitled Integrated Process Control Rolls Steel More Efficiently.

CROSS REFERENCE TO RELATED APPLICATIONS The present invention is related to the inventions disclosed in copending patent applications Ser. No. 728,469 filed May 13, 1968 and Ser. No. 787,173, filed Dec. 26, 1968 and assigned to the same assignee as the present application.

SUMMARY OF THE INVENTION In accordance with the general principles of the present invention at least one rolling mill stand is under the control of a process control computer for providing a desired delivery workpiece strip gauge or thickness from that stand in relation to stored and weighted information learned from and classified according to the previous rolling of prior similar batches or groups of workpieces. A stand operation control system is provided which takes advantage of stored rolling experience information gained from the previous rolling of prior batches of similar workpieces. Measurements are made during the rolling of each workpiece group to determine whether the general operation level should be higher or lower relative to model equation predicted stand operation values, and from this determination and for subsequent groups of similar workpieces, corrections are determined and stored to compensate when needed for each rolling mill stand operation. The target thickness to be delivered from each stand for subsequent workpiece groups is maintained in this manner better than can be determined from the original schedule calculation using provided model equations.

It is therefore a general object of the present invention to provide a new and improved gauge control system for establishing weighted operation correction information re garding previous rolling of similar groups of workpieces which is stored in classified locations to improve the rolling of subsequent similar groups of workpieces.

A further object of the present invention is to provide a new and improved workpiece gauge or thickness control system for the operational control of at least one mill stand, wherein a prediction of selected setup values is made based upon operational model equations for the known workpiece being rolled and this prediction is then corrected in relation to weighted information stored in a predetermined manner relative to the previous rolling experience with at least one similar workpiece in that same mill stand.

An additional object of the present invention is to provide a new and improved gauge control system for a known workpiece passing through at least one mill stand, wherein operation correction ratios are established between actual stand operating variables in relation to predicted values for said variables, which correction ratios are compared to weighted and previously learned corrections for those same parameters as obtained from previous rolling experience with at least one similar workpiece.

A still further object of the present invention is to provide a new and improved workpiece gauge control system whereby subsequent rolling by a given stand of a known workpiece is responsive to learned information obtained and arranged in a predetermined manner from previous rolling experience with prior similar workpieces to better control the rolling of that known workpiece.

An additional object of the present invention is to provide an improved workpiece delivery gauge or thickness control for a rolling mill, including one or more stands, wherein relative to at least one predetermined predicted mill operation variable for the previous rolling of a similar classification of workpiece, the same variable was measured during actual operation of the rolling mill when the previous workpiece was being rolled and was compared with the predicted value of this same variable to provide a stand operation correction factor; this correction factor was weighted and then stored away in a selected memory location classified by workpiece definition such that the correction factor is available to improve the sub sequent rolling of the same classification workpiece at a later time; in this way, predicted stand operation determining variables, such as roll force and unloaded roll gap settings are continually improved and better tuned relative to the utilized rolling operation model equations.

A different object of the invention is to provide an improved workpiece delivery gauge control for particularly the head end of a workpiece passing through a rolling mill which has at least one stand and wherein at least roll force stand operation correction factors are determined for the predicted operation of each of the stands as compared to the actual operation of said stands, and wherein these correction factors are stored away in accordance with a predetermined classification of the workpiece spectrum desired to be rolled by the mill, for improving model equation predicted operation of each rolling mill stand relative to similar workpieces to be rolled at some future time subsequent to the previous determination of said stand operation correction factors for at least one previous similar workpiece rolling experience.

The above and other objects and features of the invention will become apparent from the following detailed description taken in conjunction with the accompanying drawings, which form a part of this specification.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic showing of a workpiece rolling mill, including the last stand of the roughing mill, the crop shear, and a generalized showing of the finishing mill suitable for operation with the control of the present invention;

FIG. 2 is a showing of the initial storage of stand operation correction factors for a typical stand N to illustrate the classification relative to workpiece gauge and grade;

FIG. 3 shows the storage of operation correction factors for a typical stand N after some actual rolling experience has provided weighted values for many of the workpiece category classified correction factors;

FIG. 4 shows the logic flow chart for the instruction program operative before each workpiece enters the respective rolling mill stands to index the grade and gauge of this next workpiece and to provide an operation correction factor for each stand to adapt the model equation predicted operation variables such as stand roll force.

FIG. 5 shows the logic flow chart for the instruction program operative after a workpiece has entered each of the stands to determine the improved stand operation correction factors for the next succeeding similar workpiece;

FIG. 6 shows the logic flow chart for the instruction program operative after a workpiece has entered each of the stands, which workpiece involves a change of either gauge or grade relative to the previously rolled workpiece to transfer the previous workpiece stand operation correction factors as desired into the computer memory.

FIG. 7 illustrates in general the operating range of the model equations for the control of a typical stand relative to the desired operating spectrum of workpieces to be rolled by that stand to show the realized adaptation in accordance with the present invention.

DESCRIPTION OF A PREFERRED EMBODIMENT With reference to FIG. 1, a portion of a continuous strip rolling mill is shown and designated generally by the reference numeral 10. The last stand R of the roughing mill is followed by the first two stands F1 and F2 and the last stand FN of a finishing mill. Each of the rolling stands includes a pair of work rolls l2 and 14, which are given a model equation predicted unloaded screwdown setting SD to provide a desired strip reduction as a workpiece 16 passes successively through each of the several stands. A set of backup rolls l8 and for each stand provides pressure or force on the work rolls 12 and 14 in response to the operation of a screwdown mechanism 22. The stand roll force is applied through operation of a screwdown motor 24 controlled by a screwdown position regulator 26. Respective stand screwdown position detectors 28 monitor the position setting of the screwdown 22 for each stand by detecting the screwdown position change as indicated by the revolutions of the screwdown motor 24 and transmitting an output signal representative of the screwdown position setting. Following the last finishing stand FN, an X- ray gauge 30 detects the actual delivery gauge of the workpiece and provides a signal proportional thereto. The product of this X-ray determined delivery gauge of the workpiece and the exit speed of the workpiece leaving the last stand FN can be compared with the like product for each of the other stands for determining the mass flow delivery gauge from those other stands. This is compared with the roll force gauge to determine the stand ofiset correction. Associated with each of the respective rolling stands is a roll force sensing load cell 32 which measures the separating roll force FM at its associated respective stand for this purpose.

Between the last roughing stand R and the first finishing stand F1 is a crop shear 23, and a temperature measuring device such as the pyrometer to measure the actual temperature of the head end of the workpiece at a predetermined location near the crop shear 23 where the workpiece is on its way into the first finishing stand Fl. It should be understood that the stand position showing of FIG. 1 is illustrative and in normal practice, the tail end of a given workpiece strip will leave the last roughing stand R before its head end enters the first finishing stand Fl.

Control of the rolling process is determined by a programmed digital process control computer 34 which provides desired correlation between the various rolling mill input and output information in a predetermined manner. The exact mode of control is provided by at least one instruction program operative with the computer to functionally relate the respective input signals or combination of input signals to provide suitable output control signals, which are operative to provide the desired delivery of on-gauge workpiece strip from each stand and the desired operation of the rolling mill. The functional relationship between the respective inputs within the process computer 34 will be set forth in detail hereinafter. Other well-known component apparatus and structure contributing to the proper operation of the rolling mill, and ancillary to the disclosure of the present invention, have been purposely left out of this description for ease of illustration, and would include such items as drive motors, sensing potentiometers, speed controllers and the like.

Accurate online gauge control and regulation of particularly the workpiece head end is achieved by a plurality of model equation relationships operative with the process control computer to provide predicted operation determining setting reference values for each stand, such as a predicted roll force for each stand and from this a corresponding screwdown setting reference signal from the process control computer 34 to the respective stand screwdown position regulators 26 corresponding to each of the rougher stands and the finishing stands. This is initially done in relation to predetermined and empirically determined model equation information for the known characteristics of the workpiece strip.

As the initial operation of the present control, and before a given workpiece strip enters the finishing mill stands, the computer 34 predicts the roll forces for the respective stands of the finishing mill using actual delivery height from the last roughing stand in relation to known roll force and horsepower model equations and known workpiece characteristics, the desired horsepower loading of the respective stands, and in relation to the mill spring for each of those stands and the desired or target delivery gauge out of the last stand of the finishing mill. The screwdown setting for each stand is determined in relation to the calculated or predicted stand roll separating force, and the position regulator for each stand makes the required screwdown adjustment.

There is stored in the computer memory a model equation to calculate the average roll pressure for each stand as required to make the desired reduction in the workpiece strip to be made by that stand; this equation is a follows:

The stand average roll pressure is the analog determined by this equation, where T,, is the nominal workpiece strip tem perature in degrees Fahrenheit for the given stand 11, P is the average pressure across the arc of contact or the resistance to deformation in p.s.i., Hn is the desired delivery gauge out of stand n, Dn is the work roll diameter of stand n, and H(n-1 is the input workpiece strip gauge into stand n. The constants Al through A5 are parameters relating to workpiece grade and are arrived at empirically for the particular workpiece strip alloy or grade involved; these are obtained empirically by rolling many workpiece strips of a known alloy and measuring the stand roll forces for known drafts and known workpiece strip temperature, then by mathematical regression the actual constants are determined by least squares fitting relative to any resulting error. The predicted roll force for each stand is then calculated by the model equation: Fn=[P-.67TPSI(n1).33TPSI(n)] The term TPSl(n-l) is the entry tension to the stand n, the term TPSI(n) is the delivery tension from the stand in, and W is the width of the workpiece strip. A third model equation is used to calculate the desired reduction to be made at stand n in relation to the desired horsepower loading for stand n, as follows:

[TPSI(n)TPSI(n-I)]W (IPSLS) (HLS) 0.151 The term [P812 is the speed of the particular stand n, the term IPSLS is the speed of the last stand and the term HLS is the desired delivery gauge of the workpiece strip from the last stand. it should be noted that each of these equations are in relation to the draft or thickness reduction to be made in the workpiece strip by each particular stand n.

The roll force predictions are developed through the above mathematical model equations stored in the memory of the process control computer 34, and then modified for providing a desired control of the rolling mill in response to the various signals resulting from several operating condition sensing devices such as the respective load cells 32, the stand speed sensing devices 15, the screwdown position detector 28 and so forth.

Once all stands of the mill are full, such that the head end of the workpiece has passed through the last finishing stand FN, the process control computer 34 is then operative in conjunction with a well-known and conventional roll force gauge control system to maintain for the remainder of the workpiece a desired delivery gauge from each stand with the benefit of the rolling operation already being substantially on gauge due to the operation of the here described present control arrangement.

In FIG. 2 there is illustrated for a typical stand N the initially stored operation correction or adaption factors such as provided in memory before any classified gauge and grade category workpiece has passed through the stand for which FIG. 2 applies.

In FIG. 3 there is illustrated for the typical stand N the stored operation correction factors to show the modifications that occur for workpiece delivery gauge and grade combinations that have had some rolling experience with this stand. Where the correction factors have not changed relative to FIG. 2, this could indicate that the model equation was proper for the rolling by this stand of the particular gauge and grade combination workpiece or that no such workpieces have passed through this stand. The full desired spectrum of classified workpiece gauge and grade combinations is shown, and it may in actual practice with a rolling mill be a considerable period of time before a given workpiece combination passes through the mill stand, if it ever does, during the normal commercial rolling practice followed by a rolling mill.

The present control system operates to take the inertia out of the workpiece rolling process, so the workpiece can change, for example a substantial change in grade or the height can change for succeeding workpieces from 50 mils to 250 mils, and not have an undesired delay before the rolling mill operational process can adapt to this new workpiece. The prior art control systems, where the model equation variables had not been adequately manipulated through data regression techniques required the rolling of workpieces once a workpiece change occurred to be able to deliver a more-or-less acceptable height product out of the last stand of the mill.

The control system disclosed in the above-referenced copending patent application Ser. No. 787,l 73 was responsive to the grade and gauge of each succeeding workpiece relative to the present batch of similar grade and gauge workpieces being rolled to see if a change had occurred; if there was a workpiece change in this respect, the control operation went back to a one or initialized value for each stand operation correction factor. Upon a change of workpiece relative to gauge or grade, all the averaged and learned stand operation correction factors for previous workpiece rolling were not saved and returned to a one value again. It should be noted that a workpiece width change or a workpiece per unit draft change was not sensed by this prior control system and can be very important to determined the rolling operation.

The commercial significance of the present invention is that it permits, in a very short time cycle, a given rolling mill to more rapidly roll commercially acceptable work product over a total desired spectrum of work product, in reference to at least one workpiece variable such as gauge or grade, and if desired additional variables such as width and per unit draft. An empirical model equation can be provided to predict approximately the mill stand operation, but it deviates from the actual physical rolling mill operational process; by curve fitting and regression techniques, it can be made to better fit the actual process, but there still remains the need for a better adaptation and correlation of this model with the actual process. By rolling one of each workpiece category in the total desired commercial spectrum this present control system permits substantially improved rolling of all succeeding similar workpieces. The initial gross or loose fit of the model equations is in this way adapted or fitted to the actual process. It is very common for a typical rolling mill to roll under automatic control a rather narrow range of workpiece category groups for a period of time, such as 6 months, and many such mills never roll the total desired commercial spectrum of work products under automatic control because of the difficulty of fitting the model equations to the actual process through inability to obtain data and adequate operational experience relative to many workpiece categories.

In actual practice the model equations are not able to depict the actual rolling process percent, such that the stand operation correction factor here described for each stand will not always remain a one value. By regression techniques and so forth, the model equation can be made to more-or-less approximately fit the actual rolling process. The model equations used to determine the control of a tandem rolling mill have several parameters, which are adjustable through regression after data collection from monitoring of the actual process. By adjusting these parameters to better predict the operation of a particular rolling mill, an effort is made to ac ceptably fit the model equations to the actual rolling mill process. By such modification of the model equations an effort is made to more closely approach a commercially acceptable fit with the actual process such that commercially acceptable work product is thereby obtained. This fitting operation can be very time consuming, and in actual practice it never really ends; typically, the control system supplier agrees upon a reasonable operational specification such that the control system performs satisfactorily for a limited range of workpiece grades such as grades 0, l and 2. Then the customer may try to extend the model equation and control system fit for the rest of the desired workpiece grades from 0 up to 9.

The control system of the present invention improves upon this prior practice by retaining as learned information all of the acceptable past correlation between predicted and actual mill stand operation in the form of credibility indicating stand operation correction factors, and then adapts the model equations in this way in accordance with classified workpiece groupings such as gauge and gade combinations. This can be extended to classify this information according to the various width groupings such as wide, medium and narrow, and selected categories of per unit draft groupings. In this way there is established a correlation between the actual process in relation to the empirical model equation predictions for each workpiece variable, such as grade and gauge and per unit draft and width, as previously mentioned.

The temperature variable included in the model equations is utilized in a different manner, as compared to the above variables, to modify the prediction of stand scheduled settings; a comparison is made of the measured temperatures of succeeding similar workpieces and is used as a multiplying correction to compensate for temperature changes.

in FIG. 7 there is shown an illustration of a typical workpiece product spectrurn required for commercially satisfactory operation of a typical rolling mill. This is shown to be broader than the reasonably obtainable commercially satisfactory operating range of the model equations per se that are available to control such a rolling mill. The stand operation correction factors in accordance with the present invention are effective to extend as shown this range of the model equation operation to include the total desired workpiece product spectrum, for the situation where a sufiicient group of similar classification workpieces has been previously rolled. The temperature compensation in relation to the difference between the present workpiece measured temperature and the previous similar category of workpiece measured temperature is effective as shown to extend partially the range of the model equation operation.

In the prior art rolling practice, the model equation parameters were adjusted and worked on off line. The process control computer logged data for each stand during the actual rolling mill operation, and then this data was studied off line for the desired regression fitting. One set of model equation parameters was provided for each workpiece grade, and was used for the total height range within that grade. It could take 4 or 5 years of substantial effort to even partially fit the model equation to the desire spectrum of work product since a particular mill may only seldom go outside of three grades of work product such that the whole spectrum of work product might not be adequately covered for a considerable period of time.

In accordance with the present invention, if the model equation predicted value of a stand operation variable such as roll force is within 50 percent of the actual stand operation value, this is acceptable and the control system can adapt to the rolling mill operation. If outside 50 percent, then it is necessary to manipulate the model equation through data collection and so forth as previously mentioned. This 50 percent limit is arbitrarily chosen to indicate there is enough intelligence in the model equation for satisfactorily stable operation of the rolling mill. This limit operates to ignore bad data monitored from the operation of the mill and to permit a reasonable response to the process.

A weighting operation takes place regarding information gathered within a workpiece group and in addition another weighting operation takes place regarding information for a plurality of groups. One such weighting is in regard to the number of workpieces within an individual batch or group of workpieces and the other is regarding batches of similar workpieces. For a given batch of similar workpieces, any previously determined stand operation correction factor SCF relative to previously rolled batches of similar workpieces is transferred from drum memory to temporary core memory and is used in whole to modify the model equation predictions of stand settings for the rolling of the first workpiece. The weighting within core memory for the second workpiece then changes to become one-half old SCF and one-half the new SCF determined after rolling the first workpiece. The weighting for the third workpiece then changes to become two-thirds old SCF and one-third new SCF after rolling the second workpiece. The weighting for the fourth workpiece then changes to become three-fourths old SCF and one-fourth new SCF after rolling the third workpiece, and so forth up to an arbitrary limit, such as workpieces where the weighting remains nine-tenths old and one-tenth new information. After rolling all of the workpieces in a given batch, the resulting SCF is transferred from temporary core memory and stored on the drum memory of the computer in a further weighting relation ship determined by the number of such batches of similar workpieces that have been rolled, with the latter weighting being similar to the core memory weighting above described only relative to the number of batches of similar workpieces that have been rolled. In addition there is associated with the drum memory the total number of similar workpieces as well as the number of batches of similar workpieces for which a given classification of SCF information has been stored on the memory drum. A counter is provided to keep track of the number of similar workpieces in each given batch, and before this latter number is destroyed, the drum memory keeps track of the total number of workpieces in each classified workpiece category involved in the past rolling history of each stand of the rolling mill as well as the total number of workpiece batches as an index on the drum memory. For example, if there is information stored on the memory drum for 3,500 groups of a total of 35,000 similar workpieces, then after a given batch of similar category workpieces is completed, the weighted SCF information stored in temporary core would be transferred onto the drum in the weighting relationship of nine-tenths old SCF information already stored on the drum and one-tenth new SCF information, since there were more than 10 groups or batches of SCF information already stored on the drum memory. It should be understood that the number 10 is an arbitrary limit established as adequate for the learning operation to follow typical mill characteristic changes for some particular rolling mill, perhaps for a different rolling mill this limit should have a different value depending upon the particular mill to be controlled.

Thus for the rolling of a new classified workpiece category for each stand, such as would result from a gauge or a grade change, the initial SCF information transferred into core storage is whatever is on drum for the new workpiece category and all of this SCF is utilized to adapt the model equation for the corresponding rolling stand in regard to the first workpiece in this new group. Then for the second workpiece in that same group, the SCF determined for each stand after the first workpiece passes through the mill is combined in temporary core storage as one-half old SCF from drum and one-half new SCF. Then for the third workpiece in same workpiece group, the SCF determined after rolling the second workpiece is combined in temporary core as two-thirds old SCF plus onethird new SCF, and so forth. For all of this same workpiece category group after ten, the core SCF is obtained as ninetenths old SCF and one-tenth new SCF, even if 30 or 40 similar workpieces or more of this same workpiece group are rolled.

After this particular group of workpieces is completed, whatever weighted SCF information is now in core storage for each stand is then transferred onto the drum storage with the second weighting as previously set forth.

If desired, the control technique can include a 10x10 correction factor matrix for the total width range; on the other hand, if three width ranges (or even ten such categories of width are involved) then the storage matrix becomes classified by 10 gauge categories and by 10 grade categories and by the number of width categories, and so forth, per unit draft can be used instead of width, or even as 10 categories of per unit draft in addition to the 10 width categories to make the storage matrix classified according to 10 gauge by 10 grade by 10 width by 10 per unit draft categories.

In FIG. 4, there is shown a logic flow chart to illustrate the instruction program which is entered onto whenever a workpiece advances toward the rolling mill to pass through the respective stands of the rolling mill. The program begins at step 200. At step 202 the known gauge A for the new workpiece having a category of gauge 1 if the desired delivery thickness from the last stand of the rolling mill is less than 50 mils, having a gauge category 2 if the desired delivery thickness from the last stand of the rolling mill is greater than 50 mils and less than 60 mils, having a gauge classification 3 if this delivery thickness is less than 72 mils and greater than 60 mils, having a gauge classification 4 if this delivery thickness is less than 86 mils and greater than 72 mils, having a gauge classification of 5 is this desired delivery thickness is greater than 103 mils and less than 86 mils, a gauge classification of 6 if the delivery gauge is less than mils and greater than 103 mils, a gauge classification of 7 if the last stand delivery gauge is less than mils and greater than 125 mils, a gauge classification of 8 if the last stand delivery gauge is less than mils and greater than 150 mils, a gauge classification of 9 if the last stand delivery gauge is less than 220 mils and greater than 180 mils, and a gauge classification of 10 if the last stand delivery gauge is equal to or greater than 220 mils. At program step 204 there is placed into storage the grade B index of to 9 categories, which are supplied as input information relative to the known characteristics of the workpiece. At program step 206 a determination is made if this next workpiece to be rolled is the same gauge index as the last workpiece. If it is, the program advances to step 208 where a check is made to see if this next workpiece is the same grade as the last workpiece. If either of program steps 206 or 208 indicates a N0 response, the program advances to program step 210 where a transfer from drum memory to core memory is made for the respective stand operation correction factors for the new workpiece of gauge A and grade B. At program step 212 an identification is made of the location in temporary core memory of the previously calculated operation correction factor for each stand for the previous workpiece grouping of gauge C and grade D. At program step 214, there is made a transfer within temporary core memory storage of the operation correction factors for each stand as determined at program step 212 for the old gauge C and grade D workpieces. The program then advances to step 216 where a calculation is made of predicted force for the respective stands of the rolling mill using available model equations. At program step 218, a modification of the predicted force values takes place using the respective stand operation correction factors for the new gauge A and grade B workpiece. At program step 220, the process control computer determines the schedule calculations and mill set up and sequences the rolling mill to roll the new gauge A and grade B workpiece.

In the event that the check provided at each of program steps 206 and 208 indicated that the next workpiece was the same gauge and the same grade as the last workpiece, the instruction program advanced directly to block 216 where the model equation predicted variables were calculated using the modified stand correction factors for the particular gauge and grade workpiece about to enter the mill.

In FIG. 4 the instruction program is seen to provide a bookkeeping function to sense if the next workpiece is the same as the previously rolled workpiece, and if it is, then it is not necessary to access a new stand operation correction factor location in the drum storage matrix for this next workpiece and is not necessary to transfer the previous workpiece information to the temporary core memory. If either grade index or the thickness index changes, a new block of information is taken out of drum storage for determining the roll of the new category of workpiece, and it is necessary to store the old workpiece information into temporary core storage locations. The stand correction factor values regarding the new workpiece are then utilized to adapt the model equation predictions for the rolling of the new workpiece.

In FIG. 5 is shown the flow chart for the instruction program which becomes operative after a first workpiece has threaded all the stands and the sampling of feedback data is obtained on this workpiece. At program step 250, the instruction program is entered. At program step 252, a check is made to see if the number of workpieces NC(A,B) already rolled for the new gauge A and grade B workpiece is greater than 10. If NC (A, B) is greater than It), then program step 254 provides an arbitrary limit of to the number of workpieces. The program advances to step 256 where a new SCF information weighting NEW WTG(A, B) relative to the workpiece number of gauge A and grade B is set equal to one over the number of workpieces NC (A, B). The program then advances to step 258, where the weighting OLD WTG (A, B) for the old SCF information is set equal to the quantity one minus the new SCF information weighting NEW W'I G (A, B). At program step 260, the program begins with the first stand of the rolling mill by setting stand N equal to one. The program then advances to setup 262 where a check is made to see if stand N is in operation. If it is, the repredicted roll force FRP for stand N is calculated as the square root of the quantity I-IM(N1 which is the roll force measured delivery gauge from the previous stand minus HM( N) which is the roll force measured gauge from stand N, divided by the quantity H(N1) which is the predicted delivery gauge from the previous stand minus H(N) which is the predicted delivery gauge from stand N, times the quantity FP(N) which is the model equation predicted force for stand N. When N is the first finishing stand, HM(N1 is the last roughing stand delivery gauge. The instruction program then advances to step 266 where a tentative correction factor TCF(N) is determined for each stand as the measured force for stand N divided by the repredicted force for stand N. At instruction program step 268, a check is made to see if the tentative correction factor TCF (N is greater than 50 percent, and if it is not, the tentative correction factor calculation is considered to be invalid and the program advances to step 270 where N is set as the next succeeding stand and the calculation loops back to step 262. On the other hand, if the check made at step 268 is satisfied, the program advances to step 272 where a check is made to see if the tentative correction factor TCF(N) is less than percent. Again, if this check is not satisfied, the program advances to step 270 as previously described. Steps 268 and 272 provide a validity check on the assumption that if the tentative correction factor is greater or less than an indication of a 50 percent error in the measured force compared to the repredicted force, it is assumed that the information is not acceptable for the purpose of this example. If the checks made at step 268 and 272 are satisfied, the program advances to step 274, where a new stand operation correction factor relative to the new gauge A and grade B workpiece is determined, as the tentative correction factor for stand N times the new information weighting factor as determined at program step 256 for the gauge A and grade B workpiece, plus the old stand correction factor for the gauge A and grade B workpiece (as taken out of the drum memory and determined by previous rolling history of this particular stand relative to the previous groups of the gauge A and grade B workpieces) times the oid information weighting factor for this category of workpiece as determined at instruction step 258. The instruction program then advances to the step 270 where the program repeats for successive stands, until N becomes the stand after the last stand which is an inoperative stand, such that a tentative correction factor and a new stand correction factor is determined for each of the operating stands in the rolling mill. Program step 262 provides a check for stand operation; and for an operating stand the program advances to step 264; for a nonoperating stand the program advances to step 276, where the measured height for the present stand is set equal to the height out of the last previous stand and the program advances to step 278. At program step 278 a check is made to see if stand N is the last stand of the rolling mill; if it is not, the program advances to step 280 where N is advanced by one and the program returns to step 262. On the other hand if the check made in step 278 indicates that N is the last stand of the rolling mill, the program advances to step 282 where a check is made to see if gauge A index of the new workpiece is the same as the gauge index of the previous workpiece. If it is, the program advances to step 284 where a check is made to see if grade B of the new workpiece is the same grade as the previous workpiece. If it is, the program advances to step 286 which is the end of the logic flow chart. On the other hand, ifeither of

steps

282 or 284 indicate a change of gauge or change of grade respectively, the program advances to step 288, where a bid is made for the SCF update program set forth in FIG. 6.

In summary, the FIG. 5 instruction program calculated the information weighting for old SCF information and new SCF information. A check is made to see if each stand is operating, and if not that particular stand is ignored; while for each stand that is operating the stand force is repredicted on the basis of measured gauge. A tentative correction factor for each stand is calculated and limit checked and then the stand correction factors are calculated. The required determination is made to see if a given workpiece is the first workpiece of a new category group, and if it is, a bid is made to do the update operation of the FIG. 6 instruction program.

The update operation instruction program of FIG. 6 is entered at step 300. At step 302, a transfer is made from drum storage of the stand correction factor for stand N relative to old workpiece gauge C and grade D work product for each stand. At step 304, a transfer from drum is made for the weighting WTG (C, D) for the old gauge C and grade D workpiece weighting which is common for all stands. At program step 306, a new weighting NEW WTG (C, D) for each stand relative to the old workpiece gauge C and grade D work product is set equal to one divided by the weighting WTG (C, D) previously stored in drum memory. At step 308, the old information weighting OLD WTG (C, D) relative to gauge C and grade D workpiece is set equal to one minus the new information weighting NEW WTG (C, D) established at step 306. At program step 310 the calculation is started with the last stand of the rolling mill by setting stand N as equal to the last stand. At step 312, the stand operation correction factor SCF (C, D) for stand N relative to gauge C and grade D workpiece is determined as equal to the new information weighting NEW WTG (C, D) determined at instruction step 306 times the SCF (C, D) stored in temporary core plus the old information weighting OLD WTG (C, D) determined at step 308 times the stand correction factor SCF (C, D) for stand N obtained from the drum memory as part of program step 302. The program advances to step 314 where a check is made to see if N is the first stand, indicating that the stand operating correction factors for all stands have been calculated. If it is, then the instruction program advances to step 316 where the new-calculated stand correction factor SCF (C, D) for stand N is stored on the drum memory for the gauge C and the grade D workpiece, and the program advances to step 318, where a check is made to see if more than 9 batches of workpieces of information are involved in the weighting WTG (C, D). If the answer is yes, the program advances to step 320 where the weighting WTG (C, D) is limited to 10 batches of workpieces. On the other hand, if the check made at step 318 is negative, the program advances to step 322 where the weighting WTG (C, D) is set equal to the previous weighting plus one. The program advances to step 324 where the weighting WTG (C, D) is stored on drum memory for gauge C and grade D work product, and the program advances to step 326 where the previous workpiece gauge C and grade D information core storage location is overwritten with the gauge A and grade B information. The total number of similar pieces of gauge C and grade D in the last batch of workpieces is added to the total previously accumulated number of similar workpieces in program step 328. The program ends at step 330. In reference to program step 314 if the check to see if stand N is not one is negative, the program advances to step 315 where N is set equal to N-l and the program advances back through program step 312, where a calculation of the stand operating correction factor is made for the new stand N.

In summary, the update function program of FIG. 6 is operative when a change of workpiece category is identified. It takes the last previous workpiece batch SCF and adds in a weighted manner to the already stored on drum SCF for similar workpieces.

In general, an operational advantage of the present control arrangement is to store away the accumulatively learned stand N correction factor SCF (A, B) information in a classified manner for the improvement of and better model equation adaptation for the future rolling of similar gauge and grade work product. Relative to the rolling of either one of a new gauge and grade category of work product, any stand correction factor previously determined and now stored away for this new gauge and grade work product will be utilized to better control the rolling of this new work product.

Before any previous history rolling information has been built up and put away in storage locations, all the SCF information values are initialized to a one value in each storage location for each stand relative to each gauge and each grade combination category. This initially assumes that the model equation should provide the best available stand settings for the rolling mill. For the first rolling experience relative to a given gauge and grade work product, this initialization schedule calculation provides predictions for the rolling conditions and then permits actual rolling of the work product to take place using the resulting predicted operation variable, such as roll force, for each stand. After a given product of known gauge and grade has been rolled, the measured variable value is compared with the predicted variable value to determine how well the model equation predicted values accomplish the desired workpiece reductions and delivery thickness from such stand or from each pass. Whenever there is a discrepancy, high or low, a stand operation correction factor relative to the ratio of the measured value to the predicted value of the operating variable, such as stand roll force, is stored so that whenever this same gauge and grade category of work product is again to be rolled, this information is available and used, for example, 6 months later, when the same product is rolled. This results in an improvement through a weighted tuning of the rolling operations. The model equations establish the relationships for each stand relative to desired operating variables such as roll force, for desired product thickness reductions and required horsepower on a stand by standard basis. However, the model equations are predicted upon ideal conditions for the rolling operation, and the actual conditions realized in practice can vary somewhat from these ideal conditrons.

It should be understood that stand roll force is readily apparent as a suitable variable to be improved through the stand operan'on learning technique in accordance with the teachings of the present invention. However, the schedule calculation requires the prediction of other operating variables, such as torques and horsepowers and so forth. The teachings of the present invention are suitable relative to any operational variable that is first predicted and can be subsequently measured, such that a comparison can be made between the actual realized value relative to the predicted value. The improvement or operation correction information learned in this manner is classified in a predetermined manner in a storage location within the computer memory.

The control system identifies the gauge and grade category of each incoming workpiece relative to the operation of each stand from supplied input information, such as from punchcards or magnetic tape relative to the passage of each workpiece through the stand or stands of the rolling mill. After rolling several similar workpieces, the stored old SCF value is very effective to converge the stand operation to be productive of desired delivery work product out of each stand. One of the important capabilities of the technique of the present invention is to accurately compensate for inherent stand operational characteristics such as apparatus aging and other things which influence the rolling mill operation. The present stand control technique will adequately follow and correct for these inherent changes in each mill stand operation.

In accordance with the present invention when the work product changes from one grade to another, the following things happened: (I) there is put away in a selected and classified location on the storage drum of the process control computer what was learned about the gauge and grade workpiece category so that later on today or next week or whenever this same product is again rolled, there is available an accumulating improvement of stored model equation correction data. This involves the necessary bookkeeping that determines and retains improvement information about what was learned during a given work product pass or group of passes. (2) Already stored information is taken from drum memory for each stand for the same particular category of work product. (3) There is established the number of workpiece passes or number of correction information pieces in that item that has been previously weighted. (4) Another weighting takes place before the now established stand correction factor for a given work product is again stored in the selected location on the magnetic drum. It is important that proper classification of the stored SCF information takes place so that it is put away where it can be readily located when the same work product is again rolled, having the same finished gauge index and grade relative to a given stand. The present invention is particularly important where the available model equations under certain conditions are not effective to provide the correct predictions, and the resulting errors relative to a previous rolling of a batch of workpieces are corrective adjustment of future predictions by the same model equations relative to a subsequent rolling of a batch of similar workpieces.

For the convenience of a human operator to understand how well a particular stand or stands of the rolling mill is operating, the control computer can readily be programmed to provide a printed table arrangement such as shown in FIG. 3 of individual stand operation correction factors which can be displayed to show where the individual stand operations are requiring correction. This can alert an operator to examine a selected stand of the rolling mill to see why significant or even undesired stand operation correction factors are occurring relative to particular stands identified in this manner. Whenever a change occurs in the pattern of the resulting stored correction factors, this is one indication that something about the rolling mill or a particular stand or stands of the rolling mill has changed.

In accordance with the present invention, the rolling mill control operation involves learning from past rolling experience with similar workpieces and weighting the stand operation corrective influence of available new data relative to already learned and now stored data. In general, after or some other desired number of similar workpieces have been rolled, the individual stand correction factors for a given work product should be pretty well converged onto a substantially correct value; from then on, for successive passes of similar workpieces in a given batch, the weighting of new information relative to old is arbitrarily chosen for establishing and maintaining the desired credibility of the model equation stand control variable predictions.

Many operating conditions change during the performance of a metal-rolling mill that are very difiicult to identify in advance, so there is here provided along term follow technique of whatever happens to the rolling mill. The present invention provides a practical adaptation technique to adjust the available model equations by learning from any actual rolling experience with batches of similar workpieces that does occur, and classifying this learned information such that it can be later recalled when desired for the rolling of other similar workpieces. The resulting operation of the mill is such that startup of rolling relative to a changed different work product occurs in a fraction of the previously required time. The rolling mill operation improves and better tunes itself by this learning procedure. The technique of the present invention enables a more rapid startup in getting online of the given rolling mill relative to any particular work product previously rolled to produce a more commercially acceptable product. The learning technique in this way takes much inertia out of the system and more rapidly converges onto very accurate and very desirable rolling practices.

it should be understood that references herein to temporary core storage and to drum storage are for purposes of illustration. If desired both the temporary storage and the long tenn storage can be provided elsewhere as may be desired.

It should be further understood that it is within the scope of the present invention relative to the operation of some rolling mills, such as a reversing mill or other single stand mill, for the classification of the previously learned operation correction information to be in relation to some other variable than a per stand variable as one of the variables for the learning table matrix. For example, the learning table matrix shown in H0. 3 is classified by workpiece gauge and workpiece grade on a per stand basis. It may instead be desirable to classify the learned information by workpiece gauge and workpiece grade in accordance with a per unit draft variable or some other mill operation variable such as width or the like.

Although the present invention has been shown in connection with a specific embodiment, it should be readily apparent to those skilled in this art that various changes in form and in the arrangement of operational steps may be made to suit particular mill requirements without departing from the spirit and scope of the present invention.

For example, one particular embodiment of the present invention for controlling a rolling mill has been described above; however, another related embodiment would be suitable for the control of some other operating process or device, where a predetermined understanding of the process or device such as one or more operation model equations are established to enable a prediction or attempt to provide a desired functioning of that proces or device which could be related to a monitored actual functioning of the process or device for the purpose of enabling an online weighted learning of information regarding an accumulated history of that actual functioning for the purpose of adaptive improvement thereof.

We claim as our invention: 1. The method of controlling the thickness of a present workpiece passed through at least one stand of a rolling mill, after at least one similar workpiece has been previously passed through said stand and after at least one different workpiece has been passed through said stand subsequent to said one similar workpiece and before said present workpiece, including the steps of establishing a predicted value for a selected operation determining variable for said stand for the rolling of said present workpiece in accordance with a predetermined stand-operating relationship for said stand and in ac cordance with known information about said workpiece,

establishing a modified predicted value of said variable in accordance with previously learned information relating to the actual value of said variable during the previous passage of at least said one similar workpiece through said stand and which previously learned information was retained during the passage of at least said one different workpiece through said stand,

and passing the present workpiece through said stand with the operation of said stand being determined by said modified predicted value of said variable.

2. The method of controlling the operation of each operating stand of a rolling mill relative to present workpiece passed through each said stand, after one similar workpiece has been previously passed through each said stand and after one different workpiece has been passed through each said stand subsequent to said one similar workpiece and before said present workpiece, including the steps of establishing a predicted value for a selected determinative variable of each said stand operation for the rolling of said present workpiece in accordance with a predetermined stand operation model equation for each said stand and in accordance with known information about said workpiece,

establishing a modified predicted value of said variable for each said stand in accordance with previously learned information relating to the previous passage of said one similar workpiece through each said stand and which previously learned information was retained during the passage of said one different workpiece through each said stand,

and passing the present workpiece through each said stand with the operation of each said stand being in accordance with the respective modified predicted value of said varia ble for each said stand.

3. The method of claim 1, with said similar workpiece having at least one of the same gauge index, the same grade, the same width or the same per unit draft index as does the present workpiece.

4. The method of claim 2, with said similar workpiece having at least one predetermined workpiece classification in common with the present workpiece.

5. The method of claim 1, with said operation determining variable being stand roll force.

6. The method of claim 2, with said selected detenninative variable being stand roll force.

The method of claim 1, with said predetermined stand operating relationship being an empirically established model equation.

8. The method of claim 2, with said predetermined stand operation model equation being empirically established through previous monitoring of the actual operation of each said rolling mill stand relative to a plurality of workpieces having different predetermined classification categories.

9. The method of controlling the thickness of a second workpiece to be passed through at least one stand of a rolling mill after a first workpiece similar to the first workpiece has already passed through said one stand, including the steps of accumulating information regarding the actual value of a selected operation determining variable during the actual rolling of said first workpiece by at least said one stand,

establishing an operation correction factor for at least one said stand in accordance with a predetermined relationship between said actual value of said operation determining variable during the rolling of said first workpiece and a predicted value of said variable established prior to the passage of the first workpiece through said one stand,

establishing if a change in a predetermined workpiece category has occurred after the rolling of said first workpiece by said one stand,

establishing a predicted value for said variable for at least said one stand for the rolling of said second workpiece in accordance with a predetermined model equation relative to at least said one stand and with known characteristic information about said second workpiece,

with said predicted value for said variable being modified by said operation correction factor when said change has occurred.

10. The method of controlling the thickness of a second workpiece to be passed through at least one stand of a rolling mill after at least a first workpiece similar to the second workpiece has already passed through said one stand including the steps of accumulating information regarding the actual value of a selected operation determining variable for said one stand during the previous actual rolling of said first workpiece by said one stand,

establishing an operation correction factor for said one stand in accordance with a predetermined relationship between said actual value of said operation determining variable for said one stand during the rolling of said first workpiece and a previously predicted value of said variable for said one stand established prior to the passage of the first workpiece through said one stand,

establishing if a change in the workpieces rolled by said one stand relative to a predetermined workpiece category has occurred after the rolling of said first workpiece by said one stand,

establishing a predicted value for said variable for said one stand for the rolling of said second workpiece in accordance with a predetermined model operation equation relative to said one stand and in accordance with the known characteristic information about said second workpiece,

with said predicted value for said variable for said one stand being modified by said operation correction factor for said one stand when said change in said workpiece category has occurred.

11. The method of controlling the thickness of a first workpiece and a second workpiece successively passed through at least one stand of a rolling mill, including the steps of establishing a predicted operation for said one stand for the rolling of said first workpiece in accordance with a predetermined understanding of the operation of at least said one stand and with known characteristic information about said first workpiece,

passing said first workpiece through said one stand and ac cumulating information regarding the actual operation of said one stand relative to said first workpiece,

establishing an operation correction factor for said one stand in accordance with a predetermined relationship between said predicted operation and said actual operation relative to said first workpiece,

establishing if the first workpiece is similar to said second workpiece in relation to at least one selected characteristic of each said workpiece,

establishing if another workpiece which is not similar to said first workpiece in relation to at least said one selected characteristic has passed through said one stand after the passage of said first workpiece and before the passage of the second workpiece,

establishing a predicted operation for said one stand for the rolling of said second workpiece in accordance with said predetermined understanding and in accordance with known characteristic information about said second workpiece, and

modifying said predicted operation for the rolling of said second workpiece in accordance with said operation correction factor when said another workpiece has passed through said one stand after the first workpiece and before the second workpiece and when said selected characteristic of the second workpiece is substantially the same as said selected characteristic of the first workpiece. l2. Workpiece thickness control apparatus operative with at least one stand of a rolling mill having a pair of rolls for effecting a thickness reduction in each of a first and a second workpiece, the combination of first means operative with a predetermined model equation for establishing a predicted operation of said rolls relative to said first workpiece,

second means responsive to the actual operation of said rolls relative to said first workpiece,

third means for establishing a predetermined relationship between said predicted operation and said actual operation of said rolls relative to said first workpiece,

fourth means for determining the similarity between said first workpiece and said second workpiece relative to at least one of the gauge index and the grade of each said workpiece,

fifth means for determining if another workpiece not similar to said first workpiece relative to at least one of the gauge index and the grade of the first workpiece has passed between said rolls after the first workpiece and before the second workpiece,

said fourth means being operative with said first means for establishing a predicted operation of said rolls relative to said second workpiece in accordance with said model equation and said predetermined relationship when said second workpiece is determined to be similar to said first workpiece and when said another workpiece has passed between said rolls after the first workpiece and before the second workpiece.

13. In a workpiece thickness control system for a rolling mill having a plurality of roll stands and a mechanism operative with each roll stand to control the roll opening through which a present workpiece is passed subsequent to the passage of an earlier similar workpiece, the combination of first means operative with at least one model equation for predicting the operation of each operating roll stand prior to the passage of said earlier workpiece and relative to known information about said earlier workpiece, second means responsive to the actual operation of each operating stand during the passage of said earlier workpiece through the roll stands of the rolling mill,

third means for establishing an operation-controlling relationship for each of said operating roll stands in relation to said predicted operation and to said actual operation of each such stand relative to the passage of said earlier workpiece,

fourth means for weighting said operation-controlling relationship in accordance with the number of workpieces similar to said earlier workpiece that have passed through the stands of the rolling mill prior to the passage of said earlier workpiece,

fifth means for storing said weighted operation-controlling relationship for each of said stands in a memory location classified according to at least one selected characteristic of said earlier workpiece, sixth means for sensing the passage through said roll stands of a workpiece different that said earlier workpiece in relation to at least said one selected characteristic of said earlier workpiece after the passage of said earlier workpiece and prior to the passage of said present workpiece,

with said first means being operative with said model equation and with said weighted operation-controlling relationship for predicting the operation of each operating stand relative to known infonnation about said second workpiece when said different workpiece has been sensed by said sixth means.

14. The control system of claim 13, including seventh means responsive to said present workpiece being different than said earlier workpiece relative to at least said one selected characteristic for causing said first means to predict the operation of each operating stand in accordance with a second weighted operation-controlling relationship previously stored away in a classified location corresponding to at least said one selected characteristic of said present workpiece and determined relative to the actual operation of said stand with a workpiece similar to said present workpiece and passed through said stand prior to the passage of said earlier workpiece.

15. In a workpiece gauge control system including at least one stand for rolling at least first and second successive workpieces of known desired delivery gauge index and known grade, the combination of first means for determining a predicted operation for at least said one stand for rolling said first workpiece in accordance with at least a predetermined model equation relative to said one stand,

second means for sensing the actual operation of at least said one stand in rolling said first workpiece,

third means for determining a stand operation correction in relation to said predicted operation and said actual operation for the rolling of said first workpiece,

with said third means establishing a predetermined weighting of said correction in accordance with the number of workpieces of similar predetermined category as said first workpiece that has already been rolled by said one stand,

fourth means for sensing when a workpiece difierent than said first workpiece in relation to at least said one category has been rolled by at least said one stand after said first workpiece and before said second workpiece,

with said first means determining a predicted operation for at least said one stand for rolling said second workpiece in accordance with said model equation and in accordance with said correction when at least said one category of said second workpiece is similar to said first workpiece and after said different workpiece has been sensed by said fourth means.

16. In a workpiece thickness control apparatus for a rolling mill having at least one stand for rolling respective groups of similar workpieces, the combination of first means for storing a first operation correction for said one stand for each predetermined category of workpiece rolled by said stand and in accordance with the number of workpieces in a given group having a similar workpiece category that has been rolled by said stand,

second means for storing a second operation correction for said one stand for each said workpiece category to be rolled by said one stand and in accordance with the number of groups of a similar workpiece category that have been rolled by said one stand,

third means for determining a predicted operation for the rolling of a first workpiece in accordance with a predetermined model equation for said one stand and at least one known characteristic of the first workpiece,

fourth means for monitoring the actual operation of said one stand during the rolling of said first workpiece,

fifth means for establishing a third operation correction for said one stand in relation to said predicted operation and said actual operation for the rolling of said first workpiece, with said first means being operative to combine said first and third operation corrections into a predetermined weighted correction related to the number of similar workpieces in the group including said first workpiece that have been rolled by said stand and then being operative to store the resultant weighted correction. 17. The control apparatus of claim 12, with said third means subsequently determining a predicted operation for the rolling of a second workpiece similar to the first workpiece.

18. The apparatus of claim 12, with the second means being responsive to a second workpiece to be rolled and having at least said one characteristic different than said first workpiece such that said second means combines said resultant weighted correction and said second operation correction in a predeter mined weighting related to a predetermined number of similar workpieces that have been rolled by said stand.

19. The method of controlling at least one stand of a rolling mill, including the steps of storing a stand operation correction factor for at least said one stand in accordance with each of predetermined classifications of workpieces already rolled by said stand,

modifying each stored stand operation correction factor corresponding to the predetermined classification of each additional workpiece that is rolled by said stand through a comparison of a predicted operation for said stand with the actual operation of said stand relative to each said additional workpiece rolled by said stand, comparing each workpiece about to be rolled by said stand with a selected previous workpiece already rolled by said stand to determine if the workpiece about to be rolled has a different predetermined classification than did said selected previous workpiece already rolled by the stand,

and predicting the operation of said stand in relation to each workpiece about to be rolled in accordance with the stored stand operation correction factor corresponding to the classification of the workpiece about to be rolled. 20. The method of controlling at least one stand of a rolling mill, including the steps of storing stand operation corrections for at least said one stand in respective memory locations classified in accordance with at least one of the gauge index and the grade of each workpiece already rolled by said stand,

modifying the stored stand operation corrections corresponding to at least one of the gauge index and the grade of each additional workpiece rolled by said stand through a comparison of a predicted operation for said stand with the actual operation of said stand relative to each said additional workpiece rolled by said stand,

comparing each workpiece about to be rolled by said stand with a selected previous workpiece already rolled by said stand to determine if the workpiece about to be rolled has at least one of a different gauge index and a different grade than did said selected previous workpiece already rolled by the stand,

and predicting the operation of said stand in relation to each workpiece about to be rolled in accordance with the stored stand operation correction factor corresponding to at least one of the gauge index and the grade of the workpiece about to be rolled.

21. The method of controlling a product operation and including the steps of storing operation corrections in respective memory locations classified in accordance with at least one predetermined category of each monitored product already subjected to said product operation and relative to a comparison between a predicted operation with each product and a resulting actual operation with that same product comparing each product about to be subjected to said product operation with a selected previous product already subjected to said product operation to determine if the product about to be subjected to said product operation is different regarding at least said one predetermined category,

and predicting said product operation in relation to each product about to be subjected to said product operation in accordance with the stored operation correction corresponding to at least said one predetermined category of the latter product.

22. The method of controlling a predetermined operation relative to a plurality of products and including the steps of establishing a predicted value for a variable determining said operation in accordance with a predetermined understanding about said operation and in accordance with

Claims

1. The method of controlling the thickness of a present workpiece passed through at least one stand of a rolling mill, after at least one similar workpiece has been previously passed through said stand and after at least one different workpiece has been passed through said stand subsequent to said one similar workpiece and before said present workpiece, including the steps of establishing a predicted value for a selected operation determining variable for said stand for the rolling of said present workpiece in accordance with a predetermined standoperating relationship for said stand and in accordance with known information about said workpiece, establishing a modified predicted value of sAid variable in accordance with previously learned information relating to the actual value of said variable during the previous passage of at least said one similar workpiece through said stand and which previously learned information was retained during the passage of at least said one different workpiece through said stand, and passing the present workpiece through said stand with the operation of said stand being determined by said modified predicted value of said variable.

2. The method of controlling the operation of each operating stand of a rolling mill relative to present workpiece passed through each said stand, after one similar workpiece has been previously passed through each said stand and after one different workpiece has been passed through each said stand subsequent to said one similar workpiece and before said present workpiece, including the steps of establishing a predicted value for a selected determinative variable of each said stand operation for the rolling of said present workpiece in accordance with a predetermined stand operation model equation for each said stand and in accordance with known information about said workpiece, establishing a modified predicted value of said variable for each said stand in accordance with previously learned information relating to the previous passage of said one similar workpiece through each said stand and which previously learned information was retained during the passage of said one different workpiece through each said stand, and passing the present workpiece through each said stand with the operation of each said stand being in accordance with the respective modified predicted value of said variable for each said stand.

6. The method of claim 2, with said selected determinative variable being stand roll force.

7. The method of claim 1, with said predetermined stand operating relationship being an empirically established model equation.

9. The method of controlling the thickness of a second workpiece to be passed through at least one stand of a rolling mill after a first workpiece similar to the first workpiece has already passed through said one stand, including the steps of accumulating information regarding the actual value of a selected operation determining variable during the actual rolling of said first workpiece by at least said one stand, establishing an operation correction factor for at least one said stand in accordance with a predetermined relationship between said actual value of said operation determining variable during the rolling of said first workpiece and a predicted value of said variable established prior to the passage of the first workpiece through said one stand, establishing if a change in a predetermined workpiece category has occurred after the rolling of said first workpiece by said one stand, establishing a predicted value for said variable for at least said one stand for the rolling of said second workpiece in accordance with a predetermined model equation relative to at least said one stand and with known characteristic information about said second workpiece, with said predicted value for said variable being modified by said operation correction factor when said change has occurred.

10. The method of cOntrolling the thickness of a second workpiece to be passed through at least one stand of a rolling mill after at least a first workpiece similar to the second workpiece has already passed through said one stand, including the steps of accumulating information regarding the actual value of a selected operation determining variable for said one stand during the previous actual rolling of said first workpiece by said one stand, establishing an operation correction factor for said one stand in accordance with a predetermined relationship between said actual value of said operation determining variable for said one stand during the rolling of said first workpiece and a previously predicted value of said variable for said one stand established prior to the passage of the first workpiece through said one stand, establishing if a change in the workpieces rolled by said one stand relative to a predetermined workpiece category has occurred after the rolling of said first workpiece by said one stand, establishing a predicted value for said variable for said one stand for the rolling of said second workpiece in accordance with a predetermined model operation equation relative to said one stand and in accordance with the known characteristic information about said second workpiece, with said predicted value for said variable for said one stand being modified by said operation correction factor for said one stand when said change in said workpiece category has occurred.

11. The method of controlling the thickness of a first workpiece and a second workpiece successively passed through at least one stand of a rolling mill, including the steps of establishing a predicted operation for said one stand for the rolling of said first workpiece in accordance with a predetermined understanding of the operation of at least said one stand and with known characteristic information about said first workpiece, passing said first workpiece through said one stand and accumulating information regarding the actual operation of said one stand relative to said first workpiece, establishing an operation correction factor for said one stand in accordance with a predetermined relationship between said predicted operation and said actual operation relative to said first workpiece, establishing if the first workpiece is similar to said second workpiece in relation to at least one selected characteristic of each said workpiece, establishing if another workpiece which is not similar to said first workpiece in relation to at least said one selected characteristic has passed through said one stand after the passage of said first workpiece and before the passage of the second workpiece, establishing a predicted operation for said one stand for the rolling of said second workpiece in accordance with said predetermined understanding and in accordance with known characteristic information about said second workpiece, and modifying said predicted operation for the rolling of said second workpiece in accordance with said operation correction factor when said another workpiece has passed through said one stand after the first workpiece and before the second workpiece and when said selected characteristic of the second workpiece is substantially the same as said selected characteristic of the first workpiece.

12. Workpiece thickness control apparatus operative with at least one stand of a rolling mill having a pair of rolls for effecting a thickness reduction in each of a first and a second workpiece, the combination of first means operative with a predetermined model equation for establishing a predicted operation of said rolls relative to said first workpiece, second means responsive to the actual operation of said rolls relative to said first workpiece, third means for establishing a predetermined relationship between said predicted operation and said actual operation of said rolls relative to said first workpiece, fourth means for deterMining the similarity between said first workpiece and said second workpiece relative to at least one of the gauge index and the grade of each said workpiece, fifth means for determining if another workpiece not similar to said first workpiece relative to at least one of the gauge index and the grade of the first workpiece has passed between said rolls after the first workpiece and before the second workpiece, said fourth means being operative with said first means for establishing a predicted operation of said rolls relative to said second workpiece in accordance with said model equation and said predetermined relationship when said second workpiece is determined to be similar to said first workpiece and when said another workpiece has passed between said rolls after the first workpiece and before the second workpiece.

13. In a workpiece thickness control system for a rolling mill having a plurality of roll stands and a mechanism operative with each roll stand to control the roll opening through which a present workpiece is passed subsequent to the passage of an earlier similar workpiece, the combination of first means operative with at least one model equation for predicting the operation of each operating roll stand prior to the passage of said earlier workpiece and relative to known information about said earlier workpiece, second means responsive to the actual operation of each operating stand during the passage of said earlier workpiece through the roll stands of the rolling mill, third means for establishing an operation-controlling relationship for each of said operating roll stands in relation to said predicted operation and to said actual operation of each such stand relative to the passage of said earlier workpiece, fourth means for weighting said operation-controlling relationship in accordance with the number of workpieces similar to said earlier workpiece that have passed through the stands of the rolling mill prior to the passage of said earlier workpiece, fifth means for storing said weighted operation-controlling relationship for each of said stands in a memory location classified according to at least one selected characteristic of said earlier workpiece, sixth means for sensing the passage through said roll stands of a workpiece different than said earlier workpiece in relation to at least said one selected characteristic of said earlier workpiece after the passage of said earlier workpiece and prior to the passage of said present workpiece, with said first means being operative with said model equation and with said weighted operation-controlling relationship for predicting the operation of each operating stand relative to known information about said second workpiece when said different workpiece has been sensed by said sixth means.

15. In a workpiece gauge control system including at least one stand for rolling at least first and second successive workpieces of known desired delivery gauge index and known grade, the combination of first means for determining a predicted operation for at least said one stand for rolling said first workpiece in accordance with at least a predetermined model equation relative to said one stand, second means for sensing the actual operation of at least said one stand in rolling said first workpiece, thIrd means for determining a stand operation correction in relation to said predicted operation and said actual operation for the rolling of said first workpiece, with said third means establishing a predetermined weighting of said correction in accordance with the number of workpieces of similar predetermined category as said first workpiece that has already been rolled by said one stand, fourth means for sensing when a workpiece different than said first workpiece in relation to at least said one category has been rolled by at least said one stand after said first workpiece and before said second workpiece, with said first means determining a predicted operation for at least said one stand for rolling said second workpiece in accordance with said model equation and in accordance with said correction when at least said one category of said second workpiece is similar to said first workpiece and after said different workpiece has been sensed by said fourth means.

16. In a workpiece thickness control apparatus for a rolling mill having at least one stand for rolling respective groups of similar workpieces, the combination of first means for storing a first operation correction for said one stand for each predetermined category of workpiece rolled by said stand and in accordance with the number of workpieces in a given group having a similar workpiece category that has been rolled by said stand, second means for storing a second operation correction for said one stand for each said workpiece category to be rolled by said one stand and in accordance with the number of groups of a similar workpiece category that have been rolled by said one stand, third means for determining a predicted operation for the rolling of a first workpiece in accordance with a predetermined model equation for said one stand and at least one known characteristic of the first workpiece, fourth means for monitoring the actual operation of said one stand during the rolling of said first workpiece, fifth means for establishing a third operation correction for said one stand in relation to said predicted operation and said actual operation for the rolling of said first workpiece, with said first means being operative to combine said first and third operation corrections into a predetermined weighted correction related to the number of similar workpieces in the group including said first workpiece that have been rolled by said stand and then being operative to store the resultant weighted correction.

17. The control apparatus of claim 12, with said third means subsequently determining a predicted operation for the rolling of a second workpiece similar to the first workpiece.

18. The apparatus of claim 12, with the second means being responsive to a second workpiece to be rolled and having at least said one characteristic different than said first workpiece such that said second means combines said resultant weighted correction and said second operation correction in a predetermined weighting related to a predetermined number of similar workpieces that have been rolled by said stand.

19. The method of controlling at least one stand of a rolling mill, including the steps of storing a stand operation correction factor for at least said one stand in accordance with each of predetermined classifications of workpieces already rolled by said stand, modifying each stored stand operation correction factor corresponding to the predetermined classification of each additional workpiece that is rolled by said stand through a comparison of a predicted operation for said stand with the actual operation of said stand relative to each said additional workpiece rolled by said stand, comparing each workpiece about to be rolled by said stand with a selected previous workpiece already rolled by said stand to determine if the workpiece about to be rolled has a different predetermined classification than did said selected previous workpiece already rolled by the stand, and predicting the operation of said stand in relation to each workpiece about to be rolled in accordance with the stored stand operation correction factor corresponding to the classification of the workpiece about to be rolled.

20. The method of controlling at least one stand of a rolling mill, including the steps of storing stand operation corrections for at least said one stand in respective memory locations classified in accordance with at least one of the gauge index and the grade of each workpiece already rolled by said stand, modifying the stored stand operation corrections corresponding to at least one of the gauge index and the grade of each additional workpiece rolled by said stand through a comparison of a predicted operation for said stand with the actual operation of said stand relative to each said additional workpiece rolled by said stand, comparing each workpiece about to be rolled by said stand with a selected previous workpiece already rolled by said stand to determine if the workpiece about to be rolled has at least one of a different gauge index and a different grade than did said selected previous workpiece already rolled by the stand, and predicting the operation of said stand in relation to each workpiece about to be rolled in accordance with the stored stand operation correction factor corresponding to at least one of the gauge index and the grade of the workpiece about to be rolled.

21. The method of controlling a product operation and including the steps of storing operation corrections in respective memory locations classified in accordance with at least one predetermined category of each monitored product already subjected to said product operation and relative to a comparison between a predicted operation with each product and a resulting actual operation with that same product comparing each product about to be subjected to said product operation with a selected previous product already subjected to said product operation to determine if the product about to be subjected to said product operation is different regarding at least said one predetermined category, and predicting said product operation in relation to each product about to be subjected to said product operation in accordance with the stored operation correction corresponding to at least said one predetermined category of the latter product.

22. The method of controlling a predetermined operation relative to a plurality of products and including the steps of establishing a predicted value for a variable determining said operation in accordance with a predetermined understanding about said operation and in accordance with known information regarding a present product about to be subjected to said operation, establishing a modified predicted value of said variable in accordance with previously learned information relating to the actual value of said variable when a previous product was subjected to said operation and which previously learned information was retained while a later and different product was subjected to said operation, and subjecting said present product to said operation in accordance with said modified predicted value of said variable.