简介:Loadingdistributionforheavyplatemillistofindoptimalcontrolsolutionsunderthegrantedperformanceindicatorsandconstraintsincludingmillcapacityandhypothesisofrollingmodels.Thesolutionsarequitedifferentfordifferentperformanceindicators.Inthearticle,theperformanceindicatorsandsequentialquadraticprogramming(SQPforshortbelow)methodsemployedin5000mmheavyplatemillofBaoSteelarepenetratinglyanalyzed.Generally,theSQPmethodisaneffectiveandfastwaytosolvethenonlinearprogrammingproblemswithsmallormediumscaleconstraints.Earlyin1976,HanputforwardtheSQPmethodforthefirsttimeandPowellmadeitperfectandaccomplishedthealgorithmin1977.Infact,SQPmethodwastoturnanonlinearprogrammingproblemtoaseriesofsubsetofquadraticprogrammingproblems.Inthealgorithm,eachiterationstepistosolveonequadraticprogrammingproblem.Theoptimalsolutionswillbegraduallyapproachedafterquadraticprogrammingproblemsweretotallysolved.Whensolvingthequadraticprogrammingproblem,theactivesetstrategywereemployedwhichturnedtheconstrainedquadraticprogrammingproblemtounconstrainedquadraticprogrammingproblem.Theactivesetstrategymadethewholequadraticprogrammingproblembesolvedbyaleastsquareproblem.Andfinally,thematrixoftheleastsquareproblemwouldbedecomposedbyQmatrixandRmatrix.AfterQmatrixandRmatrixwereobtained,theoptimalsolutionswouldbefinallyfound.Forloadingdistribution,theperformanceindicatorswerecomposedbyplateshapeanddraftofeachpass.Plateshapeisrepresentedbyrollingforcegraduallyreducedpassbypasswithatunablefactor.Themillcapacityisanotherperformanceindicatorrepresentedbydraftofeachpass.Forheavyplatemill,themillcapacityhereisthemotormoment.Forheavydraft,themotorwouldbeoverloadedespeciallyforthefirstseveralpasses;forsmalldraft,themotorwouldbeloadedslightly.Allthesewouldnotbepermittedtohappenwhen