ebook img

Machine Learning in Industry PDF

202 Pages·2021·5.785 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Machine Learning in Industry

Management and Industrial Engineering Shubhabrata Datta J. Paulo Davim   Editors Machine Learning in Industry Management and Industrial Engineering SeriesEditor J.PauloDavim,DepartmentofMechanicalEngineering,UniversityofAveiro, Aveiro,Portugal This series fosters information exchange and discussion on management and industrial engineering and related aspects, namely global management, organiza- tional development and change, strategic management, lean production, perfor- mance management, production management, quality engineering, maintenance management, productivity improvement, materials management, human resource management, workforce behavior, innovation and change, technological and organizational flexibility, self-directed work teams, knowledge management, organizational learning, learning organizations, entrepreneurship, sustainable management, etc. The series provides discussion and the exchange of information on principles, strategies, models, techniques, methodologies and applications of management and industrial engineering in the field of the different types of organizational activities. It aims to communicate the latest developments and thinkinginwhatconcernsthelatestresearchactivityrelatingtoneworganizational challengesandchangesworld-wide.Contributionstothisbookseriesarewelcome on all subjects related with management and industrial engineering. To submit a proposal or request further information, please contact Professor J. Paulo Davim, BookSeriesEditor,[email protected] Moreinformationaboutthisseriesathttp://www.springer.com/series/11690 · Shubhabrata Datta J. Paulo Davim Editors Machine Learning in Industry Editors ShubhabrataDatta J.PauloDavim DepartmentofMechanicalEngineering DepartmentofMechanicalEngineering SRMInstituteofScienceandTechnology UniversityofAveiro Chennai,TamilNadu,India Aveiro,Portugal ISSN2365-0532 ISSN2365-0540 (electronic) ManagementandIndustrialEngineering ISBN978-3-030-75846-2 ISBN978-3-030-75847-9 (eBook) https://doi.org/10.1007/978-3-030-75847-9 ©TheEditor(s)(ifapplicable)andTheAuthor(s),underexclusivelicensetoSpringerNature SwitzerlandAG2022 Thisworkissubjecttocopyright.AllrightsaresolelyandexclusivelylicensedbythePublisher,whether thewholeorpartofthematerialisconcerned,specificallytherightsoftranslation,reprinting,reuse ofillustrations,recitation,broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,and transmissionorinformationstorageandretrieval,electronicadaptation,computersoftware,orbysimilar ordissimilarmethodologynowknownorhereafterdeveloped. Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse. Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor theeditorsgiveawarranty,expressedorimplied,withrespecttothematerialcontainedhereinorforany errorsoromissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictional claimsinpublishedmapsandinstitutionalaffiliations. ThisSpringerimprintispublishedbytheregisteredcompanySpringerNatureSwitzerlandAG Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland Preface Machinelearning(ML)isamethodfortrainingcomputersormakingthecomputer learn automatically from the supplied information or data. Different methods of machine learning originate from the nature and follow the principle of biological learning.TheapplicationsofMLintherealworldareincreasingfastandencompass ourdailylifewithoutourknowledge.Themanufacturingandotherindustrialsectors have also started using the ML in their plants effectively. With the advent of the conceptofIndustry4.0,thepaceofapplicationofMLincoreindustrieswillonly increase. The introductory chapter of this book describes the basic concepts of the most populartechniquesofmachinelearning.Itexplainsthedifferentclassesofmachine learning approaches and describes the statistical and artificial intelligence-based machinelearningtechniquesinbrief.Thetechniqueslikedecisiontree,linearregres- sion,leastsquaremethod,artificialneuralnetwork,clusteringtechniques,anddeep learningarediscussed. Practicalexamplesofapplicationsandcasestudiesusingpracticalindustrialprob- lemsorproblemsrelevanttotheindustriesaredescribedintherestofthechapters, i.e.,ChaptersNeuralNetworkModelIdentificationStudiestoPredictResidualStress ofaSteelPlateBasedonaNon-destructiveBarkhausenNoiseMeasurement–Perfor- manceImprovementinHotRollingProcesswithNovelNeuralArchitecturalSearch. In the Chapter Neural Network Model Identification Studies to Predict Residual Stress of a Steel Plate Based on a Non-destructive Barkhausen Noise Measure- ment,aML-basedprocedureusingartificialneuralnetworkandgeneticalgorithmis proposedforpredictingtheresidualstressinsteel.ChapterDataDrivenOptimiza- tionofBlastFurnaceIronMakingProcessUsingEvolutionaryDeepLearningdeals with modeling the complicated process of iron and steel making. Parameters like burdendistribution,oxygenenrichment,productivityimprovement,compositionof thetopgas,andmanyotherparametersinblastfurnacecontroltheproductivityina complexway.Inthischapter,machinelearningapproachesareemployedtomodel theblastfurnaceprocessandevolutionaryalgorithmsareemployedtooptimizethe process. In the Chapter A Brief Appraisal of Machine Learning in Industrial Sensing Probes,theauthordescribesthemethodtoimplementMLinthedigitalcontroland v vi Preface monitoring systems in industries. How the sensors can be used effectively for the integrationofindustrialdataforstandardMLisdiscussed.TheapplicationofML for searching the root cause of sliver defects in the cold rolling mill is described intheChapterMiningtheGenesisofSliverDefectsThroughRoughandFuzzySet Theories.Atypicalcaseisdescribedwherethegenesisofthedefectisanalyzedusing twomethods,viz.roughsetandfuzzysettheories.Thischaptershowshowtherough settheorycanbeusedtoselecttheimportantvariablestowhichthecauseofthedefect canbeattributed.Therulescreatedfromthedataareusedinthefuzzyframework fordevelopingapredictivemodel.AnoverviewoftheapplicationofMLmethods in the field of materials science with respect to materials—processes—knowledge formalization is provided in the Chapter Machine Learning Studies in Materials Science. ML can be used for developing surrogate models or metamodels, which can replace complex analytical models and numerical simulations for optimization, sensitivityanalysisanduncertaintyquantification.InChapterAccurate,Real-Time Replication of Governing Equations of Physical Systems with Transpose CNNs —forIndustry4.0andDigitalTwins,Convolutional-NN-likearchitecturesareused assurrogatemodelsontwodifferentapplicationsofreducedNavier–Stokesequations containinghighnonlinearitiesandabruptdiscontinuities. Deep learning is the latest paradigm of machine learning. The Chapter Deep LearninginVision-BasedAutomatedInspection:CurrentStateandFutureProspects ofthebookevaluatestheapplicationofthetechniquesforvision-basedautomated inspection.Hereadetailed discussiononthemeritsand demeritsof deep learning for automated inspection tasks in industries are made. The ninth and last chapter proposes a novel algorithm to design multi-layered feed forward neural networks withparsimonyaswellasaccuracyusingmulti-objectiveoptimization. It is quite evident that the authors of the present book covered various aspects and applications of machine learning relevant to the industry. The editors express their sincere gratitude to the authors for their excellent contributions. The editors also express their thanks to all the reviewers who have contributed immensely to theimprovementofthequalityofthechapters.Boththeeditorsaregratefultotheir colleagues,friends,andfamilymembers.TheeditorsalsoacknowledgetheSpringer teamfortheirexcellentworktowardsshapingthecompilationbeautifully. Chennai,India ShubhabrataDatta Aveiro,Portugal J.PauloDavim Contents FundamentalsofMachineLearning ................................. 1 A.VinothandShubhabrataDatta NeuralNetworkModelIdentificationStudiestoPredictResidual StressofaSteelPlateBasedonaNon-destructiveBarkhausen NoiseMeasurement ................................................ 29 TeroVuolio,OlliPesonen,AkiSorsa,andSuviSanta-aho Data-DrivenOptimizationofBlastFurnaceIronMakingProcess UsingEvolutionaryDeepLearning .................................. 47 BashistaKumarMahanta,RajeshJha,andNirupamChakraborti ABriefAppraisalofMachineLearninginIndustrialSensingProbes ... 83 R.Biswas MiningtheGenesisofSliverDefectsThroughRoughandFuzzy SetTheories ...................................................... 97 ItishreeMohanty,ParthaDey,andShubhabrataDatta MachineLearningStudiesinMaterialsScience ...................... 121 BarbaraMrzygłód,KrzysztofRegulski,andAndrzejOpalin´ski Accurate, Real-Time Replication of Governing Equations ofPhysicalSystemswithTransposeCNNs—forIndustry4.0 andDigitalTwins .................................................. 139 HritikNarayanandAryaK.Bhattacharya DeepLearninginVision-BasedAutomatedInspection:Current StateandFutureProspects ......................................... 159 R.Senthilnathan Performance Improvement inHotRollingProcesswithNovel NeuralArchitecturalSearch ........................................ 177 SrinivasSoumitriMiriyala,ItishreeMohanty,andKishalayMitra vii About the Editors ShubhabrataDatta presentlyaResearchProfessorintheDepartmentofMechan- icalEngineering,SRMInstituteofScienceandTechnology,Chennai,India,didhis Bachelors,Masters,andPh.D.inEngineeringfromIndianInstituteofEngineering ScienceandTechnology,Shibpur,India(previouslyknownasB.E.CollegeShibpur) inthefieldofMetallurgicalandMaterialsEngineering.Dr.Dattahasmorethan28 yearsofteachingandresearchexperience.Hisresearchinterestisinthedomainof designofmaterialsusingartificialintelligenceandmachinelearningtechniques.He wasbestowedwiththeExchangeScientistAwardfromtheRoyalAcademyofEngi- neering, UK and worked in the University of Sheffield, UK. He also worked in theDeptofMaterialsScienceandEngineering,HelsinkiUniversityofTechnology, Finland,DeptofMaterialsScienceandEngineering,IowaStateUniversity,Ames, USA, and Heat Engineering Lab, Dept of Chemical Engineering, Åbo Akademi University,FinlandasVisitingScientist.HeisaFellowofInstitutionofEngineers (India), Associate Editor, Journal of the Institution of Engineers (India): Series D, andeditorialboardmemberofseveralinternationaljournals. J.PauloDavim received his Ph.D. degree in Mechanical Engineering in 1997, M.Sc. degree in Mechanical Engineering (materials and manufacturing processes) in1991,MechanicalEngineeringdegree(5years)in1986,fromtheUniversityof Porto(FEUP),theAggregatetitle(FullHabilitation)fromtheUniversityofCoimbra in2005andtheD.Sc.fromLondonMetropolitanUniversityin2013.HeisSenior Chartered Engineer by the Portuguese Institution of Engineers with an MBA and Specialist title in Engineering and Industrial Management. He is also Eur Ing by FEANI-BrusselsandFellow(FIET)byIET-London.Currently,heisaProfessorat theDepartmentofMechanicalEngineeringoftheUniversityofAveiro,Portugal.He hasmorethan30yearsofteachingandresearchexperienceinManufacturing,Mate- rials,Mechanical,andIndustrialEngineering,withspecialemphasisinMachining& Tribology.HealsohasinterestinManagement,EngineeringEducation,andHigher Education for Sustainability. He has guided large numbers of postdoc, Ph.D., and master’s students as well as has coordinated and participated in several financed researchprojects.Hehasreceivedseveralscientificawards.Hehasworkedasevalu- atorofprojectsforERC-EuropeanResearchCouncilandotherinternationalresearch ix x AbouttheEditors agenciesaswellasexaminerofPh.D.thesisformanyuniversitiesindifferentcoun- tries.HeistheEditorinChiefofseveralinternationaljournals,GuestEditorofjour- nals,bookEditor,bookSeriesEditor,andScientificAdvisoryformanyinternational journalsandconferences.

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.