礦井通風(fēng)專業(yè)-畢業(yè)論文外文文獻(xiàn)翻譯

外文資料原文Anintelligenthybridsystemforsurfacecoalminesafetyanalysisabstract：Analysisofsafetyinsurfacecoalminesrepresentsaverycomplexprocess.Publishedstudiesonminesafetyanalysisareusuallybasedonresearchrelatedtoaccidentsstatisticsandhazardidentificationwithriskassessmentwithintheminingindustry.DiscussioninthispaperisfocusedontheapplicationofAImethodsintheanalysisofsafetyinminingenvironment.Complexityofthesubjectmatterrequiresahighlevelofexpertknowledgeandgreatexperience.ThesolutionwasfoundinthecreationofahybridsystemPROTECTOR,whoseknowledgebaserepresentsaformalizationoftheexpertknowledgeintheminesafetyfield.Themaingoalofthesystemistheestimationofminingenvironmentasoneofthesignificantcomponentsofgeneralsafetystateinamine.Thisglobalgoalissubdividedintoahierarchicalstructureofsubgoalswhereeachsubgoalcanbeviewedastheestimationofasetofparameters(gas,dust,climate,noise,vibration,illumination,geotechnicalhazard)whichdeterminethegeneralminesafetystateandcategoryofhazardinminingenvironment.BoththehybridnatureofthesystemandthepossibilitiesitoffersareillustratedthroughacasestudyusingfielddatarelatedtoanexistingSerbiansurfacecoalmine.1.IntroductionOneofthemostimportantandalsomostcomplexproblemsencounteredinsurfacecoalminesissafetyanalysis.Anappro-priateandreliablesolutionforthisproblemisvitalfortheworkingprocessinmineswithsurfaceexploitation.Publishedstudiesonminesafetyanalysisareusuallybasedonresearchrelatedtoaccidentsstatistics(KecojevicandRadomsky,2004;Karra,2005;BennettandPassmore,1986;Marovelli,1981).Severalstudieshavebeenconductedonhazardidentificationandriskassessment(SouthAfricanMiningIndustry,2003;JoyandGriffiths,2005)whichaimstoprovideadviceonhazardidentificationandriskassessmentwithintheminingindustry.Systemssafetyanalysismethodsprovideaproactiveapproachtoanalyzesystemsforpotentialhazardsthatmaythreatenthehealthandsafetyofminers.Thesystemsapproachtothesafetyproblemfocusesonthesystemtakenasawhole.Itinvolvestheinteractionofpeople,machines,andenvironmentwithinproce-duralconstraints(Hammer,1972).Itusesanumberoftechni-ques:thetechniqueofoperationsreview,thefailuremodeandeffectsanalysistechnique,thefaulttreeanalysistechnique.Contemporaryminingtheoryoperateswithanumberofmethodsandtechniqueswhichcanbeusedtosolveminesafetyproblems.Thesemethodsareusedincurrentengineeringpracticewiththehelpofappropriatesoftwareproducts.Softwareproductsinminesafetyneedexpertknowledgeandexperiencetobefullyexploited.Thisknowledgeconsistsofrulesandheuristicsexpertsusewhentheyapplynumericalmethods,andArtificialIntelligenceoffersformalismsandmechanismsforitsincorporationinsoftwaresystems(RussellandNorvig,2002).FormalizationofknowledgerepresentationanddevelopmentofmechanismsforusingthisknowledgeareamongmethodsandtoolsdevelopedbyAIforsolvingcomplexproblems(GiarratanoandRiley,2004).However,thecomplexityofsomeproblemsoutgrowsthepotentialsofsinglemethods.ApossiblesolutionistocombinetwoormoreAImethodsintoahybridintelligentsystem(GoonatilakeandKhebbal,1995).ThisapproachhasbeenadoptedinthecaseofPROTECTOR,ahybridsystemfortheanalysisandestimationofsafetyinminingenvironment,devel-opedattheFacultyofMiningandGeologyoftheUniversityofBelgrade.PROTECTORwasdevelopedbycombiningneuralnetworksandexpertsystemtechnology.Whiletheminingenvironmentestimationmethodologyisimplementedthroughanexpertsystem,someoftherelatedestimationparametersaredeterminedbyneuralnetworks.DiscussioninthispaperisfocusedontheapplicationofcombinedAImethodsintheanalysisofsafetyinminingenvironment.Afullunderstandingoftheprocessanduseofallcollecteddatarequiretheinvolvementofanexperiencedspecialistintheminesafetyfield.ThesolutionwasfoundinthecreationofPROTECTOR,whoseknowledgebaserepresentsaformalizationoftheexpertknowledgeintheminesafetyfield.Section2ofthispaperoutlinestheglobalproblem-solvingstrategythroughahierarchicaldecompositionofthemaingoal,theevaluationoftheminingenvironment,andtheformalizationofthisstrategybymeansofamodifiedobject-orientedanalysis(OOA)model.ThesystemstructureandthemainarchitecturalcomponentsofthePROTECTORsystemaredescribedinSection3.TheimplementationofthesystemintheKAPPA-PCexpertsystemsshellisdiscussedinSection4.Section5presentsacasestudy,followedbyaconclusioninthelastsection.2.Aformalizationofminingenvironmentevaluationproblem-solvingAnalysisofsafetyincoalminesrepresentsaverycomplexprocessbasedonestimationofnumerousandinterdependentparametersthatareclassifiedintoseveralbasiccriteriaforestimationofminingenvironment.Thesecriteriaarerelatedtothefollowingconditions:gas,dust,climate,noise,vibration,illuminationandgeotechnicalhazard(highwallstabilityandwastestability).Indiscussingoccupationalrisksintheminingindustry,itiscommonpracticetoidentifyhealthandsafetyhazardsseparately.Someoftheseenvironmentalstressesmayinteracttoproduceagreateroveralleffect.Incombinationoralone,ifenvironmentalstressesexceedhumantolerancelevelsforprolongedperiodsoftime,feelingsofdiscomfortwillarise,alertnesswilldecrease,accidentswilloccur,andperformanceandproductivitywilldrop.Miningisamongthemoretraditionalandconservativeengineeringdisciplines.Althoughvariousassessmenttechniquesandnumericalmethodsareavailable,theanalysisofsafetyinsurfacecoalminesreliesmostlyonheuristicsformulatedbyminesafetyexperts.ThiswaspreciselythemotivationfordevelopingofPROTECTORasahybridsystem,withitsexpertsystemcomponentinitscore.Thesystemarchitecturedrawsupontheexperiencegainedbysuccessfulimplementationofhybridsystemsindifferentfields,butrepresentsanovelapproachwhenminingisconcerned.Thenoveltyoftheapproachtominesafetywasfurtherreinforcedbythedevelopmentofanevaluationstrategy.Themaingoalofthesystemistheestimationofminingenvironmentasoneofthesignificantcomponentsofgeneralsafetystateinmine.Thisglobalgoalcanbesubdividedintoahierarchicalstructureofsubgoalswhereeachofthesesubgoalscanbeviewedastheestimationofasetofparameters(gas,dust,climate,noise,vibration,illuminationandgeotechnicalhazard)whichdeterminethegeneralminesafetystateandcategoryofhazardinminingenvironment.Duringthisprocess,theimpor-tance,i.e.significanceofeachparticularparametermustbetakenintoaccount.Thehierarchicaldecompositionofthemaingoalintosubgoals,representingtheproblem-solvingstrategy,makesiteasiertocopewiththecomplexitiesandtocoordinatetheuseoftheknowledgeincorporatedinthesystem.Thestrategyforevaluationofthesafetyinminingenviron-mentisformallyrepresentedusingamodificationoftheCoad-Yourdonobject-orientedanalysis(OOA)model(CoadandYourdon,1991).Inthestandardmodeleveryrealworldentityisrepresentedbyaclass(object)consistingofitsname,attributesandmethodspertainingtotheproceduresrelatedtotheobject.However,inordertoincorporatedeclarativeknowledge,wehaveresortedtoamodificationofthismodelbyincludinganew(fourth)element,featuringtheproduction(IF-THEN)rulesrelatedtoanobjectinthemodel.Thusboththeproceduralanddeclara-tiveknowledgerelatedtoaclassobjectcouldberepresented.SuchamodifiedOOAmodelwasthenusedfortherepresentationoftheminingenvironmentevaluationstrategyaswellasotherobjectsinthesystemandtheirmutualrelationships(Fig.1).Fig.1.ModifiedOOAmodeloftheminingenvironmentevaluationTheinheritancerelationsbetweenhierarchicallyconnectedobjectsrepresentingelementsofthestrategyaregivenbyfulllines,whiletheexchangeofmessagesbetweenclassesisrepresentedwithdottedlines.Themodelwasthebasisfortheimplementationofthesysteminanobject-orientedexpertsystemshell.3.ArchitectureofthePROTECTORsystemSafetyofminesandpreventionofaccidentsisanissueofgreatimportanceinminemanagement.Inordertobeabletomakeappropriateandtimelydecisioninthisdelicatematterthemanagementneedstobeadequatelyandaccuratelyinformedabouttheminesafetystateonadailybasis.Thisinformationis,naturally,onlypartoftheinformationrequiredbyminemanage-ment.Informationsystemsofferthetoolsandtechniquesforhandlingalltheinformationflowsincomplexsystemsuchasmines.Tothatend,aTechnologicalInformationSystem(TIS)hasbeendeveloped,whichcanbetailoredtosuitrequirementsofanymineinSerbia.Itintegratesthefollowingmodules:HumanResources,Maintenance,MineSafety,MinePlanningandProduc-tion,ManagementInformationDecisionSupportSystemandTechnicalData,anddeliverscriticalinternalandexternalinformationneededtosupportminingbusiness.TIScomponentsareintegratedandoperatewithauniquedatabase,andPROTECTOR,thehybridsystempresentedinthispaperispartoftheMineSafetyTISmodulenamedMISS(MineSafetyInformationSystem).TheUnifiedModelingLanguage(UML)diagraminFig.2representsthestructureofTISandtheplaceofMISSwithinTIS,aswellasthestructureofMISSandtheplaceofthehybridsystemProtectorwithinMISS.UML,asastandardlanguageforvisualization,specification,constructinganddocumentingofdataonsoftwarewasadoptedforthesoftwaredevelopmentanalysisphase.MISSmoduleasapartofTIScontainsseveralcomponentssuchas:HealthCare,InjuriesandProfessionalDiseases,PersonnelSafetyRevises,SafetyReports.Eachofthesecomponentshascomponentsofitsown,asshownbythePROTECTORsystemexample.PROTECTORcomprisesseveralcoupledsoftwarecomponents,suchasVB(VisualBasic)Application,Expertsystem,NeuroApplication(Fig.2)(Lilic′etal.,2002a,b).ThesystemconnectsnumericalmethodswithAImethods,thusintroducingheuristicsandformingaknowledgebasewithknowledgeobtainedfromengineeringpractice.Fig.2.Componentviewoftechnologicalinformationsystem.Intheproposedsystem,thecomponentVBApplicationkeepsrecordsofminingenvironmentcharacteristic,runsanalysisandpresentsaresultoftheexpertsystemanalysis.Thesecondcomponentisadiagnosticexpertsystem,whichanalysestheobtainedresultsaccordingtoaseriesofcriteria(gas,dust,climate,noise,vibration,illuminationandgeotechnicalhazard).Asaresultofexpertanalysis,anestimationofvalidityandeffectivenessoftheminingenvironmentstateisobtained,followedbysuggestionsforitsimprovement.Thearchitectureofthesystemandthesoftwareenvironmentinwhichthesystemwasdevelopedprovideforadynamiccommunicationbetweendiffer-entsegmentsofthesystemandthusforunlimitedpossibilitiesoftestingdifferentmodificationsofthesystemandobtainingafinalsolutionwhichsatisfiesalloftheestablishedcriteria.Neuroapplicationmanagestrainingofneuralnetworksfordeterminationofpermittedexposuretimeonspecificvibrationlevel.TheneuralnetworkwastrainedthroughNeuroShell(Samarasinghe,2006;MarquesdeSa′etal.,2007)withdataobtainedfromdiagramsfordeterminationofpermittedexposuretimeonspecificvibrationlevelindependenceoffrequencyandacceleration.PROTECTORcontainsthe‘‘standard’’elementsofanexpertsystem:theknowledgebase,aninferenceengine,theuserinterfaceandaworkingmemory,butalsoamodulefortheinterfacewithroutinesforrelevantparametersdetermination,theVisualBasicroutinesthemselves,andadatabaseusedbytheseroutines(Fig.3).Fig.3.ArchitectureofPROTECTORexpertsystemcomponent.Themainpurposeoftheuserinterfaceistoprovidemeansforasuccessfuldialogue,i.e.anexchangeofinformationbetweentheuserandthesystem.ItistheuserinterfacethatenablesPROTECTORtoobtainallthenecessaryinformationfromtheuser,ononehand,andthattransformssystem’sresultsandconclusionsintoinformationtheusercanunderstand,ontheother.ThePROTECTORknowledgebaseisaformalizationoftheminesafetyexpert’sknowledge.Knowledgeinexpertsystemsbasicallyconsistsoffactsandheuristicswhichcanberepresentedbymeansofrules,frames,semanticnetworksandotherformalisms.Sinceknowledgeisthekeyfactorinproblemsolutionanddecisionmaking,thequalityandusabilityofanexpertsystemisbasicallydeterminedbytheaccuracyandcompletenessofitsknowledgebase.Theselectionoftherepresentationformalismisveryimportantandplaysasignificantroleinknowledge-baseddesign.Theproblem-solvingstrategyisrealizedbytheexpertsystem’sinferenceengine.Thisreasoningmechanisminfersconclusionsbasedonknowledgefromtheknowledgebaseandtheavailableinformationpertainingtothesafetyproblemathand.Theinferenceenginestoresintermediateresultsintheworkingmemory.4.ImplementationissuesTheobject-orientedapproachinsystemstructuringandmodeling(Fowler,2007;Stevens,2006)wasusedasthestrategyfordefiningthemodelofprocessesanddatainthedevelopmentofPROTECTOR,andUMLforthesoftwaredevelopmentanalysisphase.Thetaskofvisualmodelingofthesystemistodefinetheobjectsandlogicoftherealsystemusingtheadoptedgraphicnotation.VisualStudio6.0waschosenastheprogrammingenvironmentforthedevelopmentofPROTECTOR.UMLhasbeenusedasthemostappropriatenotation,andthesystem’sarchitecturewasconceivedintheformofathree-levelclassdiagram.Thisarchitecturesupportswelltheobject-orientedapproachinmodeldevelopmentforcomplexapplications.Itsmaincharacteristicisaseparationofthedomainmodel,whichisrepresentedbybusinessservicesanddataservices,fromuserinterface,representedbyuserservices.Fig.4depictsthethree-tieredservicemodelofPROTECTOR.Fig.4.Three-tieredservicemodelofPROTECTOR.ThenineclassesidentifiedwithintheuserservicesofPROTECTORrepresentitsinterfaceforms.Theyareusedfordatamanipulation(entering,viewingandsearchingthedata),textualandgraphicalpresentationofresultsandcommunicationwithothermodulesinPROTECTOR(expertsystemandneuralnet-work).Theclassesrelatedtouserservicescommunicatewithclassesatthebusinessservicelevelbysendingmessagesthatinitiatetheexecutionofspecificapplications.TwointerfaceformsbelongingtouserservicesareshowninFig.5.Theyenabletextsearchandediting,communicationwiththedatabase,creationofbusinessdiagrams,etc.Fig.5.InterfaceformsofPROTECTOR.Threeclasseswereidentifiedwithinthebusinessservices,allofthemareVisualBasicapplicationmodulesandtheyareusedforsafetyassessment.InFig.6anactivitydiagramdepictsthedynamicmodelofoneoftheseclasses-theNeuroclass.Thisclassincludesspecificproceduresbasedonthemodelofpermittedtimedeterminationforspecifiedlevelofvibrationusinganeuralnetworkandlinearinterpolation.Fig.6.ActivitydiagramfordynamicmodelofNeuroclassDataservicesprovidedatamaintenance,dataaccessandmodificationfunctions.InviewofthecomplexityofthePROTECTORsystem’sglobalmodeldatastructure,whichhadtomodelallrelevantparametersofcomplexsurfacecoalminesafetyanalysis,thedesignandrealizationofthedatabasewasexecutedintheMSAccessrelationaldatabasemanagementsystem.Thesystemofferssafedataarchivingforcomplexdatamodelsasthisone,aswellasallproceduresfordatamanipulation.TheuseofSQLasastandardquerylanguagefordatamanipulationsecurestheopennessofthehybridsystemPROTECTORforaconnectionwithdifferentenvironments.Fig.7depictsthestructureofthedatabaserelevanttoPROTECTORthroughtheMSAccessRelationshipspanel.Fig.7.DatabasestructureofPROTECTOR.PROTECTORwasdevelopedusinganexpertsystemsshell,theKAPPA-PCapplicationsdevelopmentsystem.KAPPA-PCisaMSWindowsapplicationwhichprovidesawiderangeoftoolsforconstructingandusingapplicationsbymeansofahigh-levelgraphicalenvironmentwhichgeneratesstandardCcode.IntheKAPPA-PCsystem,thecomponentsofthedomainarerepresentedbyobjectsthatcanbeeitherclassesorinstanceswithinclasses.Therelationshipsamongtheobjectsinamodelcanberepresentedbylinkingthemtogetherintoahierarchicalstructure.ThusthemodifiedOOAmodelbasedonthestrategyforevaluationofthegeneralsafetystateofthesurfacecoalminecouldbeeasilymappedontotheappropriateelementsofKAPPA-PC.Object-orientedprogrammingtoolswithinKAPPA-PCwereusedtoendowPROTECTORobjectswithmethodsthatspecifywhatobjectscando.Firsttheobjectsandmethodsfortheknowledgebasewereconstructed.Thenmechanismswerebuildthatspecifyhowobjectsshouldbehaveandthatcanreasonabouttheobjectsbyusingrules.Eachrulespecifiesasetofconditionsandasetofconclusionstobemadeiftheconditionsaretrue.Theconclusionsmayrepresentlogicaldeductionsabouttheknowl-edgebaseorspecificationsofhowitchangesovertime.Eachruleisarelativelyindependentmodule,whichmadeitpossibletobuildthereasoningsystemsgradually,rulebyrule.TheclassesandobjectsofthemodifiedOOAmodelweretransformedtoclassesandinstancesintheKAPPA-PCsystemasshowninthesystem’sobjectbrowser(Fig.8).TheobjectbrowsershowsalsoclassesthatKAPPA-PCgeneratesforeachapplication,suchasRoot,ImageandKWindow.Fig.8.TheKAPPA-PCObjectBrowserforPROTECTORClasses/instancesaredescribedusingtheclass/instanceeditor,whileslotfacetsaredefinedbymeansofthesloteditor.Slotsrepresentclassattributeswhilemethodsintheclass/instanceeditoraccountforbothmethodsandIF-THENrulesrelatedtoaclassinthemodifiedOOAmodel.AsanexampleconsidertheGasclassgiveninFig.9.Fig.9.TheGasClass.TheclasshasaparentclassMineEnvironandsixslots.Fivemethodsarelisted.Thefirstthreearenumericalprocedures(CalcFakGas,CalcFakMix,CalcHlpFMix)usedforcalculatingthevalueofattributesFakGasaandFakMixandtheremaining(CalcKatME,CalcEstimate)containrulesfortheevaluationofthemineenvironmentonbasisofseveralparametersandtheirmutualrelationships.SinceallrulesinthesystemdonothavetoberelatedtoparticularobjectsKAPPA-PCoffersthepossibilityofspecifyingrulesindependently,usingaruleeditor.Theproblem-solvingprocessinPROTECTORunfoldsbymeansoftheKAPPA-PCbackwardchaininginferenceengine.Goalstobesatisfiedbybackwardchainingaredefinedbymeansofthegoaleditor.ThegoalsinPROTECTORpertaintoestimationofdifferentparametervalues.Goalscanalsobegeneratedandmodifiedwithinmethods.Thisfeatureenablesthecreationofnewrulesandmodificationofexistingonesdynamically,duringsystemoperation.AnexampleofarulefornoiseandthedecompositionofthegoalPara-metClimaintosubgoalsareillustratedinFig.10.Fig.10.RuleNoise1andtheparametclimategoalTheinterfacedevelopedforPROTECTORinKAPPA-PCfullyexploitstheGUI(graphicaluserinterface)technologyavailableforMSWindowsapplications.Itenablesastraightforwardandeasymanipulationofinputdataandcontroloverpartsoftheproblem-solvingprocess.Italsoofferssuggestionsandrecom-mendationstotheuserwhichcancontributetotheimprovementoftheoverallperformanceoftheminesystem.5.TheopencastmineKolubarafieldD:acasestudyThissectionillustratestheuseofthePROTECTORhybridsystemintheanalysisandestimationofthestateofminingenvironmentsafetyintheopencastmineKolubaraFieldD.OneofthemainreasonsforchoosingopencastKolubaraFieldDfortheanalysisisthescaleofthismineintermsofcoalproductionandnumberofemployees,whichmakesitthelargestcoalopencastmineinSerbia.TheKolubaraminingregionisabout50kmsouthwestofBelgradenearthetownofLazarevac.Fouropencastminesareoperatedinthisregionatpresent.Thetotaloutputin2008was29.3milliontonswhichisabout80%oftheligniteoutputfromSerbianopencastmines.TheopencastmineKolubaraFieldDisequippedexclusivelywithbucket-wheelexcavators–beltconveyor–spreadersystems(Fig.11).Draglinesofvarioussizesareusedtosupportthemainequipment.Thebucket-wheelexcavatorsusedforoverburdenremovalareconnectedtothespreadersontheinsidedumpviabeltconveyorsystems.Twoseparatebeltconveyorsystemsareoperatingonindependentworkinglevelsforlignitemining.Theopencastminecapacityisratedforaligniteoutputof15milliontonssupportedbyaworkforceof2550employees.Fig.11.PresentopencastpositioninFieldDInordertoassesstheoverallminingenvironmentsafetystateoftheFieldDopenpit,measurementsofphysicalandchemicalparametersattheworkingenvironmentwereundertaken.Measurementsincludeddeterminationofconcentrationofgasesandairbornedust,levelsofnoiseandvibrations,aswellasilluminationandclimateconditions(dryandwetbulbtempera-ture,humidity,effectivetemperature)atworkingenvironments.ObservationswereperformedatallworkingenvironmentsontheFieldDopenpitduringthesummerperiodasrequestedbySerbianminingandoccupationalsafetyregulations.Onbasisofalldatacollectedinthesystem’sdatabase,anexpertisebyPROTECTORwasinitiated.Achainofrulesleadingtoassessmentsofspecificcharacteristicsofthemineworkingenvironmentwasactivatedusingtheappropriatecommandbuttonsonthemaininterfacepanel(Fig.12).Thereportontheresultsoftheexpertisewasthendisplayedonthesamepanel.ApplicationofPROTECTORenabledevaluationandanalysesofphysicalandchemicalparametersatworkingenvironmentsforeachworkingpositionseparately,resultinginsuggestionofmeasuresforconditionimprovements.Fig.12.Maininterfacepanelwiththeresultsoftheexpertise.Some144workingenvironmentswereanalyzed,withatotalof864parametersexamined.Fig.13providesanoverviewofthenumberofworkingenvironmentswhichmeetrequirementsregardingsafetyandcomfortaccordingtosix(outofsix),five,fourandthreeexaminedparametersintheprocessofligniteminingatFieldDopenpit.Fig.13.WorkingenvironmentsparametersoverviewatFieldDopenpitItshouldbeemphasizedthatonly19examinedworkingenvironments,outof144,or13.19%,meeteachrequirement.Itshouldbeaddedthatasmuchas63(43.75%)and58(40.27%)workingenvironmentsarehaving,respectively,atleastoneandtwoparametervalueswhichareharmfulforthehealthofemployees.Asomewhatfavorablefactisthatonly4workingenvironments(2.77%)havehalfoftheexaminedparameterswithharmfulvalues.Evaluationoverview(satisfactory/unsatisfactory)byspecificparametersispresentedinFig.14.Thebenchmarksforsatis-factoryparametersarelimitvaluesdefinedbymininglegislation.Fig.14.EvaluationoverviewofharmfulparametersatworkingenvironmentsatFieldDopenpit.Airbornedustconcentration,withlevelshigherthanallowed,wasfoundat102workingenvironments.Thisparameterstandsfor53.12%ofthetotalofunsatisfactoryparameters(192).Noiselevelwashigherthanallowedat76workingenvironments(39.58%oftotalunsatisfactoryparameters).Climateparameterswereunsatisfactoryat10workingenvironments,accountingfor5.20%ofthetotalnumberofunsatisfactoryparameters.Illumina-tionwasbelowallowedminimumvalueat3workingenviron-ments,thusmakingonly1.56%oftotalunsatisfactoryparameters.Vibrationshavenotmettherequirementsatonly1workingenvironment;therefore,thisparametermakesonly0.52%oftotalunsatisfactoryparameters.Eachof144examinedenvironmentshasmetrequirementsregardingconcentrationofgasses.Theexpertsystemgivesestimatesforallofthemineenvironmentsafetycharacteristics.However,itiswellknownfromcurrentengineeringpracticethatusersareoftenunabletodeterminesolutionsforpoorlyorunfavorablyestimatedchar-acteristics.InordertosolvethisproblemPROTECTORofferssuggestionswithconcretemeasuresforimprovementofcertaincharacteristicsaswellasoftheoverallmineenvironmentsafetystate.Bysimplyactivatingthecommandbuttonsintherecommendationsgroup,suggestionsareobtainedfollowedbyproposalofpossiblemeasures.SuggestionsfortheimprovementofunfavorablyestimatedcharacteristicsofdustconcentrationatFieldDopenpit(bucket-wheelexcavatorG-1,operator’scab–measurementpoint)aregiveninFig.15.Fig.15.SuggestionsfortheimprovementofunfavorablyestimatedcharacteristicsPROTECTORwasimplementedforthefirsttimein2005intheopencastmineKolubaraFieldD.Asubstantialgrowthofthenumberofinjuriesin2005andtheprecedingyearsecuredfullsupportoftheminemanagementfortheimplementationofthesystem.Themanagementwasparticularlyinterestedintheapplicationofthemeasuresandrecommendationsfortheimprovementofsafetyconditions.Theresultsoftheapplicationofthesystemwerehighlyencouragingasthenumberofinjuriesstartedtodecreaseconsiderablyinthefollowingyears.Theinjuryexperiencedatausedthatsupporttheeffective-nessofPROTECTORwereobtainedfromtheKolubaraMineSafetyDivisionaccidentinvestigationreports.Theserecordsindicateatotalof1368injuriesforthe11-yearperiodfrom1998to2008(Fig.16).Amongtheserecords,7casesbelongtofatalinjuries,192totheseriousand1169tolightinjuries.ItshouldbeemphasizedthataccordingtoregulationsandlawsinSerbiainjuriesarecategorizedintothethreecategories:fatal,serious(morethan30workdayslost)andlight(lessthan30workdayslost)injuries.Fig.16.ThetotalnumberofinjuresattheopencastmineKolubaraFieldDThehistoricalrecordofinjuriesshowsadeclineasthetotalnumberofinjuriesdroppedfr