摘要本文主要是設(shè)計一款架空乘人裝置。在進行設(shè)計時首先分析了現(xiàn)有架空乘人裝置的工作原理及系統(tǒng)組成，以及目前架空乘人裝置的存在主要問題及未來的發(fā)展趨勢。然后根據(jù)給定的條件設(shè)計由電動機，行星減速器，錐齒輪減速器，托壓輪，摩擦盤以及尾輪等組成的常規(guī)型架空乘人裝置。設(shè)計中主要對常規(guī)型架空乘人裝置的行星減速器，錐齒輪減速器以及鋼絲繩與滑輪進行了選型與設(shè)計。同時，對于制動器，聯(lián)軸器以及架空乘人裝置的日常使用及維護進行了涉及和說明。為了使設(shè)計更加可靠，關(guān)鍵詞：架空乘人裝置;行星減速器;鋼絲繩ABSTRACTThispaperistodesignanaerialpassengerdevice.Whenmakingdesign,firstofall,analysestheexistingaerialpassengerdevice'sworkingprincipleandsystemcomposition,andanalysesthemainproblemsandthefuturetrendofdevelopmentoftheaerialpassengerdevice.Thendesignthemotor,planetarygearreducer,bevelgearreducer,pressurewheel,frictionplateandthetailwheeldeviceandsoonaccordingtothegivenconditions.Theplanetarygearreducer,bevelgearreducerandsteelwireropeandpulleyhascarriedontheselectionanddesigninthedesignoftheaerialpassengerdevice.Atthesametime,thebrakeandcouplingandaerialpassengerdeviceofdailyuseandmaintenanceareinvolvedandinstructions.Inordertomakeadesignmorereliable,usingthesolidworkssoftwaretodesignpartsofthethreemodelingandfiniteelementanalysis,finallycombiningwiththedesignspecificationsforthecaddrawingsofthedraw.Thedesignofaerialpassengerdevicecanmeetthedesignrequirements.Keywords：Aerialpassengerdevice;planetaryreducer;steelwirerope目錄1緒論 頂錐角：根錐角：外錐距：5滑輪結(jié)構(gòu)的設(shè)計5.1滑輪設(shè)計計算繩索滑輪一般用來導(dǎo)向和支承，以改變繩索及其傳遞拉力的方向或平衡繩索分支的拉力。5.1.1滑輪結(jié)構(gòu)和材料承受載荷不大的小尺寸滑輪（mm）一般制成實體的滑輪，用、或鑄鐵（如）。受大載荷的滑輪一般采用球鐵（如）或鑄鋼（如等），鑄成帶筋和孔或帶輪輻的結(jié)構(gòu)。大型滑輪（mm）一般用型鋼和鋼板焊接結(jié)構(gòu)。受力不大的滑輪直接安裝在心軸上使用，受有較大載荷的滑輪則裝在滑動軸承（軸套材料采用青銅或粉末冶金等）或滾動軸承上，后者一般用在轉(zhuǎn)速較高、載荷大的情況下。輪轂或軸套長度與直徑比一般取為。具有固定軸的滑輪成為定滑輪；具有活動軸的滑輪（隨繩索串動改變其位置）稱為動滑輪。在本設(shè)計中滑輪采用鑄鋼件鑄成有輪輻的結(jié)構(gòu)。5.1.2鋼絲繩進出滑輪時的允許偏角鋼絲繩繞進或繞出滑輪槽時偏斜的最大角度（即鋼絲繩中心線和與滑輪軸垂直的平面之間的角度）推薦不大于。5.1.3滑輪主要尺寸滑輪的主要尺寸如圖5-1所示：根據(jù)前面所選的鋼絲繩直徑mm進行設(shè)計與計算。下面的主要參數(shù)計算公式以及查表主要是根據(jù)《中華人民共和國機械行業(yè)標(biāo)準(zhǔn)》《起重機用鑄造滑輪繩槽斷面》里面的公式以及數(shù)據(jù)確定。圖5-1滑輪主要尺寸滑輪繩槽底半徑：mm查表取：，一般為這里取：所以滑輪直徑：查表取：mm5.1.4繩槽斷面尺寸本標(biāo)準(zhǔn)的繩槽半徑R是根據(jù)鋼絲繩公稱直徑d的最大允許偏差為+7%確定的。鋼絲繩繞進或繞出滑輪槽時偏斜的最大角度（即鋼絲繩中心線和與滑輪軸垂直的平面之間的角度）應(yīng)不大于。繩槽表面粗糙度分為兩級：1級：um2級：um根據(jù)工作情況，選滑輪繩槽半徑mm，表面粗糙度級的繩槽斷面。標(biāo)記為：繩槽斷面9.0-2JB/T9005.1—1999。具體尺寸參數(shù)，如圖5-2所示：圖5-2繩槽斷面圖mmmmmmmmmmmmmmmmmm5.2鑄造滑輪形式和軸承尺寸根據(jù)《中華人民共和國機械行業(yè)標(biāo)準(zhǔn)》中的《起重機用鑄造滑輪型式、輪毅和軸承尺寸》，進行滑輪形式和軸承尺寸的確定。因為此滑輪為煤礦井下架空乘人裝置所使用，考慮到井下環(huán)境，以及滑輪所承載的力較大，所以選擇JB/T9005.8-1999起重機用鑄造滑輪E2型，這種滑輪使用圓柱滾子軸承，承載力大，且對于密封的要求為一般密封、無需內(nèi)軸套，加工簡單。E2型滑輪型式結(jié)構(gòu)如圖5-3所示：圖5-3E2型滑輪型式結(jié)構(gòu)（1）輪轂尺寸的確定：根據(jù)前面計算，可以得知滑輪的最小直徑D為400mm。查《起重機用鑄造滑輪型式、輪毅和軸承尺寸》表，可以查得D5的尺寸為160mm。所以所選滑輪標(biāo)記為：滑輪E16×400–160JB/T9005.3—1999輪轂的其余尺寸如表5-1所示：表5-1輪轂尺寸基 本 尺 寸參考尺寸螺栓孔數(shù)n螺釘孔數(shù)n1D7K7D4D12D15D17D19H12B()F1()D2R5290M12143203252951158.540040016612（2）軸承型號的確定：根據(jù)計算的輪轂尺寸，同樣查《起重機用鑄造滑輪型式、輪毅和軸承尺寸》，可以選擇出E2型滑輪所使用的軸承為圓柱滾子軸承，代號為NJ232。其尺寸如表5-2所示：表5-2所選軸承尺寸B3()D5D6E 型滾動軸承代號B()B4D7k7S2E2型寬度125160180NJ232481151092909（3）滑輪技術(shù)要求：=1\*GB3①材料：ZG35Mn=2\*GB3②外觀：滑輪表面應(yīng)光滑平整，應(yīng)去除尖棱和冒口，滑輪不得有影響使用性能和有損外觀的缺陷，如氣孔，裂紋，疏松，夾渣，鑄疤等。=3\*GB3③熱處理：滑輪應(yīng)進行退火處理，以消除鑄造時產(chǎn)生的應(yīng)力。=4\*GB3④尺寸公差和表面粗糙度：加工表面未注公差尺寸的公差等級按GB/T1804中的M級（中等級）；未加工表面粗糙度值按GB/T1031中的um。=5\*GB3⑤形位公差：滑輪的形狀和位置公差應(yīng)符合《起重機用鑄造滑輪繩槽斷面》中的規(guī)定。=6\*GB3⑥裝配：裝配好的滑輪應(yīng)能靈活地旋轉(zhuǎn)。滑輪的加工部位（內(nèi)孔，繩槽表面等）和隔環(huán)的外露部位應(yīng)涂抗腐蝕的防銹油，不加工部位應(yīng)涂防銹漆。=7\*GB3⑦滑輪的加工部位（內(nèi)孔、繩槽表面等）和隔環(huán)的外漏部位應(yīng)土抗腐蝕的防銹油；不加工部位應(yīng)涂防銹漆。5.3本章小結(jié)本章主要對滑輪進行了結(jié)構(gòu)設(shè)計。主要依據(jù)《中華人民共和國機械行業(yè)標(biāo)準(zhǔn)》《起重機用鑄造滑輪繩槽斷面》選取鑄鋼件鑄成有輪輻的滑輪結(jié)構(gòu)。同時對滑輪的主要尺寸、繩槽斷面尺寸、鑄造滑輪的形式以及軸承尺寸進行了設(shè)計與驗算。6主要零部件的有限元分析6.1有限元分析簡介本章在有限元分析平臺SolidWorksSimulation環(huán)境下對架空乘人裝置部分零部件靜力學(xué)特性進行研究，分析影響加工質(zhì)量的關(guān)鍵零件的應(yīng)力、應(yīng)變及變形特性，其計算結(jié)果可為結(jié)構(gòu)優(yōu)化和改進提供理論依據(jù)。考慮到之前已經(jīng)對行星減速器高速級輸入齒輪軸進行了驗算，且此零件是一個十分重要的零件，所以本章同樣對行星減速器高速級輸入齒輪軸進行三維建模并對它進行有限元分析，已驗證我們之前的計算結(jié)果同時保證設(shè)計的準(zhǔn)確性。此外，架空乘人裝置的座椅是直接與人接觸的零部件，承擔(dān)著人的重量，之前并沒有對座椅的設(shè)計進行詳細(xì)的驗算，因此，本章同樣選用座椅作為三維建模對象并進行有限元分析。6.2主要零部件有限元分析6.2.1高速級輸入齒輪軸有限元分析根據(jù)計算可知，高速級齒輪軸主要是太陽輪，其模數(shù)為4，齒數(shù)為17，齒厚為33mm，創(chuàng)建好模型后，對模型進行簡單地受力分析并確定邊界條件。根據(jù)第四章的分析，齒輪受到一個徑向力和一個切向力，因為是直齒輪，所以軸向力為零。根據(jù)計算結(jié)果，設(shè)置徑向力為3082N，切向力為8468N。對于軸的固定，假設(shè)一種極端狀況，軸只受力，不旋轉(zhuǎn)，此時軸的工作情況最為惡劣。因此，在鍵槽處施加約束，使軸在鍵槽處固定設(shè)置好軸的邊界條件后，選擇軸的材料，選用軸常用的45號鋼作為材料，選好之后，solidworks軟件會自動的把參數(shù)付給齒輪軸，其參數(shù)如表6-1所示：表6-1齒輪軸材料參數(shù)屬性名稱數(shù)值單位彈性模量2.09x1011N/m^2泊松比0.269NA抗剪模量8.23x1010N/m^2質(zhì)量密度7890kg/m^3張力強度6x108N/m^2屈服強度3.55x108N/m^2熱擴張系數(shù)1.2x105/Kelvin熱導(dǎo)率48W/(m.K)比熱450J/(kg.K)設(shè)置好約束及材料之后，運行，即可得到應(yīng)力，應(yīng)變及位移云圖。高速輸入齒輪軸的應(yīng)力云圖如圖6-3所示：圖6-3高速齒輪軸應(yīng)力云圖高速輸入齒輪軸的位移云圖如圖6-4所示：圖6-4高速齒輪軸位移云圖高速輸入齒輪軸的應(yīng)變云圖如圖6-5所示：圖6-4高速齒輪軸應(yīng)變云圖因此可以得出做大的應(yīng)力應(yīng)變結(jié)果如表6-2所示：表6-2高速齒輪軸應(yīng)力應(yīng)變位移結(jié)果名稱類型最小最大應(yīng)力應(yīng)力94.7N/m21.2x108N/m2位移合位移0mm0.1449mm應(yīng)變對等應(yīng)變3.8x10-100.0003919346.2.2吊椅限元分析分析吊椅的邊界條件，吊椅的上部主要是固定在抱鎖器上面，人坐在吊椅的座位上，因此吊椅的邊界條件的施加如圖6-6所示，將上端面固定，座椅施加800N的力。給吊椅付常用的材料，Q235A，其屬性如表6-3所示：表6-3吊椅材料屬性屬性名稱數(shù)值單位彈性模量2.12x1011N/m2泊松比0.288NA抗剪模量8.23x1010N/m2質(zhì)量密度7860kg/m3張力強度3.9x108N/m2屈服強度2.35x108N/m2熱擴張系數(shù)1.2x105/Kelvin熱導(dǎo)率43W/(m.K)比熱440J/(kg.K)在最初設(shè)計時，設(shè)計吊椅的圓柱界面直徑為15mm2，座椅下面橫梁寬度為12mm，運行，座椅的應(yīng)力如圖6-7所示，圖6-7座椅應(yīng)力圖此時可以發(fā)現(xiàn)，座椅所能承受的最大應(yīng)力已經(jīng)超過了材料所能承受的屈服應(yīng)力，這樣不符合要求，且通過觀察，最大應(yīng)力集中在座椅與桿的焊接處，如圖6-8所示：圖6-8座椅最大應(yīng)力圖為了改善設(shè)計，通過分析，將座椅的桿截面積提高到20mm2，把座椅下橫梁寬度增加到18mm，且如圖6-5所示，焊接一個斜梁以增加強度。再次運行，吊椅的應(yīng)力應(yīng)變以及位移如圖6-9，6-10，6-10所示：圖6-9座椅應(yīng)力云圖圖6-10座椅位移云圖圖6-11座椅應(yīng)變云圖結(jié)果如表6-4所示：表6-4座椅有限元分析結(jié)果名稱類型最小最大應(yīng)力應(yīng)力5.8x10-8N/m21.3x108N/m2位移合位移0mm11.0082mm應(yīng)變對等應(yīng)變2.5x10-190.000453182改進后發(fā)現(xiàn)，做大的應(yīng)力已經(jīng)小于材料的屈服應(yīng)力，但是最大位移仍然大于了10mm，考慮到應(yīng)變，已經(jīng)可以滿足我們的設(shè)計需要了，因此可以認(rèn)為符合要求。6.3本章小結(jié)本章主要是對架空乘人裝置的部分零件進行了有限元分析。有限元分析插件對行星減速器高速輸入齒輪軸以及座椅進行了建模與分析。根據(jù)軟件給出的零件的應(yīng)力應(yīng)變以及位移云圖的分析表明之前的理論驗算完全滿足我們的設(shè)計要求。結(jié)論通過本次設(shè)計，了解到架空成人裝置及發(fā)展趨勢，確定了此架空乘人裝置的驅(qū)動部分的傳動方案為電動機帶動行星減速器與錐齒輪減速器進行傳動。其中行星減速器具有大傳動比的特點，錐齒輪減速器具有起到轉(zhuǎn)換傳動方向的作用。根據(jù)需要，確定該系統(tǒng)的傳動比為72.64，查閱行星減速器設(shè)計手冊，分配傳動比為：行星減速器為27，錐齒輪減速器為2.7。對于行星減速器選型二級NGW型的傳動結(jié)構(gòu)。設(shè)計過程中，對于行星減速器與錐齒輪減速器的齒輪，軸承，以及鍵均進行了設(shè)計計算與強度校核。設(shè)計結(jié)果均符合設(shè)計要求。分析了架空乘人裝置的發(fā)展現(xiàn)狀及發(fā)展趨勢，闡明了架空乘人裝置的組成和工作原理。對整機總體方案進行了詳盡的論證，為架空乘人裝置各個部件進行了選型設(shè)計及計算。重點完成架空乘人裝置的電機減速器的選型、鋼絲繩的選型與計算、承載托輪和壓輪的設(shè)計計算、尾輪裝置的設(shè)計計算、張緊輪的設(shè)計計算、吊椅設(shè)計計算以及緊急停車裝置的設(shè)計計算。參考文獻[1]王洪欣、李木、劉秉忠主編，《機械設(shè)計工程學(xué)》，1出版社，2004年1月[2]唐大放、馮曉寧、楊現(xiàn)卿主編，《機械設(shè)計工程學(xué)》，1出版社，2004年1月[3]成大先主編，《機械設(shè)計手冊》單行本機械傳動化學(xué)工業(yè)出版社，2000年[4]卜炎主編，《機械傳動裝置設(shè)計手冊》上冊機械工業(yè)出版社，1999年4月[5]卜炎主編，《機械傳動裝置設(shè)計手冊》下冊機械工業(yè)出版社，1997年12月[6]中國機械工程協(xié)會、中國機械設(shè)計大典編委會、朱孝錄主編，《中國機械設(shè)計大典》第四卷江西科學(xué)技術(shù)出版社，2002年1月[7]起重機設(shè)計手冊編寫組編，《起重機設(shè)計手冊》機械工業(yè)出版社，1977年8月[8]成大先主編，《機械設(shè)計手冊》單行本彈簧?起重運輸件?五金件，化學(xué)工業(yè)出版社，2004年1月[9]吳宗澤主編，《機械零件設(shè)計手冊》機械工業(yè)出版社，2003年11月[10]成大先主編，《機械設(shè)計手冊》單行本軸及其聯(lián)接，化學(xué)工業(yè)出版社，2004年1月[11]清華大學(xué)吳宗澤，北京科技大學(xué)羅圣國主編，《機械設(shè)計課程設(shè)計手冊》第3版高等教育出版社，2006年4月[12]東北大學(xué)《機械零件設(shè)計手冊》編寫組編，《機械零件設(shè)計手冊》第三版下冊冶金工業(yè)出版社，1994年4月[13]編輯委員會編，《現(xiàn)代機械傳動手冊》機械工業(yè)出版社，2002年3月[14]吳宗澤主編，《機械零件設(shè)計手冊》機械工業(yè)出版社，2004年[15]吳宗澤主編，《機械設(shè)計實用手冊》第二版，化學(xué)工業(yè)出版社，2003年10[16]單輝祖主編，《材料力學(xué)》北京，高等教育出版社，1999年[17]C.Yuksel,A.Kahraman,Dynamictoothloadsofplanetarygearsetshavingtoothpro.lewear,MechanismandMachineTheory(2004)[18]R.G.Parker,S.M.Vijayakar,T.Imajo,Non-lineardynamicresponseofaspurgearpair:modelingandexperimentalcomparisons,JournalofSoundandVibration(2000)[19]R.Maliha,U.C.Dogruer,H.N.Ozguven,Nonlineardynamicmodelingofgear-shaft-disk-bearingsystemsusing.niteelementsanddescribingfunctions,JournalofMechanicalDesign(2004)翻譯外文資料原文StrategiesforAutomatedMaintenanceofBeltConveyorSystemsProf.dr.ir.GabrielLodewijks

DeIftUniversityofTechnology,theNetherlandsSUMMARYThispaperdiscussesautomationofmaintenanceofbeltconveyorsystems,inparticularofidlerrolls.Automationofmaintenanceisapromisingalternativeforoutsourcingmaintenance,inparticularwhenlookingattheefficiency,accuracy,andcosts.Inordertooptimisemaintenanceefforts,theconceptofintelligentmaintenanceisintroduced.Thepoweredmaintenancetrolleythatcantravelautonomouslyoverthestructureofabeltconveyorisadaptedasaplatformofthemaintenancesystem.Onthistrolley,dataacquisitionequipmentforvibrationanalysisisinstalled.Dataminingcanbedoneeitheronboardofthetrolleyorinacentralcomputerdependingonthemaintenancestrategy.Theoptimummaintenancestrategyisdeterminedbyalogisticsimulationmodelthataccountsforthelay-outofthebeltconveyoritselfandtheaccuracyoftheinformationontheremaininglifetimeofitscomponents.INTRODUCTIONTodaymoreandmorecompaniesoutsourcemaintenanceinanattempttobalancethebudgetandreducethenumberofpermanentstaffmembers.Outsourcingmaintenancehoweveronlyworksifthecompanythattakesovermaintenanceemployswell-trainedandexperiencedpersonnelthatstaysonaspecificjobforaconsiderabletime.Unfortunately,realityisdifferentandmanycompanieshavepoorexperienceswithexternalcompaniesperformingmaintenance.Ingeneral,maintenanceonbeltconveyorsystemscanbedividedintoinspectionorconditionmonitoringofthetotalsystemandreplacementand/orreparation(inshortservicing)ofitscomponents.Mostproblemsexperiencedwithoutsourcementofmaintenanceareassociatedwiththeinspectionorconditionmonitoringofasystem.Itisnottrivialtoaccessthestatusofsometimesmovingcomponentsofabeltconveyor.Thesameexperiencedpersonshouldthereforecarryoutinspectionsonaregularbasis.Toovercomeoperationalproblemscausedbyalackofexperienceofexternalmaintenancepersonnel,theinspectionofbeltconveyorcomponentscanbeautomated.Inthiswayknowledgeofforexamplewearratesandreplacementschedulescanbebuiltupinadatabasesystem.Theexternalmaintenancecrewthencanbeusedtoreplacethewornoffcomponents.Alternatively,replacementofcomponentscanbeautomatedaswell.Thispaperdiscussesstrategiesandtechniquesforautomatedmaintenanceofbeltconveyorsystems.Section2definestheconceptofintelligentmaintenance,Section3discussesexistinginspectionsystemsthatcanbeusedinautomatedmaintenancesystems.Section4discussesmeansofassessingthestatusofrotatingcomponentsofbeltconveyorsbasedonvibrationbasedmonitoringconcepts.Section5presentsacasestudyandsection6finallyliststheconclusionsandrecommendations.INTELLIGENTMAINTENANCEMaintenanceonbeltconveyorsystemscanbedividedinconditionmonitoringofthetotalsystemandservicingofitscomponents.Conditionmonitoringisdefinedasthecontinuousorperiodicmeasurementandinterpretationofdatatoindicatetheconditionofacomponenttodeterminetheneedforreplacementorservicing.Conditionmonitoringthereforedealswiththeacquirementofdata(dataacquisitionorDAQ)fromsensors,theinterpretationofthatdata(dataminingorDAM)andwithtakingcorrectiveactions(ACT)oncomponentsthataretofail,thuspreventingfailsystemsfromdevelopingandpropagating.Thebasicconceptofconditionmonitoringistoidentifysubtlechangesinoperation,suchasincreasedvibrationlevels,thatindicateamechanical(orelectrical)problemisstartingtodevelop.Theseearlymessagesprovidemoretimetoplanformachinedowntimeandrepair.Therearefourtypicaltypesofmaintenance:preventivemaintenance:calendarbased,i.e.activitiesareplanneddependingonworkinghoursoratcertaintimeintervals(scheduledmaintenance);itmaybebasedonobserveddeteriorationofcomponents;nothingisrepairedbutpreventivejobsaredone.randommaintenance:opportunitybased,i.e.maintenanceisdonewhentheopportunityarises;thedecisiontomaintainacomponentbasedonopportunitiesmayormaynotbetriggeredbytheconditionofacomponent.correctivemaintenance:emergencybased,i.e.repairingwhenacomponentmalfunctions;thismaycauseageneralshutdownofthesystem;therepairactivitywasnotscheduledbeforehand.predictivemaintenance:conditionbased,i.e.componentsarebeingmonitoredandwhenirregularfactorsarediscovered,onewaitsuntilamaintenanceopportunityarises;itisaplannedandcorrectivemaintenance.Fromtheabovegivenfourtypesofmaintenanceitisclearthatonlyapredictivemaintenanceconceptqualifiesforapplicationinanintelligentmaintenancesystemthatenablesmaintenanceautomation.Intelligencehereisdefinedastheabilitytomakedecisionsbasedoninformationgatheredthroughsensorsintheequipmentorprovidedbythecontrolsystemofthetotaltransportsystem.Appliedtobeltconveyorsystemstheinformationgatheredfromasystemisinformationonthelifeexpectancyofindividualcomponentsasforexampleidlerrolls.Thisinformationleadstoadecisioneithertoinspectacertainidlerstationanditsrollsmorefrequentlyortochangearollforanewroll.Repairinginfactheremeanschangingonerollforanother.Whetherornotarollcanberepairedandtheeffectofthatonthebeltconveyor’sperformanceisoutsidethescopeofthisstudy.Themainissueinthisstudyisthequestionhowanautomatedinspectionstrategyisaffectedbytheaccuracyofthedataacquired.Intheorytherearetwoouterlimitsinpredictivemaintenance.Thefirstisthatnoaccurateinformationoftherollsisavailableatall,basicallymeaningthatanassessmentoftheremaininglifetimeismadepurelyonthebasisofhistoricaldataprovidedbytherollorbearingmanufacturers(predictivemaintenancebasedonstatistics).Thesecondisthatduringinspectionveryaccurateinformationonthestatusofrollsisgeneratedenablinganaccurateassessmentoftheremaininglifetimeofanindividualroll(predictivemaintenancebasedondata).Alogisticsimulationmodelismadetodeterminetheeffectoftheaccuracyofdataacquiredonautomatedinspectionstrategies.ThismodelisdiscussedinSection5.EXISTINGINSPECTIONSYSTEMSOneproblemfacedwithinspectionorconditionmonitoringofcomponentsofbeltconveyors,includingthebelt,pulleysandidlerrolls,isthattheyrotate.Sincetheconditionofcomponentslikerollsandpulleyscanonlybeassessedwhentheyarerotating,onlyconditionmonitoringsystemsbasedonvibrationanalysisoracousticalmonitoringcanbeused.Theoppositeholdsforthebelt.Thebelt’sconditioncanonlybeinspectedwhenthebeltconveyorsystemisnotoperating.Eitherway,aninspectorhastowalkthefulllengthoftheconveyortoinspectitscomponents.Anassociatedproblemisthatpulleysmaybefarapartfromeachotherorthattheconveyorhasagreatlength.Toeaseinspectioninthesecasesapoweredmaintenancetrolleycanbeusedforinspectionpurposes.Theconceptofapoweredmaintenancetrolleyisnotnew.Anearlyexampleofamaintenancetrolleyusedonabeltconveyorsystemwasthetrolleyusedonthe100kmPhosboucraaoverlandsystembuiltbyKruppinthe70-tiestotransportrawphosphateacrossadistanceof100kmfrominsidethewestSaharaacrossadesertofstonestotheloadingpointonthecoast.Thislong-distanceconveyorsystem,consistingofbeltsystemswithcentredistancesof6.8to11,7km.appliedamaintenancetrolleyconcepttoallowforinspectionalsoseeFigure1.TheKrupp-designturnedouttobeoccasionallyliabletoinstability.Figure1PoweredmaintenancetrolleyonKruppsysteminSahara.ArevitalisedversionofamaintenancetrolleyisshowninFigure2.Thisconcept,designedanddevelopedbyCKITofSouthAfrica,isquiterobustandstable.Ithasbeeninstalledonanumberofpipeconveyorsystems,bothinsideandoutsideSouthAfrica.Ithasthreeinspectionplatformsandissupportedonsixpoints(verticalandtransversedirection).Currentsystemsarepoweredfromthemainplatformbycombustionengines.Today’strolleysareallmenoperatedandinspectionandservicingworkiscarriedoutbymenaswell.Figure2

CKITconceptofpoweredmaintenancetrolleyTheconceptofthemaintenancetrolleyasdevelopedbyCKITisadaptedinthisstudyastheplatformforthefurtherdevelopmentofafullyautomatedmaintenancefacility.Thisdevelopmentisdividedinthreestages.Thefirststagewillbethedevelopmentofamaintenancerobotonthetrolley,enablingbothautomatedinspectionandservicing.Thedesignofsucharobot,whichisaresearchprojectcarriedoutatthesectionofTransportEngineeringandLogisticsofDelftUniversityofTechnology,isnotconsideredinthispaper.ThesecondstageistheimplementationoftheautomatedinspectionroutinesaswillbedescribedinSection5.Thethirdstageisthefullintegrationoftheautomatedmaintenancetrolleyinthetotalsystemscontrolsystem.DATAACQUISITIONANDMININGConditionmonitoringtechniquesgenerallyincludeoneorseveralalarmsthatgooffwhenaworkingpointisexceededorwhenatrenddeviatesfromtheexpectedvaluesintime.Referencesoftheworkingpointsofsignalsareprovidedbyknowledge-basedsystemsandnotbycomparisonwithamodelofthesystem.Signalsareacquiredbysensorsystems.DATAACQUISITIONTECHNIQUESChoosingtheproperdataacquisitiontechniquehasalargeimpactontheefficiencyofthemaintenancestrategy.Good,reliablemeasurements,aswellasproperanalysesoftheresultsofthosemeasurements,areessentialforreliableactionsofthemaintenancesystem.Forrotatingcomponentsmostoftenasignal-basedconditionmonitoringsystemisappliedbasedonvibrationand/oracousticsmeasurementtechniques.Anotheroptionisusingforceandtorquemeasurementsasabasisforconditionmonitoring.Forrotatingcomponentshowever,theapplicationofwirelesstorquemeasurementequipmentisrequired,whichisexpensive.Itissuitableasatemporarymonitoringsystembuttodaystillnotfitforlarge-scalepermanentmonitoringsystems.Spectralanalysisisanimportanttoolinvibrationbasedconditionmonitoring.Ingeneralvibrationbasedmonitoringmeansmeasuringaccelerationlevelsusingthree-dimensionalaccelerometers.Thesignalacquiredfromthesesensorsthereforeisaccelerationasafunctionoftime.InaspectralanalysisthissignalistransformedfromthetimedomaintothefrequencydomainbyapplyingaFastFourierTransformtechnique(FFT).Withthesignalinthefrequencydomaintherootsourceofthesignalcanbeeasilydetermined.Theanalysisofthespectraldensity,whichrelatestheenergyinthevibrationsignaltoaspecificfrequency,isagoodmeansofdeterminationoffaultsadvancingintime.Vibrationanalysisisoneofthemainformsofconditionmonitoringand,ingeneral,isoftenappliedintheindustry.Thespectraldensityofvibrationlevelsofagoodworkingcomponentwillgenerallybelow.Whenwear-outoccurs,orwhenloadsappearonspecificcomponents,thensomesmallbutnotablechangeswilloccurinthedynamicalbehaviourofthecomponent.Bymakingthesechangesvisibleandanalysingthem,adiagnosisoftheproblemcanbemade.Themonitoringtechniquesusedinpracticecanbedividedintotwomaincategories:signalRMS(RootMeanSquare)basedmonitoringdetailedsignalspectrumbasedmonitoringAcousticalanalysisstronglyresemblesvibrationanalysis.Dataminingfollowsafterdataacquisition.Dataminingconsistsofthreemainsteps.Thefirststepisthedetectionofdefectsthatisbasedonknowledgeofthedynamicsofthecomponentsmonitored.Thesecondstepisdataprocessing,transformingtheacquireddataindatathatisbetterfitforanalysis.Thethirdstepistheactualanalysisofthedataitselfrequiredtomakeadecisiontotakecertainactions.DATAMININGI-DETECTIONOFDEFECTSInthispaperthemainfocusisonbearingssincebearingsare,byfar,themajorsourceofthemalfunctioningofrotatingcomponentsincludingidlerrolls.Obviously,idlerrollscanalsofailasaresultofshellwear.Themechanismofshellwearhoweverdiffersfrombearingwearandassuchrequiresadifferentdetectionprocedure.Inthispaperonlythedetectionprocedureforbearingfailuresisdiscussed.Therearemanywaysinwhichbearingdynamics,whichmayleadtodefects,canbeclassified.OneofthemisbydefiningcomponentfrequenciesasafunctionoftherotatingspeedfrotandofsomegeometryparametersincludingthenumberofrollingelementsN,thediameteroftherollingelementsD,thecontactangleφ,andthebearingpitchdiameterP.Thefrequenciesidentifyingthefourmaindynamiceffectsinbearingsare:thecagerotationalorthefundamentaltrainfrequencyfcage:theinnerringorballpassinnerringfrequencyfir:theouterringorballpassouterringfrequencyfor:therollingelementorballspinfrequencyfre:Bearingdefectsshowupasanincreaseinspectraldensityfordefectrelatedfrequencies.Bearingdefectfrequenciesarearesultofimpactsduetotherollingelementspassingoverthedefectsastheypassthroughtheloadedzoneofabearing.Thedefectfrequencies,exceptforthecagerotationalfrequency,aresurroundedbysidebandsinarealsignal.Ifthedefectfrequencyoriginatesfromasignalthatpassedtheloadedzone,thereareksidebandswherek€N.Thenextfrequenciescouldappearinaspectrum:outerringdefect:at

innerringdefect:atrollingelementdefect:atcagedefect:atwherenisthenumberofharmonics.Asthedefectissmaller,themeasuredaccelerationsignalismorelikeapulsethanlikeasinewaveandtheenergycontentdecreaseswhilethedefectfrequencyincreasesinthespectrum.DATAMININGII-TECHNIQUEOFDATAPROCESSINGBandenvelopingistheprocessoftransformingavibrationsignalwithsmallsuperimposeddisturbancesintoisolateddisturbanceinformation.Themainreasonforusinganenvelopeofasignalisthatonecandetectdevelopingdefectslikesmallcracksinaveryearlystage.Theprocessofbandenvelopingconsistsofthreesteps:high-passfiltering,rectification,andlow-passfiltering.Astheenergyofadisturbancecomparedtotheenergyofthesinewaveisverylowthenthepulseishardlydetectableinthefrequencyspectrum.Thefirststepthereforeistouseahigh-passfiltertofilteroutthe(lowfrequency)sinusoidalcomponent.Theremainingsignalcontainsonlytherepetitivedisturbances.Thesignalthenisrectifiedandpassedthroughalowpassfilter.Thepeakinthefrequencyspectrumthenrepresentsthedefectfrequencyofthecomponentthatisdefective.Thepulseslosethehighfrequencycomponentsbecauseofthelow-passfilter.Therepetitionperiodhoweverremains.DATAMININGIII—DATAANALYSISDataanalysiscanbequitecomplicated.IfthescopeofanalysisisrestrictedtobearingsandthefouridentifiedpossibledefectsaslistedSection4.2,thentheprocedurecanbeasfollows.Firstthefrequencyspectrumisscannedforanomalies.Ifpeaksaredetectedinthisspectrumthentheequations(5)till(8)canbeusedtoidentifytherootcause.Knowingtherootcause,forexampleouterringproblems,thentheacquiredsignalleveliscomparedtoadatabaseidentifyingtheseriousnessofthedefect(s)anddeterminingapropercourseofaction.Partofthelatterdeterminationisbasedoncommonif-then-elsestructuresenablingastructured(andautomated)analysisofpossiblecausesandfutureeffectsonoperation.Ifmorethanonepossiblecauseforadefectisknown,forexampleaspecificsignalcanidentifyadefectinabearingbutalsointhesensoritself,thenconfidencefactorshavetobeappliedtorule-outthemostpossiblecause.ANINTELLIGENTMAINTENANCECONCEPTInthissectionaconceptforthelogisticcontrolofanintelligentmaintenancesystemisgiven.Themaintenanceconceptisbasedonthepredictivemaintenanceconcept,usingeitherstatisticsortheresultsofadetaileddataanalysis,whichwasintroducedinSection2.Thetechnicallay-outofthemaintenancesystemisbasedontheapplicationofanautomatedmaintenancetrolleyincludingamonitoringandservicingrobotasdiscussedinSection3.Thedataacquisitionandminingtechniquesusedwerediscussedintheprevioussection.MODELInthelogisticmodelanumberofelementsaredetailedincluding:thebeltconveyoritselfthebearingsthemaintenancerobottheinspectionrequirementstheservicingaspectsandthedataanalysisBeltconveyorInthemodelthebeltconveyorcanbespecifiedintermsofitslengthandtheidlerpitches.Thenumberofidlerstheniscalculatedautomaticallyassumingthatthepitchisconstant.Itisassumedthatacarryingidlerhas3rollsandareturnidler2.Eachrollhastwobearings,whichhaveaminimumlifelengthasspecifiedbytherollandbearingmanufacturer.Thenumberoftherollsthatfailbeforetheminimumlifelengthcanbespecified.Asastandardthisnumberis10%.Ifonasystemusedrolls,insteadofnewrolls,areinstalledthentheprogramaccountsforthiseffectbyallocatingremaininglifelengthstoindividualrolls.BearingsThelifelengthofaspecificbearinginarollisallocatedviaatabularizeddistribution.Underandupperlimitscanbespecifiedassumingauniformdistribution(minimumandmaximumlifelengthasspecifiedbyrollandbearingmanufacturer).Thechanceoffailurebeforereachingtheminimumlifelengthcanbespecified,againaccordingtoauniformdistribution.Alldistributionscanbechangedforthemiddleandthesiderollsofthecarryingaswellasthereturnidlersets.MaintenancerobotThemaintenancerobottravelsoverthestructureofthebeltconveyorinthedirectionfromtheheadtothetailataconstantspeed.Itisassumedthattherobotisavailable24hoursperday.InspectionOninspectionofanidlerset,thetotallifelengthestimationofanindividualrollisbasedeitheronhistoricaldata(statistics)orbasedonaccuratevibrationmeasurements.Thetotalinspectiontimeconsistsofafixedset-uptimeandtheinspectiontimeitself.Allrollsinoneidlersetareinspectedatthesametimeusingamultiplesensorrobotarm.ServiceIfthemaintenancerobotdecidestochangearollthenitisalwaysreplacedbyarollofthesametype.Thetotalreparationtimeconsistsofafixedset-uptimeandthetimeforrepairingor,inthiscase,changingouttheidlerroll.EstimationofresiduallifetimesTherobotestimatesthelifetimeofarollusingaFourieranalysisofthevibrationsintheroll.Forsimulationofthisprocess,amodelwith2parametersisused.Theestimationisasamplefromanormaldistributionwithasmeanthelifetimeoftheroll.Thedeviationofthisdistributiondeterminestheaccuracyoftheestimationandiscontrolledbythefirstparameter(d).Withthesecondparameter(f),abiasisintroduced.Theestimatorbecomesconservative,biasedtowardsunderestimatingthelifetime.TheestimatorS1isdefinedby:S1=L+d(L-A)X-f(L-A)WhereL,LifetimeoftherollA,CurrentageoftherollX,Randomvariable,sampledfromaNormal(0,1)-distributiond,Deviation,asfractionoftheresiduallifetimef,Safetyfactor,asfractionoftheresiduallifetime

AsampleofXisdrawnforthenormaldistributioneachtimeestimationisrequired.Iffequalszerothenestimatorisunbiased.Theprobabilitythattheestimatoroverestimatesthelifetimeis50%.Withf>0theestimatorbecomesbia