Class7技術(shù)-2風(fēng)險分析的方法體系

TheToolsandMethodologiesof

RiskAnalysis工程項目風(fēng)險管理與保險雙語課ContentsofRiskAnalysisAnalysisofriskeventsevolution;RiskEvaluationfor:

Ⅰ

occurringprobabilityofriskevents,Ⅱvariedpossibleriskconsequences,Ⅲandtheirconditionalprobabilitiesgivenoriginaleventoccurring.

AnalysisofriskeventevolutionFaultTreeAnalysis(FTA,故障樹)EventTreeAnalysis（ETA，事件樹）Whatisafaulttree

？Isaspecialkindofinvertedtreelogiccausalitydiagram,itUSESeventsymbols,transferoflogicgatesignsandsymbolstodescribeacause-and-effectrelationshipbetweenvariouseventsinthesystem.Logicgateinputoutputevents\"because\",logicgateoutputistheinputevents\"fruit\".name

symbolCausalrelationship

WiththedoorInputalleventsoccuratthesametimehaveaoutputOrgateTheinputaslongasthereisaneventoccursthatisoutputTheforbiddendoorTheinputconditionaleventisoutputOrderthedoorInputalleventsoccurinfromlefttorightorderisoutputXorgateTheinputeventiswhenonlyoneeventintheoutputsymbolnamemeaningcircularBasicevent,thereareenoughrawdatarectangularBylogicgatesexpressedeventoffailureThediamondForreasonsunknownfailureeventsDoublediamondHaveanimpactonthefaulttree,needsfurtherstudythecauseofthefailureofunknowneventTheroomtypePossiblefailureeventsmaynotappeartriangleConnectionandthesymbolsTheprocedureoffaulttreeanalysis1.Befamiliarwiththesystem2.Investigateaccidents3.Determinethetopevent4.Determinethetargetvalue5.Reasonsforinvestigation6.DrawthefaulttreeTheprocedureoffaulttreeanalysis

7.Analysis8.Theaccidentprobability9.Comparing10.Analysis

Supervisionofprojectfailure

Fromtheowner'srisk

Fromthecontractor'sriskFromtheriskofsupervision

X1X2X6X3X4X5X7X11X10X8X9ACase:TypicalriskX1:feasibilitystudyreportlackofseriousnessX2:macromanagement,investmentdeficiencyTheX3:blindintervention,deprivedoftherightsoftheengineeringconsultantX4:don'trespectcontract,norespectfortherightsandinterestsofthecontractorX5:projectpaymentdoesnotreachthedesignatedpositionorsettlementinatimelymannerFromtheowner'srisk

X6:contractorbiddingnothonestBiddersX7:lackofbusinessethicsTheunder-fundedbyX8:contractorsX9:contractorqualityistoolowFromthecontractor'srisk

EventtreeanalysiscaseThewatersystemoftheaccidenteventtreeanalysis

ProbabilityMethodsUsingtheexistingdata

analysisriskeventprobabilitydistributionTheoryofprobabilitydistributionisusedtodeterminetheprobabilityofriskeventsTheprobabilityofriskeventsusingsubjectiveprobabilityanalysisAnalogyisusedtoderivetheprobabilityofriskevents

Usingtheexistingdataanalysisriskeventprobabilitydistribution

Case:theemergenceofconstructionperioddelayriskprobabilitycalculationAconstructioncompanyinthepasttocomplete72constructionprojects,duetovariousreasons,somepartoftheprojecthasdelayedtimelimitforaproject.Draggedprojecttimelimitforaproject,arrangethemgetsituationasshowninthetable.Thetimelimitforaprojectinthetablearetherelativevalueofdelaytime;Thenumberoffrequencyfortheproject.Theconstructionperioddelaydatastatistics

Datapacketinterval（%）Inthegroupvalues（%）frequencyfrequency（%）Thecumulativefrequency（%）-34~-30-32.500.00.00-29~-25-27.522.782.78-24~-20-22.511.394.17-19~-15-17.534.178.34-14~-10-12.579.7218.06-9~-5-7.51013.8931.95-4~0-2.51520.8352.781~52.51216.6769.456~107.5912.5081.9511~1512.5811.1193.0616~2017.545.5698.6221~2522.500.0098.6226~3027.511.39100.0131~3532.500.00100.01Timelimitforaprojectdelaydistributionexperience

-30-25-20-15-10-5051015202530%Frequency(probability)

0.2083Estimatetheconstructionperioddelayeventprobability

Forexample,acompanyplannedconstructionperiodof16months,ifyouwanttoknowtheprobabilityofconstructionperioddelay3months,shallbecalculatedonthefollowingsteps

：①3/16×100%=18.8%②Fromthetableorfigurecanchecktheconstructionperioddelayistheprobabilitythat3months

5.56%。Theoryofprobabilitydistributionisusedtodeterminetheprobabilityofriskevents

Casestudy:Qualitydataarevolatile,thefactoriscausedbymanyfactors,andundertheconditionofnormalproduction,theroleofeachfactorisnotobvious,andthereisnoonefactorplaysaleadingrole.SothefluctuationsNormaldistribution.Intheconstructionofwaterconservancyprojects,manyhydrologyworkersstudyofriverfloodpeakflowofrandomness,foundthatmostoftheriversobeyedP-Ⅲtypedistribution.Theprobabilityofriskeventsusingsubjectiveprobabilityanalysis

Casestudy:Aprojectisdividedintofivestandard,itemizedbid,abidderintendstoparticipateinoneofthebid,assumethatabidderthewinningprobabilityisdifferentforeachtarget,namelynotthebiddingriskisdifferent.Bidderspleasefourexpertofbiddingriskanalysis,evaluationforwhichstandardcanbidmore,lessriskyormissthemark.Expertsontheriskofnotwinningfromsmalltolarge,theresultofthedischargesequenceasshownintabletenderingdecisionmakersfortheexperts'trustdegreebycoefficienta,were0.25,0.28,0.22,0.25,andfourNumbersand1.Expertstotheevaluationorderofdifferentprojects專家

名次12345aAⅠⅢⅤⅡⅣ0.25BⅢⅡⅠⅤⅣ0.28CⅡⅠⅢⅣⅤ0.22DⅢⅡⅤⅠⅣ0.25Ⅰ：1×0.25+2×0.22+3×0.28+4×0.25=2.53Ⅱ：1×0.22+2×0.28+2×0.25+4×0.25=2.28Ⅲ：1×0.28+1×0.25+2×0.25+3×0.22=1.69Ⅳ：4×0.22+5×0.25+5×0.28+5×0.25=4.78Ⅴ：3×0.25+3×0.25+4×0.28+5×0.22=3.72Differenttargetscannotbethewinningprobabilitycalculation：Thetotalscore：2.53+2.28+1.69+4.78+3.72=15Can'tthewinningprobability,respectively：2.53/15=0.172.28/15=0.151.69/15=0.114.78/15=0.323.72/15=0.25Conclusion:Formostatriskwhenthefourthstandard,forthethirdwhentheriskisminimalAnalogymethodtocalculatetheriskprobability,hydrologicanalogymethod

Casestudy:TobuildAport,tocomputetheprobabilityofAonce-in-a-centurystormsurge,butlackofMarinedisastersstatistics.AndBislocatedinhasnotfarfromthedetailedhistoricalMarinemeteorologicalstatisticaldata,socantheregistrydataisusedtocalculatetheprobabilityofoccurrenceofstormsurge,andcomparewithAriskofprobability:Pa/Pb=ɡ×(Ha/La)/(Hb/Lb)

Amongthem:H-significantwaveheight;L-wavelength;G-comparedifferencecoefficient.

BayesMethods-基本步驟確定先驗概率-根據(jù)歷史資料或主觀判斷，未經(jīng)實驗證實所確定的概率。進(jìn)行預(yù)后驗分析，決定是否值得搜集補充資料以及從補充資料可能得到的結(jié)果和如何決定最優(yōu)對策。BayesMethods-基本步驟搜集補充資料，取得條件概率，包括歷史概率和邏輯概率，對歷史概率要加以檢驗，辨明其是否適合計算后驗概率。用概率的乘法定理計算聯(lián)合概率，用概率的加法定理計算邊際概率,用貝葉斯定理計算后驗概率。

發(fā)生的條件概率公式為：

表示在一個樣本空間中的兩個事件，給定和下，和的聯(lián)合概率公式為：BayesMethods-貝葉斯定理中的一個出現(xiàn)是事件構(gòu)成互斥和完整的兩事件，和概率的邊際概率公式為：和若發(fā)生的必要條件，那么事件回總目錄回本章目錄BayesMethods-貝葉斯定理中的一個出現(xiàn)是事件構(gòu)成互斥和完整的兩個事件，和事件的貝葉斯公式為：和若發(fā)生的必要條件，那么兩個兩個事件的貝葉斯定理為：回總目錄回本章目錄Ai中的某一個出現(xiàn)是事件B個事件的貝葉斯公式為：假定存在一個完整的和互斥的事件發(fā)生的必要條件，那么n

n個事件的貝葉斯定理為：回總目錄回本章目錄

優(yōu)點：

（1）貝葉斯決策能對信息的價值或是否需要采集新的信息做出科學(xué)的判斷。（2）它能對調(diào)查結(jié)果的可能性加以數(shù)量化的評價，而不是像一般的決策方法那樣，對調(diào)查結(jié)果或者是完全相信，或者是完全不相信。

回總目錄回本章目錄BayesMethods-優(yōu)缺點

（3）如果說任何調(diào)查結(jié)果都不可能完全準(zhǔn)確，先驗知識或主觀概率也不是完全可以相信的，那么貝葉斯決策則巧妙地將這兩種信息有機(jī)地結(jié)合起來了。（4）它可以在決策過程中根據(jù)具體情況下不斷地使用，使決策逐步完善和更加科學(xué)?；乜偰夸浕乇菊履夸?/p>

局限性：（1）它需要的數(shù)據(jù)多，分析