北航iSIGHT软件培训资料

1、iSIGHT720863/CIMSEngineousiSIGHTISIGHTISIGHT1.3.4.5.iSIGHTiSIGHTiSIGHT1SIGHTISIGHT1SIGHT2 41SIGHT2 51SIGHT1SIGHT#1SIGHT1SIGHT1SIGHT21.CAD CAEExcel ISIGHT1SIGHTFortran C+ VisualBasic Unix1SIGHT3ISIGHTISIGHT4541SIGHT1SIGHT name-valueCORBAISIGHT(FDC)MDOL1SIGHTInput fTSimulation-basedProcessesOutputklEx

2、plore EnaineISIGHTISIGHT1S1GHT1SIGHT1SIGHTIPC InterprocessCommunicationiSIGHTCuMcmTd7Case1SIGHT1SIGHT1SIGHTMSC/NASTRANIfWhileTel#ExcelExcelMD0L1SIGHTExcel 1SIGHT1SIGHT Visual BasicISIGHT1SIGHT1SIGHTEPCCORBA1SIGHTKIDOL720863/Cl 胎WDONational Aeronautices & Space Administration NASA Langley Multidi

3、sciphnaiy Optimization Branch NfDOB Multidisciplinaiy Design Optimization 1 Multidisciplinary Design Optimization (MDO)ls a methodology for the design of complex engineering systems and subsystems that coherently exploits the synergism of mutually interacting phenomena.MX)MDOHjI t i Di sci pl i nary

4、Feasi bl e血FAl I -1 n- OneA-l-0I ndi vi dualDi sci pl ne Feasi bl eI DFGSEConcur r ent SubSpace Opt i ni zat i oncssoCol I abor at i veOpt i rri zat i on CO BL I SS Bi - Level I nt egrat ed Syst emSynt hesi sBL I SS/ RSM9#1.M)F All-In-One#LOOmil t i di sci pl i nary anal ysi s血AXoXoM)A#XoXd U Xd#g(X

5、°0(XJ)WOXoLDA11IDFIDFI DF9jz9jzIDFIDFF( XD. U( X)X=( XD, X )g(XD,u(x) oc(X) = Xm - m = QXD9jz9jz3.cssoCSSOMX)csso4.coJj = C； <0.0001 j -1,2?2 IDF血ASobi eski(CO)(CO)CSSOCSSOGSECSSORenaud Bat i 11CSSOKr oo h0MX)Fxiyj9jzXsj,Xj ,J s<=|X?.-Z； L +匕z;T，zsZfcococo5 BL I SS Bi - Level I nt egrated

6、 Syst em Synt hesi sBLISSBLISSt he opt i num sensi t i vi t y anal ysi s dataBLISSI SI GHTBLISSBLISSA-1-0ZBLISS1./Y2.sensi t i vi t y anal ysi s, syst emopt i rri zat i on.BLI SSsyst em anal ysi sI ocal di sci pl i nary opt i rri zat i ons6 BLISSRSMBLISSbl ack box/di sci pl i naryBLI SS RSMBBOPTBLIS

7、S20BLISSBLI SS/ RS1BLI SS/ RS2BLI SS/ RS1BLI SS/ RS2BL I SS/ RS2syst em anal ysi s SABl ack- Boxsensi t i vi t y anal ysi s BBSASyst emsensi t i vi t y anal ysi s SSABl ack- Box opt i ni zat i on BBOPTsyst emopt i rri zat i onSOPT(Opt i rri zat i on)720863/CMi SIGHTi SI GHTwm士 o加(LB_gj(X)<0 SF;SF

8、LB iSlGHTlnputParameter / UBSFSSFW1.01.0i SGHTi SIGHTh(x)± DeltaForEqualityConstra int Violation0.00001i SIMiSIOITXi SI O1Tg(x)xX1E15i SIGHTI SI加APIi SIGHTi SIGHTNiuncncal Optimization Tecluiiques Expl ora toiy TecluiiquesExpert System Tecluiiques ADS (Automated Design Synthesis)-bascd Tecluiiq

9、ues Exterior Penalty Modified Metliod of Feasible Directions ( Sequential Linear Progianuiiing)(Generalized Reduced Gradient LSGRG2)Hooke- Jeeves( Hooke-Jeeves Direct Search Metliod)一 CONMIN Method of Feasible Directions 一 CONMIN- MOST Mixed Integer Optimization - MOST一 DONLP Sequential Quadiatic Pr

10、ogramming- DONLP一 NLPQL Sequential Quadratic Prograimiiiiig 一 NLPQL Successive Approximation MetliodDirect metliods ( Generalized Reduced Gradient LSGRG2) - CONMIN Metliod of Feasible Directions - CONMIN 一 MOST Mixed Integer Optimization 一 MOST Modified Metliod ofFcasiblc Directions ADS ( Sequential

11、 Linear Progranmiing- ADS) 一 DONLP Sequential Quadratic Prograiimniig 一 DONLP 一 NLPQL Sequential Quadratic Prograiimiiiig 一 NLPQL Successive Approximation Metliod2Penalty metliods(!)=尸(丄)+ P(x)G)# Exterior Penalty Hooke- Jeeves(Hooke-Jeeves Diicct Search Mctliod)2i SIGHT Genetic Algoiitlun Genetic A

12、lgontlun witli Bulk Evaluation Simulated Aimealing3i SIGHTDirectedHeuristic Search DHS(Exterior Penalty)MLP(x) = 7p m"(9j(5),0)2+7p 工j = 1Jc= I(Modified Method of Feasi bl e Di recti ons)1. a = 0.x = x°尸(卫 gj j = i,2,.,a/jVF(x) Vgj(x) jeJ£a *VF(xql)xSqV(x)xr <O;yeJSq xSq <1PF0)7

13、 一卩Vg, 0) x Y +50;jwJ$ V VIJ60y =007 >0CT<(x)<CTM/N g&)3.4.5.6.7.8.9.123Sj(x)>CTM/N Sj(x)ox(6U Ee6Od ss 二-e 二 uenbs)cosUJf-l*勺 QA+(J) d 川9G 划二。卫A + (%|订川)42LSGRG2尸(X)2(LSGRG2)hk(X) = 0k=MXLi< Xi < XUii = 1,"Xj + n >0j = 5nmmgj(X) + X/ + /? = 0 j = l.mn+m17LSGRG2GRG#5 Ho

14、oke-JeevesHooke-JeevesfO(Hooke-Jeeves Di rect Search ftfethod)fOr ho,fOHooke Jeevesrho1 g = 0,x = x°2 q = q +13.竺/(M)f（M）6.-CONM NCOWNDesign： = Design + A* SearchDirectioniiA# CONMN19 CONMN1.q = 0.x = x°# CONMN# CONMNq = q+# CONMN# CONMN吃)gj(x)j = 1,2,、M# CONMN# CONMN4.5.VF(x)S°7.a*8.

15、9.# CONMN-CONhflNVgy(x)x50;jeJSqxSq <1-CONhllN阻也5丫+ 60SO;j"SqxSq <160y =0®. >0CT <y(x) < CTMIN gy(x)#gj©>CTMlN gj MOSTMOST Multifunctional OptumzationSystem Tool MOSTbranch-and-bound1SIGHT1SIGHT-NLPQL-DONLPSQPXXXv EN N+l1SIGHTSQP8. DOhLPDONLPK-TDONLPLagr ange Hessi a

16、n Pant oj a- lAynear ni j i -1 ype-DONLP(DO NonLinear Programming) Kuhn-Tuckei* DONLPLagrangian HessianPantoja-Maynear ni j i t ypeDONLPX= (XpX-,a:n) 弘)g/(I)= o；丿=1,gf(x)>Oj = me+ly.9mE 2 2“Vhk (xk)T x S + hk (xk) = 0; R = 1,2,.,厶(凶 Tx§+g 赵(xk) = QMPJ:J(E1,2,.,M 沁)509. M.PQLNLPQL(SCP) NLPQL#

17、Hessi anBFGS#NLPQLx=(.v1,x2,.xN) 化)幻= 0;_/ = l,.,九Xt X<XUSQP#SQPSQP21#L(x. u)SQPB,Hessian-dTBkd + Vf(xky d deR" 2Vg j g r d + g j (E ) = 0 J = 1,.,叫，Vg j (耳 r d + g / g) n o j = i,.叫，xt - xk <d <xu 一 忑#NLPQLdkSchittkovzskixk = xk+akdkB,SQPBFGS血+i -忑 VL(xi+1,ha ) - VE(无，叭)NLPQLB#SQPKh-科

18、|M儿卜上-巧7kSQPn+mn+mKulin-TuckerHessianBk(Successi ve Appr oxi mat i on Kfet hod)10.#M Ber kal aar and J. J. Di r ksLP- SOLVE1SIGHTdnvejpsolve cmaxminv, s,=,n, >< < > >int < var > +尸hk (x) = 0;k = 1.2,.,Lxx,i = ly.,n g.hk环叫-士罟叫+ £薯z1=1 1=1003 1SIGHT100101100100231002510010030

19、(0)Hol I and 6015521552110000100021OOOI1OO1111OO11OI11OOO11O1OOOI110001100110110011110#1000100111101000100101101g op g2P93P9P9 19 9 1422(Genetic Algolitlun with Bulk Evaluation)i SI HGTi SIGHTBui kEval uat i oni SIGHT LDOLiSI GdT3 (SA)c（s）STTT恥t r叩ol i sss'c(s')<qs)Te(c(s )-c(s)T(hferkov)

20、i SIGHT问 ropol i sDHSi SIGHTDHSDHSDHSDHS27DHSDHSDHSDHSDHSX1X2X3XnnDHS180DHS Conpet eDHS Sequent i alSi nul t aneousSeparabl e and Si mul t aneousl nsepar abl eDHSRandom29LM)FModified Method of Feasible Directions - ADS Method of Feasible1关冨性崎也相关尺熾ABcDABcDR耳Tl.lX方同*+I¥f1t “Scq弘lR inssSIScqs'

21、-'-R:in1+Itf+f+1H.D4i rDHSDHSDHSDHS仁DHS2.DHS3.Gi vebackG vebackG vebackDHSGi veback#LM)FModified Method of Feasible Directions - ADS Method of Feasible#LM)FModified Method of Feasible Directions - ADS Method of FeasibleMMFDMMFDSLPSQPSQPHJSAMDHSGAMOSTLSGRG2Modified Method of Feasible Directions -

22、 ADS Method of Feasible Directions 一 CONLUN Sequential Linear Programming ADS Sequential Quadratic Programming -DONLP Sequential Quadratic Programming -NLPQL Hooke-Jeeves Direct Search Method Successive Approximation Method Directed Heuristic Search (DHS) Genetic Algorithm14ixed Integer Optimizauon

23、一 MOST Generalized Reduced Gradient LSGRG2Directions 一 CONMINSQPSequential Quadratic Programming 一 DONLPSequential Quadratic Programming 一 NLPQLGA Genetic Algorithm Genet i CAl gor i t hm wi t h Bui k Eval uat i on2Pen. 煦h.hWFSLPSQPHJSAMDHSGASi mAnnl M3STLSGRG2XXXXXXXXXXXXXXXXXXXXXXX20XXXXXX1000XXXX

24、XXXXXXXxXXXNLPQL X"3hWFSQPHJSAMDHSSi mAnnl.MOSTLSGRG2XXXXXX*XXXXXXXXKuhn-TuckerXXXXx-XXXXXXXXXXXXXXXXXX*Modified Method of Feasible Directions 一 ADSX*iSIGHTiSIGHT Coarse-1 o- Fi ne Search Est abl i sh Feasi bi I i t y, Then Search Feasi bl eRegi on Expl oi t at i on and Expl or at i on Co叩I err

25、ent ary Landscape Assu叩t i ons Procedural For mil at i on Rul e- Dr i veniSIGHTISIGHT1 Coarse-1o-Fi ne SearchSQP1SCP2 SQP1SQP22 Est abl i sh Feasi bi I i tyf Then Search Feasi bl e Regi oni SI GHT3 Expl oi t at i on and Expl or at i on4 Co叩I errentary Landscape Assurrpti ons5 Procedural For mil at i

26、 oniSIGHT6 Rul e- Dri veni SI GHT31Approximation720863/CIMSL. A. Schrrl tAppr oxi mat i on Concepts f or Ef f i ci ent St r uct ural Synthesis, NASA ContractorReport 2552, NASA Langl ey5 Wrch 1976(DOE)i SIGHTRSMRSMRSM尸(X) = 5 +工切兀NNF(X) = q + 工 b£ + 工 ctix; + 工 CjMZ=11=1v(f< /)N33RSMi SIGHT#

27、(N + l)木(N + 2)/2N + l,N#RSM#35R2R2i SIOfTR2R2 1.00NN+1R2TSA1234i SIGHTMDOL Gr adi enti SIGHTN+1i SIGHT尸(X) = F(X°) +工 g,(X°)(*-X)/=1X。£MTFF2F(x)=r(x0)+f gj(x°)a -x*)4r=lX3F(X) = F(X0) + E g,(X°)(f -站)0(f,对)1 ifxf.(Xo)>O0X，站)Xq4% = < r匸(X) = F(X°) + 丄士(x；)-匕(X J(x

28、V - (x；) i«iX?尸(XJ"(XJ = O537#N (J古人1N*i/!p.£(亦(XJ/%)_gj(XJ = OF(X1)-F(X1) = OEuler CFDNavi er St okes CFDNavi er - St okes CFDEul er CFDi SIGHTonoffObj ect i veAndPenal t yRSMTSA VCMTrr _ F _ F? _ Actual change in Obj ectiveAndPenalty -A2 Approximate change in Obj ectiveAndPenaltyF1

29、F2A1 A2 TRR1.0 TRR0.0 TRRTRR TRR TRR TRR1234567TRR#103RSM39u7 1720863/CIMS#u7 1#u7 11SIGHTMont e Car I oTaguchi sSi x Si gmaLbnt e Carl o#u7 1#u7 1#u7 1#u7 1Monte CarloM)nt e Car I oi SIGHT血 nt e CarloMont e Carlo110,000(1,000)345/63527Saliby, 1990( )vari ance reduct i on t echni ques1SIGHTLatinFigu

30、re 7-/. Munte < urlo Sampiinif TeehHiques41u7 1#u7 1Simple Random SamplingDescriptive Sampling#u7 1#u7 11SIGHT Monte CarloMonte CarloTaguchis X (Z)y yTaguchi#7 2-2. Prifdriut AnyftH Rcjwnh Mean it nd I aritiuce Kstiinatio/iMo«fi Anay1234afswO-IW3yl3yz3ywy« a w a y V y V fr41 y y y yn n-

31、 n.x1, x2, x3?z1,z2.z3,*12211y>2yiyi7Taguchi72SI Nyi7 2L»iS/Nnoni nal i sbestI oweri s betterhi gheris bett erS/N7 143Table 7-/. 5ZV RufivRespond ly|M-S/N RmioNofniiial h hcvt:J4V y V 81° %；< eLvuct if better0 V y V 8loginI n .-I 丿HLghcr is bettenf 1 ft 0<y <oo-I0log1(1A告片/i SI

32、GHTNominal is bestS/N-g吩辛i#-IGIoy.丄f厂加小I1"仃Lower is betterHigher is betterHigher is betterLower is better max|y| Nominal is bestTaguchi100%9.910.r9.9”10JM100%i(rTaguchiFignrti 7-J. Taguchi % Lw F u fictionpared io Coitventiotuil OitalityAementIMLu *l IIM(onv cnlional(No I oss Rm45#L(y)yTalyorL(

33、y)=k(y-T)2Ty-t1.0Response lypcLoss FunctionNtHiiinal is best:-£心寸 8 V < «JLower is Feder:n IUH 8Higher is better:丄办丄以幺y：0 S v S «S/N"Lower i s bett er”MH gher i s bet t er”0 yi SIGHTResponse lx pe1 -Oss FunctionLcnvii i、belter;1亍如后II8 < V < OOHigher is butter:丄f ”用卜18 &l

34、t; V 00i SI GHTObjective (0lll|MHWntRvsjHime 1)卩 tPvrfbrnuinctCharacteristicYlfamj 4S- N RitioY2(any)Loss ruiKliuiiY3Higher is BetterMcuii and XnfuinccY4Lower is rUttcrMean and VnrijinxrcY5Nottiinal is Bc、【Mean Only S/ Nyi + Ly2*Y3 +2Y3+Y4+<+(Y5* TY5) 247Taguchi1.0Six Sigma#iSIGHTSix SigmaTaguchi

35、Six SigmaiSIGHTSix Sigma#si gma0.683#F0n91ut Distr/hirtuHi for Six Siwa-60 -5ct -4a *3cr -2n -1n p *ln +2n +.kF *4a +5ct +octsi gmaTuble 7-2. Argrwrf Z.rr/ .4*hifHi/ww n/ic/ Defect Per MillionSima Le> vlPercent VariationDew per millioiiDefects per millionbh«rl lerml(lonu (ertn)±ln6H263

36、m)6Q7.7(IO£954645,400308,733亠3(7M.732.70066.8(13±4<t4 9937636.2<M)99.QQQO430.57233i为如恥i朋0.0023.47 433399.73%36Harry, 1997si gma叩 m7 232700/ ppm100. 1010.051.5si gmasi gma7 2si gma1.5366?8036Si x Si gma0.0027 2si gma3.466DFSSDesign For Six SigmaHarry,1998DFSS6Si x Si gma (6 )7 57 5a7 4336100%17 5bVxEninernn: Design fttr Srx-Sifintaa 3g Designb» 6