Զѧ
ACTA AUTOMATICA SINICA
1997 23 3 vol.23 No.3 1997



ԤֵƲԼʵ1
ͿζƽSheldon XC Lou
ժҪԻΪһʷֲĲɿϵͳһԤֵƲԺʵһƲԵĲŻ.ӦŶõָԿƲȹƣ֤˹ƵƫԣƽŻƲ.֤˿ƲԵЧ.
ؼʡϵͳŶƫƣƽ.
PREVENTIVE HEDGING POINT CONTROL AND ITS REALIZATION
TU FENGSHENG
(Dept. of Computer and System Science, Nankai Univ., Tianjin 300071)
SONG DONGPINGSHELDON XC LOU
(National Lab. of Ind. Control Tech., Zhejiang Univ., Hangzhou 310027)
(California State University)
AbstractWe consider a failure-prone machine whose up and down times follows general probability distributions. A policy called preventive hedging point control, as well as a parameter optimization algorithm, is proposed. Using perturbation analysis technique, the sensitivity estimate of the performance index with respect to control parameters is derived, and it is proved that the estimate is unbiased. Then, the stochastic approximation method is applied to determine the optimal control parameters. Simulations examples show the effectiveness of the policy.
Key words Manufacturing system, perturbation analysis, unbiased estimate, stochastic approximation.
1
вѧо˲ɿϵͳ13ݣǼĲɿһ״̬Markov̻̿Ĺʺ޸ʾΪ㶨.ף2֤˶һ̨״ֻ̬ӹһƷϵͳֵ(hedging point)Ʋŵģҵ˿Ʋľȷ.ͨĹʺ޸ʾΪ㶨ʱȼڻÿδ϶ά޵ʱͻÿڹ֮ʱ䶼Ӳ㶨ĸֲָ.ĽǻÿʧЧά޵ʱͻÿʧЧ֮ʱֱͬһĸʷֲҪΪֲָ.ȽһϵͳĽ˵HPƲԲŵģƹHPԵõԤֵ(PHP)Ʋ.ЧHPԣͬʱʵPHPԵһּ㷨.
2ģƲ
һֻ̨ӹһƷϵͳ.ǲɿģÿδϵʱֱΪ,ǵķֲ0E=,0E=b.ʼΪu(t)ʱڣ0,UUΪ.ƷǺ㶨ģΪd;ۻۻ֮Ϊx(t).ϵͳFlow Rate Control Model1£
x(t)=u(t)-d,(1)
0u(t)ܦ(t).U(2)
 (t)=1,ʱ(t)=0ʱ.Ŀ꺯Ϊƽ
(3)
x+(t)=max(x(t),0),x-(t)max(-x(t),0),c+,c-ֱǲƷǷȱķϵ.ͻΪѰҿƲu(t)ʹĿ꺯JﵽС.Ϊʹ壬ٶϵͳڳ㹻
a.U(a+b).d.(4)
ģ1HP£(t)=1,
(5)
hΪƲΪֵ.ƹɵõԤֵ(PHP)Ʋԣ
(t-0)0Ҧ()=1,ӡʣt0,tݣ
(6)
h,sΪƲ.ÿһʱڣt0,t0+sݳΪֵƽ׶Σt>t0+sΪԤƽ׶.
˵.
1)s=ޣPHP˻ΪHPԣˣPHPHPƹ㣻ܦ,ΪʲôֲPHPԵĿƲhsŻЧضHPԣǿƲռ.
2)Ρa,ǡbʱϵͳȷԵ.ָ(3)(4)£ѵſƹɾPHPԣh=0,s=a-bdc-/(U-d)(c-+c+).
3)ΣΪһĸʷֲʱſƲӦǷʽģԤ¼ķΪԤķ.PHPеԤƲʵԤϷ.ûܸPHPԵ֤ΪPHPڸΧԤ˼룬˱HPԸӽſʽ.
3ŻʵPHP
ϵͳt=0ĳֵΪ x(0)=0,(0)=1.PHPԿ£ϵͳĲ·x(t)̦(t)ͿƲ=(s,h)Ψһȷͼ1ʾ.

ͼ1ϵͳPHPµһβ·
һβ·Ŀ꺯Ϊ,JT()=ELT(ȣ(t)).ܹݶJT()Ż=(s,h).Ŷ(PA)4ݵŵǻһβ·ܸJT()Ĺ.еǰҪƵƫ.
1.ϵͳ(1)(2)PHPԿ£JT()=ELT(ȣ(t))ΪĿ꺯

Ϊ֤1ȸ¼Ķ
1)ϡ(t)=1,(t)=0;
2)޸(t)=0,(t)=1;
3)ֵʱx(t)ӲhΪh
4)Ԥӻÿ޸ʼʱʱsʱΪԤ¼
5)S=:ڣ0Tʱڣĳһʱ̣ͬʱͬ¼ЦΪϵͳPHPԿµ.
ǣ0Tе¼ΪN¼ʱΪt0=0,t1,t2,tN=TΪ¼..
1.϶NһصķǸPأN=ޣ=0.
2.SĸΪ㣬PأءS=0.
̶hs΢СŶΪs+sЦsΪС.ͼ2SN޵ĳβ̵ı·x(t)Ŷ·x(t).ΪSNޣԶԳСĦsܿԱ֤ƹ켣x(t)Ŷ켣x(t)ȷ.Ȼ¼֮Ŷǲ.ǦxiΪi¼i+1¼֮x(t)Ŷxi=xi(.)-xi(.)óֵx0=0ɵs΢СŶʱϵͳŶ£

ͼ2sŶǰ·
1. 1)tiΪԤ¼ʱ̣x(t-i)=hx(t)ӲŶs(U-d),xixi-1-s(U-d);x(t-i)hxi=xi-1-s.Ux(t-i)h,xixi-1. 2)tiΪֵ¼ʱ̣xi=0. 3)tiΪ¼ʱ̣xixi-1.
֪Ŷxi(i=1,2,N)ǦsԺxi/s(i=1,2,N)릤s޹.ΪԤ¼Ϊʱ䳬sڣ0,TԤ¼ۼӴT/s.ڳconstʹãxiconst.sԦһ³.1ͱƹ켣Ŷ켣ԣѵõ
(7)
(ti-ti-1)+Ϊ(ti-1ti)x(t)0䳤ȣ(ti-ti-1)-Ϊ(ti-1ti)x(t)0䳤ȣʽ릤s޹.
3.ϵͳ(1)(2)PHPԿ£
a.s.,(8)
(7)ʽ0(s,(t))
(9)
֤.12ֻ迼N޺ͲS.ʱȸأ(7)ʽȻ༴
(10)
sȸܿд(7)(8)ʽʽܱ֤(10)ʽԦһ³.һ⵽ƻԭ¼ʱСڦs(x(t)ŶĲֻŶʧĲȫ).,ΪĲһ棬(10)ʽһ⵽ƻע⵽xiconst.sԦһ³ԶN޺ͲSĦһµУ0(s,(t))/sC,CΪһ.
(11)
֤.
̶s,hŶΪhhЦhΪС.IΪӻ״ηֱֵT.ǣi.ΪʾԺ.ƿɵõ24.
2.ӻ״ηֱֵTʱx(t)ŶΪhʱΪ0.
4.ϵͳ(1)(2)PHPԿ£
a.s.,(12)

(13),(14)
1֤.(8)ʽͬԦsȡѧæs0õ(a);(b)(7)ʽֱӵõ.(12)(13)ʽƴ֤ڿƲhĽ.֤.
1(t),2(t),k(t)Ϊ(t)һϵвɵ¿޷ƣǿɺͶ1T/JT/ȵƫ.ȻõĿ ڿƲݶȹƣͿƽŻ.һƽŻ㷨£5ݣ
JT()=ELT,(t)),k+1k-kdk.dkΪJT(k)һƣkΪһзǸĲϵkʱЦk0.ڱĸϵͳŶõJTkһƫƣdkT=(T/h,T/s)ͿŻh,sʵPHP.
4Ӻͽ
k=100T=360U=2,d=1c+=1c-=5..б1.
1PAŻPHPHP

PHPHP
shT/sT/hThT/hT
18.330-0.00800.67550.277010.01391.1625
216.51-0.0069-0.03561.80291.30.01791.8437
37.8500.00810.53680.76841.10.01281.5198
41.7--0.00513.34311.7-0.00513.3431

ע.1Ϊ=10,=2ȷԷֲ.2Ϊ-U(0,20),-U(0,4)ľȷֲ.3Ϊβֲ̬Ρ-N(10,1),ǡ-N(2,1)Ϊֲ̬.ЦΣǶΪ=0,Ρ䣼0;=Ρ,Ρʣ020ݣ=20,Ρ䣾20.=0ǡ䣼0=ǡ䣬ǡʣ04ݣ=4,ǡ䣾4.4ΪΡ0.1e-0.1t,-0.5e0.5tֲָʱPHP˻HPԣs=+.
PHPHPԣرǵϵͳԲ̫ʱ13.PAƲȣȻŻƲķ㷺һϵͳұȵķ淨Լ˴ĹPAҪT.
1)õȻѧ͹ҽίʿר.12th IFAC World Congress (Sydney, 1993).
߼飺Ϳ216.
ζƽ1967.1992Ͽѧϵͳѧϵ˶ʿѧλΪ㽭ѧҵƼصʵҽʦְʿ.Ȥоϵͳɢ¼̬ϵͳ.
ߵλͿϿѧϵͳѧϵ300071
ζƽ㽭ѧҵƹصʵҡݡ310027
Sheldon XC Lou:ݴѧ
ο
1Kimemia J G, Gershwin S B. An algorithm for the computer control of production in flexible manufacturing systems. IIE Trans., 1983, 15(4):353-362
2Bielicki T, Kumar P R. Optimality of zero-inventory policy for manufacturing systems. Operations Research, 1988, 36(4):532-541.
3Sharifnia A. Production control of a manufacturing system with multiple machine states. IEEE Trans., Automat. Contrl. 1988, 33(7):620-625
4Ho Y C, Cao X R. Pterturbation analysis of discrete-event dynamic systems, Boston:Kluwer, 1991.
5Rubinstein R Y. Monte carlo optimization, simulation and sensitivity of queuing networks. Wilay & Sons, 1986.
ոڡ1995-09-07
