Ϣ
INFORMATION AND CONTROL
1999 28 3 Vol.28 No.3 1999



ھƵŴ㷨Ӧ
ʥ˵»
ժҪMonte Carloķ룬ڸ÷Ŵ㷨иĽ𽺲ƷеӦʵʵŴ㷨ģŻ֮Чģ
ؼʣŻ,,Ŵ㷨,䷽
ͼţTP13ױʶ룺B
GENETIC ALGORITHMS BASED ON UNIFORM DESIGN
AND ITS APPLICATION
GAO QishengPAN Dehui
(Institute of Information Science and Technology, Northeaster University, Shenyang 110006)
AbstractBased on uniform design technique, a new genetic algorithm, UDGA, is proposed, and it is applied to solve a complicated optimization problem that appears in rubber prescription. Experiment results show its validity. 
Key wordsoptimal design, uniform design, genetic algorithms, rubber prescription
1
Ŵ㷨(Genetic algorithms: GA)MichiganѧJohn Holland60[1,2]һȻӦŻ㷨ǻȻŴȻѡѰŷνȻŴȻѡڴĵĽѧ˵ѧ˵ΪУһֲﾭɴŴͱ죬ʹܹ֮ӦµĻʤ̭ĽȻŴ˼ҲŻ⣮
GAѡ(selection)(crossover)ͱ(mutation)ʵ֡ﾺ,桱һȻģ⣬չѸ٣Ŀǰоͳı׼Ŵ㷨(SGA)ȫŽ⣬SGAĸĽ㷨ѡǰѸGAܹȫŽ⣬ͬʱָȫŽҪʱܺܳGAЧĿǰ⣮
ĻھƷһܹЧЧʵŴ㷨UDGA(genetic algorithms based on uniform design)𽺲ƷеһӦʵ
2
ơ80йѧԺѧԪͷ̩ڽۺͶԪͳƷϴһӱ鷽[3]ǵӾԭƣԶƫ(discrepancy)
(1)
Уv(X)Ϊ[0,X]mXΪX1,X2,,Xm[0,x]ĵCnΪĿռ䣮
ͳƵ鷽ơȣ٣ܵԱձ黶ӭƵĺ˼ȷԷѰҿռоȷֲĵ㼯Monte CarloеMonte Carloķ룬ɿͳƳЩGAĽ֮ݣ
ƵʵʩһȱUm(mn)ɣ mΪ nΪ (
)Ժøӵ㷨(good lattice point)Ϊ˵Ԫuijɹ̣
һ(Ⱥģ)m,ѰұmСhʹ(m,h)=1,ЩһH=(h1,h2,,hn);
ڶȱĵj淽
(2)
[mod m]ʾͬ㣮m=9ʱH(1,2,4,5,7,8),ɾȱ(1)
1U9(96)

123456
1124578
2248157
3363636
4487215
5512784
6636363
7751842
8875421
9999999

3ĽŴ㷨UDGA
3.1ʼȺ
Ⱥģmm=|A(0)|, A(t)ʾtȺ(ĸ)ʼȺA(0)ɲþUm(ms)sΪ
3.2
tȺA(t)еiΪ
(3)
 aij{1,2,,m}(I=1,2,,m; j=1,2,,s).Ȼ˱벻ǵȳģҲʵֵͱӦϵΪ
(4)
[xjmin, xjmax]ΪtxjΧX(t)={xij}ms
3.3Ŵ
1) ѡ
fi, ΪȺиӦȺȺӦƽֵѡȺӦȵи
(5)
һΪȡ㣮A(t)ʵʲΪ
(6)
2) ͱ
ʽСԭͱӦȽǿĻнУǻX(t)(A(t))γɵռʵֽͱ죮
(7)
ѡþƱUp(ps)p壬Ϊ
(8)
A(t)ΪǾͱɵ¸壬ǶӦʵʲΪ
(9)

(10)
A(t+1), x(t+1)һȺıʵʲ
3.4ֹͣ׼
(11)
f*ʾȺѸӦȣ1,2ΪԤȸľȣ
UDGA£
һt=0ѡ1,2,m,Um(ms)ɳʼȺA(0);
ڶ۵tȺA(t)Ӧ
(5)(10)ѡ񡢽ͱһȺA(t+1);
Ĳ(11)ֹжֹת岽,tt+1,תڶ
岽ĸ
ϿɼھƻϵŴ㷨(UDGA)ص㣺
1)ɽŴģѡȡ
2)ÿһŴܱ֤ǰŸԸ1תƵһȥ
3)ÿһȺγɵ򲻶ЧGAȫŽٶȣ
4Ӧʵ
ĳ𽺳ڿչQC(Quality Control)Уѡ񶡻𽺲Ʒ䷽ƿչĿڱ֤ƷҪǰ½ͲƷĳɱҪ仯ɷΪǿ̼ڡsʹٽTMTDֱ֣Ϊx1,x2,x3,շΧΪx1[20, 40],x2[0.8, 2], x3[0.8, 2.2]ĲƷָгǿy1,쳤y2,˺ǿy3ϻϵy4,ό(120)y5300% y6ñy7T90(168)y8 ʱ׼Ϊy*1=12.00, y*2=600.00, y*3=40.00, y*4=0.80, y*5=37.00, y*6=30.00, y*7=33.00, y*8=11.00,ǰ6ΪͣΪͣ
(12)
yi(x1,x2,x3)ƷiƱ֮Ľƺϵͨͱʶģ͵õyi(x1,x2,x3)ṹ£2
2ģϵһ

b0b1b2b3b11b12b13b22b23b33
y11753.74-36.6-55.80.6764-0.10360.31644.5-0.878613.5
y260717.1-59.4-11.2-0.3071-0.27660.184629.96-0.1558-1.46
y355.41.91-11.1-20.6-0.05170.36890.36920.92-3.51441.69
y40.96-0.030.170.430.00030.00350.0012-0.120.0373-0.18
y530.2-1.132.5834.10.01320.15530.0773-4.490.4349-9.46
y641.2-1.2821.0-11.00.0215-0.33820.3428-6.526.5895-0.71
y727.00.43-0.42-10.70.00070.0307-0.1320.191.31795.24
y820.7-0.093.32-14.3-0.00110.03840.0725-0.65-0.113.10

(13)


(14)
MΪͷӣڱУM=100, =1ˣ(13)ȼʽ
(15)
ӦȶΪ
(16)
ȡm=30, 1=0.005, 2=0.00510UDGAõΪx*1=29.00, x*2=1.04, x*3=1.21.
5
ھƼһ͵Ŵ㷨UDGA÷Ӧõ䷽ŻУʵ÷⸴ӵĴģŻʱЧؼӿٶȣʵӦã
ߵλѧϢѧ빤ѧԺ110006
߼飺ʥ32꣬ʿڣоΪŻƼӦã
˵»ݣ70꣬ڣʿʦоΪӦѧԶۼӦۣ
ο
1Holland J H. Adaption in Natural and Artificial Systems. Univ of Michigan Press, 1975
2Goldberg D E. Genetic Algorithms in Search. Optimization and Machine Learning, Addison-wesley, 1989
3Ԫ̩. ھȷֲ(۷). ѧͨ 1981 26: 6570
4̩. Ʊ. ѧ磬1994
5ʥ˵». ӦTQCĳƷ䷽. ѧ(¼ã) 
6ΰʥ. ӦþŻ䷽. 𽺹ҵ 1997 44(1)
7ʥ. ĽSA䷽ŻеӦ. ϳ𽺹ҵ1998, 21(2)
1998-04-08ո
