Sie sind auf Seite 1von 6

28 4

2011 4

Control Theory & Applications

Vol. 28 No. 4
Apr. 2011

: 10008152(2011)04056106

1,2,3 , 1,2 , 3
(1. , 110004;
2. , 110004; 3. , 110004)

: (robust collaborative optimization, RCO),


(non-dominated sorting genetic algorithm, NSGA--II)RCO. NSGA--II,
, . , ,
, ; , ,
. , RCO.
, .
: ; NSGA--II; ;
: TP301.6
: A

Multidisciplinary robust collaborative optimization based on


non-dominated sorting genetic algorithm
LI Hai-yan1,2,3 , MA Ming-xu1,2 , JING Yuan-wei3
(1. Liaoning Province Key Laboratory of Multidisciplinary Optimal Design for Complex Equipment,
Northeastern University, Shenyang Liaoning 110004, China;
2. Key Laboratory of Integrated Automation of Process Industry, Ministry of Education,
Northeastern University, Shenyang Liaoning 110004, China;
3. School of Information Science and Engineering, Northeastern University, Shenyang Liaoning 110004, China)

Abstract: To the robust collaborative optimization(RCO) scheme with two-level multiobjective optimization structure, a solution
strategy employing the non-dominated sorting genetic algorithm(NSGA--II) is proposed. In the process of non-dominated sorting,
the feasibility of an individual is determined by its infeasibility degree and the threshold of infeasibility degree. The threshold of
infeasibility degree is reduced gradually in the process of evolution. At the initial stage of genetic evolution, the individuals with smaller
values of objective function and standard deviation are more likely to be preserved to ensure the optimization process for reaching the
neighborhood of the global extremum. In the following stages of genetic evolutionthe individuals with smaller value of infeasibility
degree are more likely to be preserved to enhance the interdisciplinary compatibility. The convergence of the results of RCO to the local
extremum is usually avoided while keeping the desired interdisciplinary consistency. The results of validation by using typical examples
show that the proposed approach is efficient.
Key words: robust collaborative optimization; NSGA--II algorithm; multiobjective; interdisciplinary consistency

1 (Introduction)
(multidisciplinary design optimization, MDO)
, [1] .
(collaborative optimization, CO)
MDO, , CO
.
, , ,
; ,
, CO
: 20091129; : 20100601.
: 863(2009AA04Z104).

[2] . , CO.
CO, ;
,
[3] .
,
, .
RCOCO, RCO
CO, RCO
[48] ,
.

562

,
, . RCO
, .
RCO3,
[46] , ,
CO,
, ;
, (decision support problem, DSP) (linear physical
programming, LPP)[7] ,
, ,
; (multiobjective evolutionary algorithm, MOEA),
[8] . RCO,
RCO.
RCO
, NSGA--IIRCO.
, ,
, RCO
, RCO
. ,
, .

28

, ,
.
n, CO
.
:

min F (z),
s.t. Ji (z) =

sP
ish
j=1

(zj xij )2 +

sishP
+siaux
j=sish +1

(zj xij )2 = 0.

(1)

: i = 1, 2, , n, z ,
zj j , sish i
, siaux i(),
xij ij .
i:
sishP
+siaux
sP
ish
(xij zj )2 ,
(xij zj )2 +
min Ji (xi ) =
j=sish +1

j=1

s.t. ci (xi ) 6 0.

(2)

: xi i, xij j
, zj j
, ci (xi ).

2 RCO (The formation of RCO


model)

2.2 SUA RCO (RCO model based on


SUA)

,
. MDO
,
.
,
,
. MOD
,
CORCO,
.

DU,
(system uncertainty analysis,
SUA)(concurrent subsystem uncertainty analysis, CSSUA)
[4] ; GU
,
(implicit uncertainty propagation, IUP)[5] ; LI
RCO,
[8] .
MDO,
DUSUA, RCO,
.

2.1 CO
(CO algorithm description)
CO,
, .
, ,
,
,
. ,
.
,
, ,
.
, . CO

min f , f ,
s.t. Ji (z) =

sP
ish
j=1

(zj xij )2 +

sish +2s
P iaux
j=sish +1

f2 =

(zj xij )2 = 0,

ssh f (z)
P
(
)2 z2j +
zj
j=1
sshP
+saux f (z)
2
)2 z2j + f
,
(
zj
j=ssh +1

(3)

f f
, ssh saux
2
, z2j j , f

.
i:
sP
sish +2s
ish
P iaux
min Ji (xi ) =
(xij zj )2 +
(xij zj )2 ,
j=1

j=sish +1

s.t. gi + kgi 6 0,
sP
ish g (x )
i
i 2 2
) xij +
g2i = (
x
ij
j=1
sishP
+siaux g (x )
i
i 2 2
(
) xij +
x
ij
j=sish +1
sish +siaux
P+silocal gi (xi ) 2 2
(
) xij .
xij
j=sish +siaux +1

,
.
,
.
, ; ,

. ,
, (5).

(zi ) = J1 (x1 , zi )+J2 (x2 , zi )+ +Jn (xn , zi ).


(5)

(4)

: x2ij ij , silocal
, k
[4] , gi , k = 1
0.8413, k = 20.9772.
RCO, ,
,
.
,
, ,

563

NSGA--II
RCO
(RCO
solving method based on NSGA--II)

3.1 RCO
(RCO solving scheme)
NSGA--II,
,
.
RCO,
RCO. CO
,
, RCOCO,
RCO, .
,
. NSGA--II
, ,
, ,
RCO.
RCO,
,
. , ,
,
, ,

: zi , x1 xn
, J1 Jn ,
zi , (zi )zi
.

, ,
; , . ,
, ,
,
; ,
, ,
, .
,
(6).
popsize
P
(zi )
i=1
1 = T
,
(6)
popsize
gen + n0
,
(7)
T0 = 1
maxgen
(
T 0 , T 0 > 0,
T =
(8)
0, T 0 < 0,
: popsize, gen,
maxgen, n0
, maxgen0.1, T
.
, ,
. genn
2 (9)(10).
maxgen gen
n = dmax{ popsize
,
maxgen
popsize}e,
(9)

2 = chromosome(n, constrainpos).

(10)

: ( > )
, < 1, < 0.1, ,
,
, ,

564

; chromosome
; constrainpos
.
, 1 2
, = max{1 , 2 }, NSGA--II
[9] . ,
,
zi zj , 3,
zi zj .
1) zi (), zj (
);
2) zi zj , zi zj ;
3) zi zj ,
, zi zj .
,
, ,
,
, ,
.

3.2 (Implementation steps)


Step 1 popsize
P0 , ,
, .
, ,
, P0
:
Step 2 ,
,
, (simulated binary
crossover, SBX),
(polynomial mutation)[10] , c
m 20, Qt ;
Step 3 Rt = Pt Qt ,
,
, , .
(6)(10),
NSGA--II, ;
Step 4 Rt ,
popsizePt , t = t + 1,
Step 2, .

(Engineering example)

4.1 (Mathematical description)


NASAMDO
10.
,
(). 7,

28

min f (x) = 0.7854x1 x22 (3.3333x23 +14.9334x3


43.0934) 1.5079x1 (x26 + x27 ) +
7.477(x36 +x37 )+0.7854(x4 x26 +x5 x27 ),
s.t.

397.5
27
1.0 6 0, g2 =
1.0 6 0,
2
x1 x2 x3
x1 x22 x23
1.93x34
1.93x35
g3 =

1.0
6
0,
g
=
1.0 6 0,
4
x2 x3 x46
x2 x3 x47
r
745x4 2
(
) + 16.9 106
x2 x3
g5 =
1.0 6 0,
110x36
r
745x5 2
) + 157.5 106
(
x2 x3
g6 =
1.0 6 0,
85x37
x2 x3
5x2
g7 =
1.0 6 0, g8 =
1.0 6 0,
40
x1
x1
g9 =
1.0 6 0,
12x2
1.5x6 + 1.9
g10 =
1.0 6 0,
x4
1.1x7 + 1.9
g11 =
1.0 6 0,
x5
2.6 6 x1 6 3.6, 0.7 6 x2 6 0.8,
g1 =

17 6 x3 6 28, 7.3 6 x4 6 8.3,


7.3 6 x5 6 8.3, 2.9 6 x6 6 3.9, 5.0 6 x7 6 5.5,
(11)
, x1 , x2 x3
, x4 x5 , x6 x7
, g1 , g2
, g3 g4 ,
g5 g6 , g7 , g8 g9
, g10 g11 .

4.2

RCO
(RCO model solving)

31
, 1g1 , g2 , g7 g9 , 2
g1 , g2 , g4 , g6 g9 , g11 , 3g1 g3 ,g5 ,
g7 g10 . ,
0.1, k 2,
0, 0.1. popsize80,
maxgen 150, T
n0 10, n0.4,
0.08. NSGA--IIRCO
, ,
, 1
2; ,
0.5, 3.

565

1
Table 1 Results of the fixed threshold
x1

x2

x3

x4

x5

x6

x7

3.5168 0.7724 18.9448 7.5114 7.8540 3.5404 5.1684 3761.3 532.6834 0.0002891
3.1109 0.7304 18.6418 7.9475 8.1065 3.2789 5.0062 3076.9 433.1959 0.8297000 0.0001
3.1108 0.7305 18.6409 7.9475 8.1063 3.2789 5.0061 3077.1 433.1953 0.8303000
3.1212 0.7250 18.6880 7.9511 8.0593 3.2802 5.0094 3065.0 433.4223 0.8041000
3.1117 0.7307 18.6346 7.9474 8.1074 3.2789 5.0062 3077.7 433.1828 0.8287000 1.0000
3.1219 0.7249 18.6875 7.9508 8.0616 3.2802 5.0088 3064.5 433.4310 0.8017000

2
Table 2 Results of the dynamic threshold
x1

x2

x3

x4

x5

x6

x7

3.5954 0.7000 18.7117 7.3000 8.1595 3.3009 5.3716 3401.5 482.4221 0.00025241
3.5954 0.7000 18.7158 7.3000 8.1595 3.3009 5.3717 3402.4 482.2900 0.00024551
3.5953 0.7000 18.7093 7.3000 8.1597 3.3008 5.3716 3401.0 482.6194 0.00024849

3
Table 3 Results of the weighted sum

3, 1, 17, 7.3, 7.5, 3, 5

3425.3
3056.5

488.0234
445.5307

0.00026322
0.24670000

0.0001
0.1000

3.5, 0.7, 17, 7.3, 7.715, 3.35,5.287

3428.6
3056.5

488.8364
445.5307

0.00028808
0.24670000

0.0001
0.1000

2.65, 0.63, 20, 6.8, 6.4, 3.0, 5.099

4381.2
4104.7

727.7270
695.7016

0.00403000
0.25950000

0.0001
0.1000

2.8, 0.71, 25, 7.9, 7.599, 3.0, 5.099

4383.4
4104.7

727.7375
695.7014

0.00401000
0.25950000

0.0001
0.1000

1, NSGA--II
RCO ,
. 0.0001,
Pareto,
, ; 1,
Pareto,
, . ,
, , Pareto
. 2
,
, 1
, .
,
,
,
, .
3,

, RCO[5, 6] ,
. 12,
, ,
, ; 34,
, ,
, .
,
. NSGA--II,
,
,
,
, .

, ,
,

566

.
,
, 0,
0,
.
, 4
, [11]
x = [3.5, 0.7, 17, 7.3, 7.71, 3.35, 5.29],
f (x ) = 2994. 4
, ,

28

, , ;
, ,
. ,
,
, ,
NSGA--II
, ,
, ,
, ,
.

4
Table 4 Results of the deterministic optimization
x1

x2

x3

x4

x5

x6

x7

3.5000
3.4956
3.4291

0.7000
0.7000
0.7000

17.0351
17.9480
17.0000

7.3165
7.3000
7.3000

7.7164
7.7149
7.7607

3.3502
3.3407
3.3364

5.2866
5.2777
5.2344

3000.6
3152.2
2931.1

0.00016996
0.00021274
0.20050000

0.0001
1.0000

5 (Conclusion)
SUARCO
, NSGA--IIRCO.
,

, ;
,
.
:
NSGA--II,
,
; ,
;
,
. ,
RCO
.
, RCO
. ,
, RCO
.
(References):
[1] OLIVIER D W, JEREMY A. State-of-the-art and future trends
in multidisciplinary design optimization[C] //The 48th AIAA/ASME/
ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference. USA: AIAA Press, 2007, 3: 2467 2487.

tion[J]. Structural and Multidisciplinary Optimization, 2008, 35(4): 301


313.
[4] DU X P, CHEN W. Efficient uncertainty analysis methods for multidisciplinary robust design[J]. AIAA Journal, 2002, 40(3): 545 552.
[5] GU X Y, RENAUD J E, PENNINGER C L. Implicit uncertainty propagation for robust collaborative optimization[J]. Journal of Mechanical
Design, 2006, 128(4): 1001 1013.
[6] WAN W M, PENG Y H, HU J, et al. Collaborative robust optimization
under uncertainty based on generalized dynamic constraints network[J].
Structural and Multidisciplinary Optimization, 2009, 38(2): 159 170.
[7] MCALLISTER C D, SIMPSON T W, HACKER K, et al. Integrating
linear physical programming within collaborative optimization for multiobjective multidisciplinary design[J]. Structural and Multidisciplinary
Optimization, 2005, 29(3): 178 189.
[8] LI M, AZARM S. Multiobjective collaborative robust optimization with
interval uncertainty and interdisciplinary uncertainty propagation[J].
Journal of Mechanical Design, 2008, 130(8): 1 11.
[9] , . [J].
, 2009, 26(12): 1452 1454.
(YU Chunyue, WANG Chengen. Multi-objective order-planning model
and algorithm for integrated steel production[J]. Control Theory & Applications, 2009, 26(12): 1452 1454.)
[10] DEB K, GOYAL M. A combined genetic adaptive search (GeneAS) for
engineering design[J]. Computer Science and Informatics, 1996, 26(4):
30 45.
[11] AZAM S, LI W C. Multi-level design optimization using global monotonicity analysis[J]. ASME Journal of Mechanisms and Automation in
Design, 1989, 111(2): 259 263.

(1979), , , ,

, E-mail: lihaiyan0313@163.com;

(1973), , , ,

[2] ALEXANDROV N, LEWIS R M. Analytical computational aspects of


collaborative optimization for multidisciplinary design[J]. AIAA Journal, 2002, 40(2): 301 309.

[3] LI X, LI W J, LIU C A. Geometric analysis of collaborative optimiza-

(1956), , , ,

Das könnte Ihnen auch gefallen