Progressively Interactive Evolutionary Multi-Objective Optimization ...
Progressively Interactive Evolutionary Multi-Objective Optimization ...
Progressively Interactive Evolutionary Multi-Objective Optimization ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
are K + L + 1real-valuedvariablesin this problemaswell:<br />
Minimize F(x, y) =<br />
<br />
(1 − x1)(1 + K<br />
j=2 x2 j )y1<br />
x1(1 + K<br />
j=2 x2 j )y1<br />
<br />
subject to (x) ∈ argmin (x) f(x) =<br />
<br />
(1 − x1)(1 + K+L<br />
j=K+1 x2 j)y1<br />
x1(1 + K+L<br />
j=K+1 x2 j)y1<br />
,<br />
,<br />
G1(x) = (1 − x1)y1 + 1<br />
(12)<br />
1<br />
x1y1 − 2 + [5(1 − x1)y1 + 0.2] ≥ 0, [·] denotesgreatestint. function,<br />
2 5<br />
−1 ≤ x1 ≤ 1, 1 ≤ y1 ≤ 2,<br />
−(K + L) ≤ xi ≤ (K + L), i = 2, . . . , (K + L).<br />
For the upper level Pareto-optimal front, xi = 0 for i = 2, . . . , (K + L), x1 ∈ [2(1 −<br />
1/y1), 2(1 − 0.9/y1)], y1 ∈ {1, 1.2, 1.4, 1.6, 1.8}(Figure8). For this test problemwe have<br />
chosen K = 5 and L = 4 (an overall 10-variable problem). This problem has similar<br />
difficulties as in DS4, except that only a finite number of y1 qualifies at the upper level<br />
Pareto-optimal front and that a consecutive set of lower level Pareto-optimal solutions<br />
now qualify tobeon theupper levelPareto-optimalfront.<br />
5 HybridBilevel <strong>Evolutionary</strong><strong>Multi</strong>-<strong>Objective</strong> <strong>Optimization</strong> (H-BLEMO)<br />
Algorithm<br />
The proposed hybrid BLEMO procedure is motivated from our previously suggested<br />
algorithms (Deb and Sinha, 2009a,b),but differs in many different fundamental ways.<br />
Beforewe describethe differences,we first outline the proposed hybridprocedure.<br />
A sketch of the population structure is shown in Figure 9. The initial population<br />
Lower level<br />
NSGA−II Local search<br />
x_u<br />
t=0<br />
x_l<br />
t=1<br />
x_l<br />
t=t*<br />
ND<br />
x_l<br />
Archive<br />
T=0<br />
ND<br />
ND<br />
Upper level NSGA−II<br />
Archive<br />
Figure9: A sketchof theproposed bileveloptimization algorithm.<br />
(markedwith upperlevel generationcounter T = 0 ofsize Nu) has asubpopulation of<br />
lower level variable set xl for each upper level variable set xu. Initially the subpopulation<br />
size (N (0)<br />
l ) is kept identical for each xu variable set, but it is allowed to change<br />
adaptivelywithgeneration T. Initially,anemptyarchive A0 iscreated. Foreach xu,we<br />
perform a lower level NSGA-II operation on the corresponding subpopulation having<br />
variables xl alone, not till the true lower level Pareto-optimal front is found, but only<br />
till a small number of generations at which the specified lower level termination criterion<br />
(discussed in subsection 5.2) is satisfied. Thereafter, a local search is performed<br />
90<br />
T=1