Information Technology Reference
In-Depth Information
·
1
1
·
1
32.768
32.768, 1,…,.
(
(22)
The parameters of this f
global minimum
0
moderate complications in
tima, the algorithm only ba
tion. However, the ability o
that will be able to cross the
function are recommended to set a=20, b=0.2, c=2
π
. T
0
is located on
0
for
i=1,…,n
. This function produ
the search of optimal solutions. Due to multiple local
ased on the gradient will be trapped in local optimal so
of search and analysis in wider area is helpful to algorith
e valley among the optima and to achieve better results.
The
uces
op-
olu-
hms
Fi
ig. 3.
Ackley function in 2-dimension
The initial parameters a
population size, frequency
vious position, current posit
each bat are randomly gene
(d=64, 128) to produce the
dimension increment leads
to discover the global optim
est dimension is set accordi
The initial positions of b
pattern is executed by 1000
this pattern is repeated exa
age error (
Avg.
), standard
These statistics appraise the
of convergent solution foun
the reliability of average er
re the same as the original bat algorithm in [6] includ
range, emission pulse rate and loudness. Then, the p
tion, location, velocity, emission pulse rate and loudnes
erated in initialization. There are two kinds of dimensi
e different complexity for two benchmark functions. T
to the expanded solution space which requires more ti
mal solution in these functions. In this simulation, the hi
ing to the previous work [6].
bats are randomly generated and regarded as a pattern.
0 iteration as a round to obtain a convergent solution. A
amined by 100 rounds to derive the statistics include av
deviation (
Std.
) maximum (
Max.
) and minimum (
Mi
e solving efficiency of GBA. The
Avg.
presents the qua
nd by GBA. The
Std.
,
Max.
and
Min.
are applied to supp
rror. The
Std.
is variance of convergent solutions found
ding
pre-
s of
ions
The
ime
igh-
A
And,
ver-
in.
).
ality
port
d in
Search WWH ::
Custom Search