Professional Documents
Culture Documents
CHAPTER 3
3.1 INTRODUCTION
() , = 1,2, . (3.1)
where F(x) is the objective function, x is the set of design variables ,Xi is the
range set of the possible values for each design variable. The following HS
algorithm parameters are also specified. The harmony memory size(HMS) or
number of solution vectors in the harmony memory; the harmony memory
considering rate (HMCR); the pitch adjusting rate (PAR); the number of
decision variables (N); the number of improvisations (NI) and the stopping
criterion.
37
11 21 1
.. 1 1 ( (1) )
12 22 2
.... 1 2 ( (2) )
= .. ..
.... .. .
. .. (3.2)
11 21 ..1
1 1
( (1) )
[ 1 2 1
] ( () )
( ( ) < )
{1 , 2 , . . }
(3.3)
( ( ) < )
= ( )
else
end (3.4)
( )
() = + (3.5)
() = exp(. ) (3.6)
( )
= (3.7)
Sphere function (1 )
(3.8)
() = 2 ()
=1
(3.9)
() = |()| + |()|
=1 =1
Rosenbrock function (3 )
1
2 (3.10)
() = (100( ( + 1) 2 ()) + ( () 1)2 )
=1
Step function (4 )
(3.11)
() = (| () + 0.5|)2
=1
(3.12)
() = ( ())2
=1 =1
Rastrigin function (7 )
(3.14)
() = ( 2 () 10 cos(2 ()) + 10)
=1
42
1 1
() = 20 exp (0.2 2 ()) ( cos(2())) + 20 +
=1 =1
(3.15)
Griewank function (9 )
1 ()
() = 2 () cos ( )+1 (3.16)
4000
=1 =1
(3.17)
() = 2 () + _1 (3.18)
=1
() = ( () ) + _2 (3.19)
=1 =1
1
2
() = (100(( + 1) 2 ()) + (() 1)2 ) + _6 (3.20)
=1
Shifted Sphere function and Shifted Schwefels problem 1.2 are unimodal.
Step function is discontinuous. Rosenbrock function, Schwefels problem
2.26, Rastrigin function, Ackley function, Griewank function, Shifted
Rosenbrock function are difficult multimodal problems where the number of
local optima increases with the problem dimension. Six-hump Camel-back
function is a low-dimensional function with only a few local optima.
Parameters Values
Parameters Values
Population size 50
Max iteration 100
wmin 0.4
wmax 0.9
C1=C2
1.4
Velocity bounds
(-3,7)
where UB and LB are the upper and lower bounds of the design variable.
46
Table 3.4 Mean and standard deviation of the benchmark function optimization results (n=30)
4
x 10
5
HSA
4.5 IHSA
GA
4 PSO
3.5
Objective function
2.5
1.5
0.5
0 100 200 300 400 500 600 700 800 900 1000
No of Iterations
small average iteration number. That is to say, IHSA has stronger ability and
faster convergence speed to find better solutions than the compared HSA.
Hence, it can be concluded that IHSA is much effective and reliable for
solving global optimization problems.
In this section, the effects of the HMS, HMCR and PAR on the
performance of the IHSA are examined. The averages and SD obtained by
using different HMS, HMCR and PAR values for the number of dimensions n
= 30 are reported in Tables 3.6 to Table 3.8 respectively. It can be found
from Tables 3.6 that for most benchmark functions, a small value of HMS
(i.e., 5 or 10) is superior to a large value (i.e., 20 or 50). This can be explained
by the basic principle of the original HSA. Since HM is emulating musicians
short-term memory and the short-term memory of the human is known to be
small, it is logical to use a small HMS. From Table 3.7, it can be seen that the
performance of the IHSA degrades with the HMCR value decreasing for most
of the test functions. Although a small HMCR value can increase the diversity
of the harmony memory, a much slower convergence rate is obtained
accordingly. Hence, it is a good choice to use a large value for the HMCR
(i.e. HMCR > = 0.95). The effect of using different constant values of PAR
on the performance of IHSA Table 3.8 is investigated. According to
Table 3.8, there is no one setting of PAR is better than the other settings. It is
generally better to use a PAR value between 0.3 and 0.7.
49
Table 3.5 Mean and standard deviation of the benchmark function optimization results (n=50)
390 2472150
1344100.0 2368300 1343900.0
2368161.2 1343572.9
4300.483 4872.476
27 72
50
Table 3.6 Mean and standard deviation of the benchmark function optimization results (n=100)
Function Global GA PSO HSA IHSA
optimum
0 20003.000 2219.600 19181.000 1987.400 19166.359 1969.024 19239.760 1954.350
52
Table 3.9 The effect of HMCR (n=30)
53
Table 3.10 The effect of PAR (n=30)
Function PAR =0.01 PAR =0.1 PAR =0.3 PAR =0.5 PAR =0.7 PAR =0.9
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
(0.000000) (0.000000) (0.000000) (0.000000) (0.000000) (0.000000)
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
(0.000000) (0.000000) (0.000000) (0.000000) (0.000000) (0.000000)
282.653364 107.440357 32.833197 30.660391 25.328236 29.568903
(489.187424) (106.216073) (15.307607) (13.342002) (0.313470) (0.233126)
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
(0.000000) (0.000000) (0.000000) (0.000000) (0.000000) (0.000000)
986.842377 0.096724 0.000000 0.000000 0.000000 0.000000
(407.467545) (0.053309) (0.000000) (0.000000) (0.000000) (0.000000)
0.000747 0.000974 0.026382 0.027708 0.009598 0.0023337
(0.000123) (0.000219) (0.006004) (0.006397) (0.004128) (0.001134)
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
(0.000000) (0.000000) (0.000000) (0.000000) (0.000000) (0.000000)
54
Table 3.10 (Continued)
Function PAR =0.01 PAR =0.1 PAR =0.3 PAR =0.5 PAR =0.7 PAR =0.9
0.000000 0.000000 0.000000 0.135848 0.073089 0.000000
(0.000000) (0.000000) (0.000000) (0.493023) (0.415800) (0.000000)
0.035402 0.041795 0.010084 0.003705 0.001881 0.000000
(0.036172) (0.035617) (0.017694) (0.007580) (0.006472) (0.000000)
-1.031628 -1.031628 -1.031628 -1.031628 -1.031728 -1.031628
(0.000000) (0.000000) (0.000000) (0.000000) (0.000000) (0.000000)
-459.999882 -449.899808 -449.782512 -449.922743 -449.08060 -448.999535
(0.000055) (0.000066) (0.001794) (0.002824) (0.000669) (0.000298)
2974.495515 -443.731060 -449.435561 -448.663221 -447.142620 178.454056
(1454.665369) (19.753425) (0.716192) (2.989611) (5.355981) (369.538765)
55
56
3.4 CONCLUSION