Anda di halaman 1dari 11

Journal of Microwaves and Optoelectronics, Vol. 6, No.

1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

167
Abstract The aim of this work is to compare and investigate
mathematical methodologies to deal with multiobjective
optimization problems. Four methods are considered: the Weight
Sum Approach, the Nondominated Sorting Genetic Algorithm
(NSGA), the Strength Pareto Evolutionary Algorithm (SPEA) and
the Pareto Archive Evolutionary Strategy (PAES). The work
proposes some indexes in order to compare the performance of the
algorithms. An analytical benchmark problem was investigated to
validate the methods and their implementations. For real-world
problems, the electromagnetic field computation was performed by
the finite element method and Kriging models were used as
surrogate functions to approximate the objective functions of the
problems. Three real-world problems were analyzed: the first one
consists in the design of a die press model, the second problem deals
with the design of superconducting magnetic energy storage
(SMES) and the third one dealt with the design of a special
reluctance motor.

Index Terms Optimization Methods, Finite Element Method
I. INTRODUCTION
Multiobjective optimization methods deal with the conflict of the requirements of an
electromagnetic design, instead of mono-objective optimization method, that deals only with a
maximization (or minimization) of one design requirement. In this paper, we will compare some
multiobjective optimization methods and propose indexes to measure the performance of a
multiobjective optimization method on the Pareto set achievement. These indexes must reflect two
important attributes: the Pareto set must converge to an optimal-Pareto set and the diversity of the
Pareto set must be as high as possible.
II. THE MULTIOBJECTIVE OPTIMIZATION METHODS
Four multiobjective optimization methods will be analyzed: the Weighted Sum Method, the Non-
dominated Sorting Genetic Algorithm (NSGA) [1], the Strength Pareto Evolutionary Algorithm
(SPEA) [2] and Pareto Archived Evolutionary Strategy (PAES) [3]. In this section, we will briefly
describe these methods.
A. Weighted Sum Method
The Weighted Sum Method is, by far, the most intuitive approach to solve a multiobjective
optimization. Instead of deal with the conflict of the requirements, a weighted sum of the functions is
performed and a standard mono-optimization is performed. For each set of weights one point of the
Multiobjective Optimization Applied to
Electromagnetic Devices
Fbio Augusto Baria Perdiz and Luiz Lebensztajn
Escola Politcnica da Universidade de So Paulo
perdiz@gmail.com, leb@pea.usp.br
Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

168
Pareto set is achieved.
B. Nondominated Sorting Genetic Algorithm
The selection operator on NSGA [1] differs from the usual genetic algorithm. It ranks the
population and identifies the nondominated individuals. They are assumed to be the first
nondominated front and a large dummy fitness value is assigned to them. To preserve diversity in the
population, a fitness sharing process is applied to these individuals.
The process divides the fitness by a quantity proportional to the number of individuals around it.
This process punishes the individuals located in high populated areas.
The fitness value degradation of neighbor individuals may be executed using (1) e (2), where the
parameter d
ij
is the variable distance between two individuals i e j, and
shared
is the maximum distance
allowed between any two individuals to become members of the same niche. The term df
i
is the
dummy fitness value assigned to individual i. The df
'i
is its corresponding shared value. N
pop
is the
size of the population.
( )

<

=
shared ij
shared ij
shared
ij
ij
d if
d if
d
d Sh

, 0
, 1
2

(1)
and
1
1
) (

=

=

pop
N
j
ij i i
d Sh df df

(2)
The dominated individuals are now assumed to constitute a new population who will suffer the
same process. The new nondominated front is assigned a dummy fitness value, which is kept smaller
than the minimum shared dummy fitness of the previous front. The process continues until the entire
population is classified into fronts.
The individuals assigned share values are used as parameter for the stochastic remainder
proportional selection for the crossover and mutation operations. During the selection, the best ones
(higher share values) get more copies. This results in quick convergence of the population towards
nondominated regions and sharing helps to distribute it over this region.
C. Strength Pareto Evolutionary Algorithm - SPEA
The SPEA [2] stores the nondominated solutions found so far externally. A fitness assignment
method is used, where one population serves as basis for anothers evaluation using the concept of
Pareto dominance. Clustering is performed to reduce the Pareto-optimal front to the allowed maximal
size, preserving its original characteristics. A Pareto-based niching method is provided in order to
preserve diversity in the population.
SPEA starts updating the Pareto set. All nondominated solutions in the population are copied to the
Pareto set, and possible dominated ones are removed from it. This set is called the extended Pareto
set, and if its size exceeds a given maximum, a reduction clustering base method is applied. The
Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

169
population is used to calculate the strength (fitness). In a first step each solution in the Pareto set is
assigned a value s [0,1) called strength. S is proportional to the number of population members
covered. In the second step, for each individual in the population, the strengths of all external Pareto
solutions by which it is covered are summed up. Adding one to this result, the fitness value is
obtained. This guarantees that Pareto solutions are most likely to be reproduced.
Finally, both population and external Pareto set are considered for reproduction. A binary
tournament selection is applied: the smaller the fitness, the higher the probability to be selected.
Crossover and mutation are then used to generate the new population.
D. Pareto Archived Evolution Strategy - PAES
The algorithm proposed by Knowles and Corne [3] begins with the initialization of a single
chromosome (the current solution), which is evaluated using the objective functions. A mutation
operator is then applied to current solution to obtain a copy. This new individual is evaluated and is
called the new candidate solution.
The acceptance criterion is straightforward, when one solution dominates the other, but when there
is no dominance between the solutions, the candidate is compared to a reference population, called the
archive. On the acceptance criterion, the algorithm takes into account the location of the solution:
there is a preference to solutions located on the least crowded regions.
So the archive stores all the nondominated generated solutions and helps the accurate selection
between the candidate and the current solution. The result of this process is a front closer to the
optimal solution and more diverse. This diversity is achieved thanks to a d-dimensional grid, where d
is the number of objectives.
III. THE INDEXES AND THE PROBLEMS
The first index was created to show the convergence of the Pareto-set to an optimal-Pareto set. It
measures the non-dominated area under the Pareto surface. The second index is linked with the Pareto
front diversity. The individuals that belong to the Pareto set are ordered according to one function.
Then the distance between neighbors is performed and the variance of the distances is calculated. If
the distribution on the Pareto set is uniform then a low variance is achieved. The third one is
associated to the coverage concept and shows that the outcomes of one algorithm dominate the
outcomes of another algorithm, although it does not tell how much better it is.
A. The Analytical Function
The analytical problem was proposed by Deb [4] and deals with the minimization of two functions,
1 1 1
x ) x ( f = (3)
( ) ( ) g / f 1 ) 1 m /( x 9 1 ) x ,..., x , x ( f
1
m
2 i
i m 2 1 2
+ =
=

(4)
where m=30 e x
i
[0,1].

Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

170
B. TEAM Workshop Problem 25
The geometry of TEAM Workshop Problem 25 is shown on Fig.1. The goal is to obtain a
homogeneous radial magnetic induction distribution on the path e-f. Four parameters (R1, L2, L3 and
L4) could be changed in a specified range without any constraint. The complete description of the
problem is presented in [5], so it is omitted here.

Fig. 1. Team Workshop problem 25: die press model
The homogeneity of the magnetic distribution along the path, if the index p shows calculated points
and the index o prescribed points, could be written as:
( ) ( ) { } + =
=
10
1 i
2
yxio yip
2
xio xip
B B B B W

(5)
It is possible to define the local quality of the induction magnetic distribution, as proposed on [6].
The error on the amplitude is defined as 100 / ) ( max
max
=
p o p B
B B B E and the error on the angle of
magnetic induction vector is computed by
0 max
max
B Bp
E

= . The multiobjective problem is then


to minimize W,
max
E and
max B
E . Every function was approximated with a Kriging model, based on a
regular 7x7x7 grid for the optimization variables (R1, L2, L4).
C. TEAM Workshop Problem 22
A Superconducting Magnetic Energy Storage (SMES) configuration (Fig.2) shall be optimized.
There are three design parameters (R2, h2, d2). The complete description of the problem is presented
in [7], so it is omitted here. There are two objectives: minimize the stray field (measured at a distance
of 10 m from the device) and the stored energy in the device should be 180 MJ. The maximum
magnetic induction (B
max
) in the coils must not violate the quench condition, so the magnetic
induction on the coils could not be greater than 4.92 T, because on both coils the absolute value of the
density current was set to 22.5MA/m2. The quench condition is, therefore, a constraint of the
Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

171
optimization problem.

Fig. 2. Team Workshop problem 22, SMES
This problem is, by far, the most popular benchmark on multiobjective optimization of
electromagnetic devices. Several methodologies were adopted to solve it. For example, on [8] the
problem is solved as an unconstrained multiobjective optimization problem, using the NSGA. A
constrained solution could be found on [9], when the constraint is treated as an objective, using
Niched Pareto Genetic Algorithm (NPGA). Another constrained solution is proposed on [10], when a
Multiobjective Clonal Selection Algorithm is adopted.
In this works a penalty approach will be used to insert the constraint on the objectives. Thus, the
problem could be written as a unconstraint minimization problem and the quench condition will
appear as a penalty function, as follows:
( ) 92 . 4 , 0 max 100 1
10 . 180
1
max 6
+ = B
Energy
F

(6)
( ) 92 . 4 , 0 max 100 2
max
+ = B B F
stray
(7)
Every function and constraint was approximated with a Kriging model, based on a regular
7x7x7 grid for the optimization variables (R2, h2, d2).
D. The Switched Reluctance Motor
.The 4:2 pole 2 phase Switched Reluctance Motor (SRM) has a configuration , which is
particularly well suited for unidirectional low cost drives due to its simple motor construction and
simple drive topology with only four power switches. In order to have a better torque characteristic,
the SRM has an asymmetric designed rotor poles with an original geometry that ensures starting
torque in a defined direction at any rotor position. For design proposes, the problem is the
Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

172
minimization the torque ripple, while the average torque is maximized. This is a typical multi-
objective problem, due to the contradiction between those requirements.
Fig. 3 shows the motor geometry and the optimization parameters are
0
, l
g1
e l
g2
.


Fig. 3 Motor Geometry and rotor details
To compute the torque ripple and the average torque, the finite element method was used and
to save computation time, a Kriging approximation was used. A regular 5x5x5 grid was performed,
e.g, 125 triples (
0
, l
g1
, l
g2
) were generated and rotor was placed in 19 different positions, and this
results in 125 x 19 Finite Element Analysis.
IV. RESULTS
For every problem, we will adopt the following parameters: 250 generations, population equal to
100, crossover probability equals to 0.8 and mutation probability equals to 0.01.
A. The Analytical Problem
Fig. 4 shows the results for the analytical problem. We can see the Pareto set for the four methods
and the optimal front. This figure shows that only the evolutionary methods achieved a good Pareto
set. The nondominated area under the Pareto surface for the NSGA, SPEA and PAES as a function of
the number of iterations shows the convergence of the methods. Another way to measure the quality
of the Pareto set is the variance of the distance, as shown in Table I. The most effective method
(PAES) has the lower variance.
Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

173
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1
1.2
1.4
F1
F
2
OTIMO
WEIGHT
NSGA
SPEA
PAES

(a) The Pareto Set
0 50 100 150 200 250
0
0.5
1
1.5
2
2.5
3
Iteration
N
o
n
d
o
m
i
n
a
t
e
d

A
r
e
a
NSGA
SPEA
PAES

(b) The nondominated area under the Pareto surface
Fig. 4 The Analytical function
TABLE I THE VARIANCES
Method Weighted NSGA SPEA PAES
Variance 5.29x10
-4
2.6x10
-4
4.3x10
-5
1.7x10
-5


B. TEAM Workshop Problem 25
The TEAM workshop problem 25 was analyzed as a multi-objective problem. Every functions were
computed by the Finite Element Method and surrogate functions were used to save computation time
and Kriging models were adopted, as used on [6], when the authors adopted a min-max approach to
solve the multi-objective problem.
The Pareto-set for SPEA, NSGA and PAES are shown on Fig.5 and seems to be similar, but Table
II shows that SPEA achieves the best value for variance on the Pareto set.
Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

174
0
2
4
6
0
10
20
30
40
-0.02
0
0.02
0.04
0.06
0.08
0.1
Local Error ->Angle
Local Error ->Amplitude(%)
G
l
o
b
a
l

E
r
r
o
r

Fig. 5 The Pareto Set - TEAM Workshop Problem 25
TABLE II TEAM 25 VARIANCES
METHOD NSGA SPEA PAES
VARIANCE 1.2X10
-3
0.86X10
-3
9.1X10
-3

C. TEAM Workshop Problem 22
Fig.6 shows the three different Pareto sets. NSGA and PAES have a better performance then SPEA
in this case, as the coverage data could demonstrate on Table III, which also shows that the PAES
front is better than NSGA one and Table IV shows the quality of the Nondominated Front. PAES
front is the most equally spaced one.

0 0.01 0.02 0.03 0.04 0.05
0.5
1
1.5
2
2.5
x 10
-3
F1
F
2
NSGA
SPEA
PAES

Fig. 6 The Pareto Set for the TEAM Workshop Problem 22
The nondominated area under the Pareto surface TEAM-22 is shown on Fig 7, and for NSGA and
PAES we can observe that there is a very little improvement on the area in the last iterations. This
Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

175
improvement is followed by an augment on the diversity of the Pareto set.
TABLE III TEAM 22: COVERAGE
Method NSGA SPEA PAES
NSGA - 39,08 28,57
SPEA 9,84 - 6,59
PAES 22,95 66,67 -

TABLE IV TEAM 22 VARIANCES
Method NSGA SPEA PAES
Variance 4.0 x10
-3
5.6 x10
-3
3.8 x10
-3

0 50 100 150 200 250
0
0.5
1
1.5
2
2.5
3
x 10
-4
Iteration
N
o
n
d
o
m
i
n
a
t
e
d

A
r
e
a
NSGA
SPEA
PAES

Fig. 7 The nondominated area under the Pareto surface TEAM-22
D. The Switched Reluctance Motor
Fig.8 shows the three different Pareto sets for the Switched Reluctance Motor. A simple visual
analysis shows that all the method attain the Pareto front. Nevertheless, when coverage and the
variance of the Pareto set is analyzed, PAES outperforms NSGA and SPEA.
TABLE V TEAM 22: COVERAGE
Method NSGA SPEA PAES
NSGA - 0 0,52
SPEA 30,77 - 1,30
PAES 71,79 53,33 -

TABLE VI SRM VARIANCES
Method NSGA SPEA PAES
Variance 16 x10
-4
3.53 x10
-3
0.44 x10
-3



Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

176
0.265 0.27 0.275 0.28 0.285 0.29 0.295
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
3
Torque(N.m)
R
i
p
p
l
e
(
%
)
NSGA
SPEA
PAES

Fig 8 The Pareto Set The Switched Reluctance Motor
V. COMPUTATIONAL ASPECTS
The computational cost is a very important point to analyze when the finite element method is
associated to an optimization process. Nevertheless, when surrogate functions are used to replace the
objective functions a lot of time is saved. An important question is the precision and in our case, we
have checked the data from the Pareto set: the maximum deviation between the Kriging model and the
Finite Element Analysis was only 1.4%. As an example, for the Problem 22 solution, only 343 Finite
Element Method simulations were computed to create the Kriging model. If no surrogate functions
were used, 25000 finite element analyses will be necessary and the computational cost will be
prohibitive.
VI. CONCLUSIONS
Multi-objective methods were performed and the tests showed that evolutionary algorithms have a
very good performance on real world problems.
PAES has a better performance for the analytical function proposed by Deb. This better behavior
could be observed by the area index or by the variance index. On real-world problems, PAES remains
with the best Pareto set, but the quality between the Paretos front obtained by NSGA, SPEA and
PAES is not so different. On Problem 25 the best variance index was obtained by SPEA and on
Problem 22 the coverage index shows that the results from PAES and NSGA are similar. For the
Switched Reluctance Motor, the best performance is still associated to PAES.
The use of surrogate functions was the key for the efficiency of every multi-objective optimization
method. The computation effort was highly decreased with Kriging models, particularly for the
Switched Reluctance Motor, when the multi-objective optimization was performed in a scenario
Journal of Microwaves and Optoelectronics, Vol. 6, No. 1, June 2007
Brazilian Microwave and Optoelectronics Society-SBMO Received 10 Aug., 2006; revised 12 Feb., 2007; accepted 23 March, 2007
ISSN 1516-7399 2007 SBMO

177
where each solution evaluation is temporally expensive.
REFERENCES
[1] N. Srinivas and K. Deb, Multiobjective Optimization Using Nondominated Sorting in Genetic Algorithms,
Evolutionary Computation,(2), 3,221-248,1994
[2] E. Zitzler and L. Thiele. An Evolutionary Algorithm for Multiobjective Optimization: The Strength Pareto Approach,
Technical Report 43, ETH Zurich, May 1998.
[3] J. D Knowles, D. W. Corne, The Pareto Archived Evolution Strategy: A new baseline algorithm for Pareto
multiobjective optimization. Congress on Evolutionary Computation, v.1, p.98-105. IEEE Press, Piscataway, NJ,
1999.
[4] K. Deb, Multi-objective genetic algorithms: Problem difficulties and construction of test functions. Technical Report CI
49/98, Kanpur, India
[5] http://ics.ec-lyon.fr/team.html [On-line]
[6] L. Lebensztajn, J.-L. Coulomb TEAM Workshop Problem 25: A Multiobjective Analysis IEEE Trans on MAG 40, 2,
1402-1405.
[7] http://www.igte.tugraz.at/archive/team/index.htm [On-line]
[8] A. H. F. Dias and J. A. Vasconcelos, Multiobjective genetic algorithms applied to solve optimization problems, IEEE
Trans. Magn., vol. 38, no. 2, pp. 11331136, Mar. 2002.
[9] D. A. G. Vieira, R. L. S. Adriano, J. A. Vasconcelos, and L. Krhenbhl, Treating constraints as objectives in
multiobjective optimization problems using niched pareto genetic algorithm, IEEE Trans. Magn., vol. 40, no. 2, pp.
11881191, Mar. 2004.
[10] F. G. Guimares, F. Campelo, R. R. Saldanha, H. Igarashi, R. H. C. Takahashi and Jaime A. Ramrez A Multiobjective
Proposal for the TEAM Benchmark Problem 22 IEEE Trans. Magn., vol. 42, no. 4, pp. 14711474, Apr. 2006.

Anda mungkin juga menyukai