Abstract
Arriving optimal solutions is one of the
important tasks in engineering design. Many real-world
design optimization problems involve multiple conflicting
objectives. The design variables are of continuous or
discrete in nature. In general, for solving Multi Objective
Optimization methods weight method is preferred. In this
method, all the objective functions are converted into a
single objective function by assigning suitable weights to
each objective functions. The main drawback lies in the
selection of proper weights. Recently, evolutionary
algorithms are used to find the nondominated optimal
solutions called as Pareto optimal front in a single run. In
recent years, Non-dominated Sorting Genetic Algorithm II
(NSGA-II) finds increasing applications in solving multi
objective problems comprising of conflicting objectives
because of low computational requirements, elitism and
parameter-less sharing approach. In this work, we propose
a methodology which integrates NSGA-II and Technique
for Order Preference by Similarity to Ideal Solution
(TOPSIS) for solving a two bar truss problem. NSGA-II
searches for the Pareto set where two bar truss is evaluated
in terms of minimizing the weight of the truss and
minimizing the total displacement of the joint under the
given load. Subsequently, TOPSIS selects the best
compromise solution.
Key words Two bar truss, NSGA II, TOPSIS
I. INTRODUCTION
Many real-world engineering design optimization problems
involve many objectives which are often conflicting in
nature. Because of these conflicting objectives, it gives
way for a number of optimal solutions to be generated,
generally called as Pareto-optimal solutions. It is always
preferred to find all possible Pareto-optimal solutions,
since all the Pareto optimal solutions cannot be better than
another without further analysis. Work on the multiobjective engineering design optimization problems for the
design of machine elements involves a weighted
combination of individual objective functions. Generally a
single objective scalar optimisation problem is created
from the weighted approach of the multi-objective
problem. Then either deterministic or stochastic algorithms
are used for optimizing the constraint single objective
scalar optimisation problem. For deterministic approach,
the interior penalty function method was preferred, while
algorithms such as simulated annealing and genetic
algorithms were used for stochastic approaches. The
ISBN 978-93-80609-17-1
627
International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]
F ( x) =
2
P.h. 1 X 2
1
1 .5
g ( x) =
1
g ( x) =
2
1 X4
1
1 X12
2 2.X .X
1 2
P. X 1 1 X 2
1
1
2 2.X .X
1 2
The ranges of parameters are
III. APPLICATION of
(1)
all
all
(2)
2 2 .E . X 2 . X
1 2
Subjected to the following constraints
P. 1 X
(3)
(4)
B. NSGA-II operators
Based on the previous experience, NSGA-II parameters
have been selected. The values of the parameters of NSGAII algorithm used in this study give the best optimal results.
The following are the values of the parameters of NSGA-II
algorithm used in this study:
Variable type = Real variable, Population size = 50,
Crossover probability = 0.8, Real-parameter mutation
probability = 0.01, Real-parameter SBX parameter = 10,
Real-parameter mutation parameter = 100, Total number of
generations = 100.
ISBN 978-93-80609-17-1
628
International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]
S S 11 , S 12 ; S 21 , S 22 ;...; S m1 , S m 2
h
(12)
H i i for i = 1,2,,m
hi hi
where 0 < Hi < 1
Step 7. Choose the best compromise solution whose
relative closeness Hi is the closest to 1.
IV RESULTS and DISCUSSIONS
The bi-objective design optimization of two bar truss
problem is solved by NSGA II algorithm. This algorithm is
implemented using MATLAB 2009 to run on a PC
compatible with Pentium IV, a 3.2 GHz processor and 2
GB of RAM (Random Access Memory). The graphical
method reported only few solutions with the following
spread: (36.12 lb, 0.0959 in.) and (190.368 lb, 0.0718in).
We apply the proposed NSGA II method for the same
problem as per the parameters reported. The nondominated
solutions obtained by NSGA II algorithm is plotted as
Pareto optimal front as shown in Figure 2. The solutions
are spread in the following range: (7.5916 lb, 0.4372 in)
and (47.509 lb, 0.07297 in) which shows the superiority of
NSGA-II compared to the graphical method. Table 1 shows
the comparison of extreme solutions obtained by using
graphical method and the proposed NSGA II approach. The
graphical method could not find wide variety of solutions
in terms of the objectives. If minimization of weight is
important, NSGA-II finds a solution with weight as low as
7.59 lb, whereas the graphical method has found a solution
with minimum weight of 36.1299 lb, an order of magnitude
higher than that found in NSGA-II. NSGA-II solutions are
also better than CGA solutions, both in terms of closeness
to the optimum front and in their spread. The main
advantage of NSGA II is that all the solutions have been
obtained in one iteration. Finally, TOPSIS finds the best
compromise solution with different weights to the objective
functions as seen in fig.2. The results of best compromise
solution obtained by using TOPSIS method for different
weights considered to each objective functions is reported
in Table 2.
Sij
for I = 1,2,,m and j = 1,2
S ij
2
im1 Sij
(6)
S S 11 , S 12 ; S 21 , S 22 ;...; S m1 , S m 2
(7)
S ij W j S ij for I = 1,2,,m and j = 1,2
Step 4. Use the following equations (8) & (9) to determine
S+ and S
and
S
max S11, S 21,..., S m1 , min S12 , S 22 ,..., S m2 (8)
0.45
(10)
hi S ij S j for i = 1,2,,m
j 1
(11)
hi S ij S j for i = 1,2,,m
j 1
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
10
15
20
25
30
35
40
45
W eight of truss, lb
ISBN 978-93-80609-17-1
629
International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]
Method
Graphical
Method
[11]
NSGA II
(present
work)
TABLE I
COMPARISON of EXTREME SOLUTIONS
Solution
X1
X2
F1
Min.Weight
0.65
0.53521
36.1299
Min.
0.9
2.5
190.368
Displacement
Min.Weight
0.7578
0.1069
7.5916
Min.
0.6504
0.7037
47.509
Displacement
REFERENCES
F2
0.0959
0.078237
[1] C.C.A. Coello, Handling Preferences in Evolutionary MultiObjective Optimisation: A Survey. Congress on Evolutionary
Computation, vol. 1, IEEE Service Center, Piscataway, NJ, Julio del 2000,
pp. 3037.
[2] Aimin Zhou, Bo-Yang Qu, Hui Li, Shi-Zheng Zhao, Ponnuthurai
Nagaratnam Suganthan, Qingfu Zhang, Multiobjective evolutionary
algorithms: A survey of the state of the art, Swarm and Evolutionary
Computation 1 (2011) 32490
[3] Srinivas, N. and Deb, K. Multi-Objective function optimization using
non-dominated sorting genetic algorithms, Evolutionary Computation,
2(3):221248 (1995)
[4] K. Deb, A. Pratap, S. Agarwal, T. Meyarivan, A fast and elitist
multiobjective genetic algorithm: NSGA-II, IEEE Transactions on
Evolutionary Computation 6 (2) (2002) 182197.
[5] Hwang, C.L., Yoon, K., Multiple Attribute Decision Making
Methods and Applications. Springer-Verlag Press, Heidelberg. 1981
[6] Ali Wagdy Mohamed & Hegazy Zaher Sabry, Constrained
optimization based on modified differential evolution algorithm
Information Sciences, 194, pp.171208 (2012)
[7] Kaveh, M. Khayatazad, A new meta-heuristic method: Ray
Optimization, Computers and Structures 112-113 283294 (2012)
[8] Deb K, Multi-objective optimization using evolutionary algorithms.
Wiley, New York (2001)
[9] Li M, Azarm S, Aute V A multi-objective genetic algorithm for robust
design optimization. Mech Eng 771778 (2005a)
[10] Antonio Gaspar-Cunha, Jose Ferreira, Gustavo Recio, Evolutionary
robustness analysis for multi-objective optimization: benchmark
problems, Struct Multidisc Optim, 2013
[11] S.S. Rao, Optimization Theory and Applications, third ed., John
Wiley and Sons, New York, 1996.
[12] Olcay Ersel Canyurt , Prabhat Hajela, Cellular genetic algorithm
technique for the Multicriterion design optimization, Struct Multidisc
Optim (2010) 40:201214
0.4372
0.07297
TABLE II
THE BEST COMPROMISE SOLUTION DETERMINED by TOPSIS
Weight
Best Compromise Solution
X1
X2
Weight, lb
Displacement, in
(0.8, 0.2)
17.93545
0.185042
0.763206
0.251898
(0.5, 0.5)
20.08868
0.165287
0.781807
0.279612
(0.2, 0.8)
22.95307
0.144851
0.793640
0.317649
V. CONCLUSION
In this paper, we have addressed the application of NSGAII algorithm to solve the bi-objective design optimization
problem of a two bar truss. The two objective functions
considered in this problem are minimization of weight of
the truss and minimization of vertical displacement of joint
3. The concept of controlled elitism applied in this
algorithm helps to improve the quality of the solutions
obtained by using NSGA-II, which is evident from the
Pareto-optimal front plotted for this problem. The plot of
Pareto-optimal front has uniform distribution and deviation
is very less. The algorithm performs better in our problem
since the number of design variables is minimum and also
the ability of scalar operator of SBX crossover to create
new generations. Unlike the weight method of combining
the objectives into a single objective function, NSGA-II
uses the concept of domination count and crowding
distance operator which helps to improve the results and
hence the performance is better than that found by
traditional MOEAs. A multi criteria decision making tool
namely TOPSIS is applied to select the best compromise
solution from the obtained Pareto optimal front. It is
evident from the plot of results that even after assigning
different weights to each objective, the best compromise
solutions obtained by TOPSIS are within the close
proximity. In future, the combined approach of NSGA II
algorithm and TOPSIS can be used to solve real world
multi objective design optimization of mechanical
components such as flywheel, gears, spring and shaft.
ACKNOWLEDGMENT
The authors would like to thank the Management of K.L.N
College of Engineering for providing the facilities to carry
out the research work. The team has great interest to honor
Dr.S.Miruna Joe Amali, Associate Professor, Computer
Science Engineering Department of K.L.N College of
Engineering for her help in the coding of NSGA II
Algorithm using Matlab.
ISBN 978-93-80609-17-1
630