Está en la página 1de 11

G MANIKANDAN et al.

/ (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES


Vol No. 3, Issue No. 2, 078 - 088

Aerodynamic Multi-Objective Optimization


Using Parallel Genetic Algorithm

G MANIKANDAN M ANANDA RAO1


Professor Professor and Principal
SS Institute of Technology SS Institute of Technology
Dundigal, Hyderabad Dundigal, Hyderabad
Andhra Pradesh, India Andhra Pradesh, India
manii731@yahoo.co.in profanandarao @yahoo.com

T
ES
A
IJ

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 78


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

Abstract - Shape optimization of airfoil R2= User specified parameter which controls
for the aerodynamic analysis of a low the probability test of global random number
speed and low Reynolds number mutation operator
unmanned aerial vehicle wing is R (0, 1) = Random number generator which
performed using parallel Genetic returns a random value between 0 and 1.
Algorithm. NACA 2412 chambered airfoil ith gene from the jth chromosome from
is chosen as zero generation airfoil. Real the nth GA generation.
number coding is implemented for jth chromosome from nth GA
inputting seed value. Four modification generation
operators are applied in this design space
User specified maximum limits on
search method. The design space genes th
are control points of airfoil. Multiple the i gene
fitness functions are utilized. Genetic User specified minimum limits on

T
th
Algorithm optimized airfoil profiles are the i gene
used for the fabrication of composite ϵ = User specified parameter which controls
material wing and are tested in the the size of perturbation mutation parameters
subsonic wind tunnel. The aerodynamic Subscripts
characteristics gleaned from experimental i = Gene Index
analysis are compared with base line

airfoil.
ES
airfoil and genetic algorithm optimized

Keywords: Parallel Genetic Algorithm;


Cambered Aerofoil; Fitness Function;
j = Chromosome Index
k = Objective function index
m = No of scalar objective function
Superscripts
n = Population Index
t = Temporary chromosome and gene values
Composite Material; Wind Tunnel; obtained after initial selection and before
Aerodynamic characteristics. modification operator.
A
Nomenclature I Introduction

A = Set of Scalar Chromosome The objective of airfoil design


L = Set of Vector Lift values optimization is to enhance the lift and L/D
L/D = Lift by Drag ratio ratio and minimize the drag. There is a
F – Set of Scalar Objective Function tradeoff between drag and lift because one
IJ

f = Scalar Objective Function of the drag components called Induced drag


No = nth GA generation increases in proportion to the square of lift.
M= User specified vector with four Therefore the design airfoil profile is a
elements that controls modification challenging problem. Very precise shape
operators optimization using very sensitive control
mpt = Pass through operator points is needed. Aerodynamic evaluation
mc = Random average cross over operator using high fidelity model using Navier
mpm= Perturbation mutation operator Stroke equation leads to very expensive
mm = Random mutation operator function evolution. Gradient based
R1= User specified parameter which controls numerical method for optimizing the airfoil
the probability test of perturbation mutation shape was in practice for many years. The
operator efficiency of gradient based optimization
generally requires a smooth design space

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 79


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

and single extreme or initial guess very design optimization [7]. Direct and inverse
close to global extreme for quick and proper airfoil design is carried out using multi
convergence. The number of function objective genetic algorithm [8]. Multi
evaluation required for the convergence of objective optimization based on Pareto front
Genetic Algorithm (GA) optimization technique and neural network by reduced
process exceeds the finite difference based cost was developed [9]. Shape optimization
gradient optimization. Genetic algorithm has by the use of Voxel (N dimensional pixel)
capability of finding a global optimum from based presentation using series of binary
multiple design variables effectively because number was proposed by Peter Baron,
it does not use any derivative information. Robert Fischer and R Smith [10]. To solve
Therefore in this paper a promising GA problems with large number of real design
approach is used for airfoil shape parameters, Stochastic Genetic Algorithm
optimization. are used effectively and efficiently [11]. For

T
Air Combat Tactics optimization, Stochastic
Profound knowledge and quite GA has been successfully applied [12]. The
essential idea of optimization by genetic dynamic coding for binary coded GAs to
algorithm is delivered by DE Goldberg [1]. treat continuous design space is a novel
Parallel computing for genetic algorithm technique adopted by Adaptive Range GA
optimization is used for the fitness

effort required for


ES
evaluation of computational fluid dynamics
analysis because of large computational
aerodynamic
optimization. In aerospace most of the
airfoil optimization have however employed
(ARGA) [13]. The aerodynamic airfoil
shape optimization can be performed better
than real coded GA by ARGA [14]. Airfoil
shape optimization was carried out by multi
objective optimization technique [15].
Missile aerodynamic shape optimization and
sequential genetic algorithm than the Wing shape optimization was also carried
parallel. [2, 3, 4]. In this paper parallel out by multi objective optimization [16],
computing method and real number coding [17]
is implemented. The chromosomes are
A
coded as finite length string of real numbers II Parallel Genetic Algorithm for Airfoil
corresponding to the design variables. The Shape Optimization
real coded genetic algorithm outperformed
binary coded genetic algorithm in many For a single objective optimization
design problems [5, 6]. Hybrid genetic problem involving lesser number of design
IJ

algorithms have been one of the advanced variables for the airfoil shape optimization
techniques adapted for improving GA generally follows sequential genetic
performance. It requires care in balancing algorithm. For the optimization problem
various elements of search space. It involving more than one objective is a very
adversely affects population and force the difficult situation because each objective
evaluation in wrong direction if the high must be simultaneously optimized and each
rated solutions are injected in the population objective plays a vital role in deriving
at the earlier evolution stage. Moreover it optimal solution. In multi objective airfoil
requires special care in encoding. Therefore optimization the concept of dominance is
it is worthwhile to extend the optimization utilized. Three vectors lift, drag, and L/D are
by genetic algorithm. Coupling genetic used as scalar objective functions. The
algorithms on gradient based optimization vector lift (L1)=L(l1, …, li ,…, lN) is said to
techniques gives flexibility in design airfoil dominate another vector lift (L2)=L(m1,…,

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 80


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

mi,…, mN ) if and only if li ≥ mi for all i and number sequence is reset. The fitness values
there exist at least one value of i such that for each chromosome are calculated by
li>mi. The airfoil multi objective fitness function evaluation denoted by
optimization problem is defined by F=F
(f1(A),… fk(A),… fm(A)). The decision ( ) …. (4)
variable vector A consists of 35 independent
co-ordinates. The multi objective
There are three fitness functions used
optimization of airfoil shape profoundly
namely lift, drag and lift by drag ratio are
involve in finding the set of A= ̅ that
defined by
produce non dominated values of F= ̅ ; ̅ is
known as Pareto Front. The idea behind
( ) …. (5)
Pareto Front is for many events; roughly
80% of the effects come from 20% of the

T
causes. In GA optimization design space is ( ) …. (6)
discreetly described by decision variables
i.e. control points Ai. These parameters are ( ) …. (7)
called genes in GA parlance. The decision
variable vector, A is known as chromosome The function represents quantitative
and is denoted by

The j subscript indicates chromosome


ES
number and n superscript indicates genetic
)… (1)
evaluation of lift, drag and lift-drag ratio.
The chromosome with highest fitness i.e.
high lift and L/D ratio and low drag is
ranked 1 with the second highest fitness
ranked 2 and so on. The highest fitness
function chromosome is passed through the
algorithm generation number. Real number next generation. In this paper four
encoding is used to represent all genes. The
modification operators-pass through,
initial generation using the real number random average cross over, perturbation
genes is represented by
A
mutation and mutation are used. The number
of chromosomes modified with each
( ) ... (2) operator is controlled by M vector. The
vector consists of 4
The population size considered is 30. parameters . The value of
Each gene with each chromosome is each M vector element ranges from 0 to 1
IJ

assigned with an initial real number value by and the sum of all four elements is equal to
random number generation between fixed 1. The M vectors are in the ratio 1:3:3:3.
upper and lower limits. The ith gene in an The pass through operator is performed first
arbitrary chromosome is computed using and then the other operator until the airfoil
shape optimization is converged. The
( )( ) … (3) highest individual fitness valued
chromosome is passed to the next
The random number generator used in generation. Thereby guaranteeing that none
this paper provides an integer input seed of the maximum fitness valued
value. If the integer is positive the current chromosomes will get dropped during GA
random number sequence is selected or else iteration. The random average cross over
random number sequence is reset. If same operator is applied on randomly selected two
seed value is selected then also the random chromosomes from the population. The gene

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 81


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

by gene basis combination of the two ribs (one in the root and tip chord and two at
selected chromosomes is achieved by: the mid chord). The skin is made of two
layers. First layer is 2mm balsa sheet and the
( ) second layer is 1 mm fiber glass reinforced
with epoxy resins. Pressure tapings are
…… (5)
provided in the mid chord for the
investigation of the pressure distribution
The perturbation mutation operator is over the wing model. Load cells are used to
applied by first selecting a random find the aerodynamic characteristics such as
chromosome from the population. lift, drag, etc. The composite wing model is
Probability test is performed on each gene in tested at an angle of attack of 2 deg and
the selected chromosome Aj using random Mach number of 0.06. The total weight of
number generator. If the random number is the wing is 500 gram. Various stages of

T
greater than the user defined random number wing fabrication are shown in figure 1 to 2.
R1 then the gene is not modified or else it is
modified by

( )[ ( )
] ES …. (8)

The value of „ϵ‟, a user specified tolerance


which controls the perturbation operator lies
between 0 and 1.0. The random number
mutation is applied by selecting a random
chromosome from the population and a
probability test is performed on each gene in
the selected chromosome Ai. If the random
number is less than the user specified
A
random number R2 then the gene is modified Figure 1: Wind Tunnel Scaled Wing
by Model Structure.

( ) ( )
….. (9)
IJ

In this paper ϵ is assumed as 0.9, R1 as .9and


R2 as .6.

III Wind Tunnel Model preparation,


Testing and Analysis

The optimized aerofoil profile by


GA is used for the fabrication of scaled wing
model of Unmanned Aerial Vehicle (UAV)
having a chord of 15 cm and span of 21 cm
using Balsa wood reinforced by S fiber glass
with epoxy resins. The wing is a single spar Figure 2: Wind Tunnel Scaled Wing
multi rib type having I section spar and 4 Model S fiber Lamination Procedure

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 82


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

IV Results and Discussions Table 1: Upper and Lower Limit of


Design Variables.
Genetic Algorithms are stochastically
based search algorithms which produce Parameter Maximum Minimum
results with statistical variations from (Control Point) Ordinate Ordinate
generation to generation. The initial seed 1 0.0 0.0
depends on random number generator. A 2 0.0208 0.001000
total of 9999 random number valued
3 0.037500 0.011400
chromosomes are generated out of which 30
chromosomes are selected as initial seed 4 0.051800 0.020800
based on their fitness function ranking. The 5 0.0636 0.037500
maximum and minimum range of values for 6 0.072400 0.051800
the design space variables (control points) 7 0.078000 0.063600

T
are fixed based on the co-ordinates of the 8 0.0788 0.072400
base line aerofoil NACA 2412. The 9 0.078000 0.078000
optimized aerofoil profile obtained in each 10 0.078000 0.078800
generation is tested for aerodynamic 11 0.076 0.076700
characteristics by panel method using
ES 12 0.072600 0.056300
Design Foil software. The leading edge,
maximum thickness location and trailing 13 0.066100 0.049600
edge genes are fixed and rest 31 genes are 14 0.056300 0.041300
altered by parallel GA as shown in table 1. 15 0.049600 0.029900
The aerodynamic characteristics of 30 16 0.041300 0.021500
chromosomes for the first generation are 17 0.029900 0.010000
shown in the graph 1 to 3. It was found that 18 0.0 0.0
the highest lift was produced by g1c18 19 -0.010000 -0.022700
chromosome and lowest by g1c29, highest 20 -0.016500 -0.030100
L/D ratio was achieved by g1c18 and lowest
A
21 -0.022700 -0.034600
by g1c2 and lowest drag is obtained by
g1c22 and highest drag is achieved by 22 -0.030100 -0.037500
g1c26. The comparative study of lift, drag 23 -0.034600 -0.041000
and L/D ratio of first generation is presented 24 -0.037500 -0.042300
in table 2. The generation wise optimized 25 -0.041000 -0.042200
IJ

aerofoil profile generated is shown in figure 26 -0.041200 -0.041200


3. 27 -0.038000 -0.040000
28 -0.033400 -0.041200
29 -0.027600 -0.038000
30 -0.021400 -0.033400
31 -0.015000 -0.027600
32 -0.008200 -0.021400
33 -0.004800 -0.015000
34 -0.002000 -0.008200
35 0.000000 0.000000
Figure 3: Optimized Aerofoil Profiles

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 83


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

Table 2: First Generation Lift, Drag and


L/D ratio of 30 Chromosomes.

Airfoil Lift Drag L/D


2412 0.528 0.0084 62.85714
g1c1 0.537 0.0077 69.74026
g1c2 0.171 0.0097 17.62887
g1c3 0.502 0.0077 65.19481
g1c4 0.262 0.0097 27.01031
g1c5 0.547 0.0077 71.03896
g1c6 0.559 0.0077 72.5974

T
Graph 1: l/d vs. chromosome number of
g1c7 0.171 0.0096 17.8125
first generation
g1c8 0.276 0.0098 28.16327
g1c9 0.542 0.0077 70.38961
g1c10 0.544 0.0077 70.64935
ES
g1c11 0.171 0.0096 17.8125
g1c12 0.172 0.0096 17.91667
g1c13 0.575 0.0077 74.67532
g1c14 0.504 0.0077 65.45455
g1c15 0.262 0.0097 27.01031
g1c16 0.261 0.0097 26.90722
g1c17 0.524 0.0077 68.05195
A
g1c18 0.619 0.0077 80.38961
Graph 2: Drag vs. Chromosome number
g1c19 0.274 0.0098 27.95918 of first generation
g1c20 0.509 0.0077 66.1039
g1c21 0.282 0.0098 28.77551
g1c22 0.44 0.0072 61.11111
IJ

g1c23 0.263 0.0097 27.1134


g1c24 0.519 0.0077 67.4026
g1c25 0.586 0.0077 76.1039
g1c26 0.272 0.0098 27.7551
g1c27 0.592 0.0077 76.88312
g1c28 0.564 0.0077 73.24675
g1c29 0.171 0.0096 17.8125
g1c30 0.519 0.0077 67.4026
Graph 3: Lift vs. Chromosome number
for first generation

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 84


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

The aerodynamic characteristics of It was found that the highest lift was
30 chromosomes for 300th generations are produced by g300c4 chromosome and
shown in the graph 4 to 6. lowest by g300c6, highest L/D ratio was
achieved by g300c4 and lowest by g300c6
and lowest drag is obtained by g300c3 and
highest drag is achieved by g300c6. The
comparative study of lift, drag and L/D ratio
is presented in table 3.

Table 3: 300th Generation Lift, Drag and


L/D ratio of 30 Chromosomes.

Airfoil Lift Drag L/D

T
g300c1 0.619 0.0077 80.38961
g300c2 0.592 0.0077 76.88312
g300c3 0.586 0.0077 76.1039
Graph 4: Lift vs. Chromosome for 300th g300c4 0.668 0.0072 92.77778
generation
ES g300c5 0.377 0.008 47.125
g300c6 0.312 0.01 31.2
g300c7 0.432 0.0072 60.00
g300c8 0.506 0.0077 65.71429
g300c9 0.516 0.0077 67.01299
g300c10 0.532 0.0077 69.09091
g300c11 0.536 0.0077 69.61039
g300c12 0.513 0.0077 66.62338
g300c13 0.574 0.0077 74.54545
g300c14 0.523 0.0077 67.92208
A
g300c15 0.584 0.0077 75.84416
g300c16 0.56 0.0077 72.72727
g300c17 0.526 0.0077 68.31169
Graph 5: Drag vs. Chromosome for 300th g300c18 0.48 0.0078 61.53846
generation g300c19 0.626 0.0077 81.2987
IJ

g300c20 0.585 0.0077 75.97403


g300c21 0.593 0.0077 77.01299
g300c22 0.439 0.0072 60.97222
g300c23 0.456 0.0072 63.33333
g300c24 0.482 0.0072 66.94444
g300c25 0.506 0.0075 67.46667
g300c26 0.507 0.0077 65.84416
g300c27 0.524 0.0077 68.05195
g300c28 0.529 0.0077 68.7013
g300c29 0.533 0.0077 69.22078
Graph 6: L/D vs. Chromosome for 300th g300c30 0.544 0.0077 70.64935
generation.

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 85


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

Lift increases from 0.171 for the first


generation to 0.668 for the 300th generation
for a fixed speed of 0.06 Mach, Reynolds
Number 90000 and angle of attack 2 deg.
The L/D ratio also increases from 17.62 to
92.7 from first to 300th generation as shown
in graph 7 and 8. No remarkable reduction
in drag is achieved from first to 300th
generation.

T
Graph 9: Comparative study of
Aerodynamic Characteristics of 2412
Baseline Aerofoil, GA Optimized Aerofoil
with Experimental analysis

Graph 7: L/D vs. Generation


ES V Conclusion

A parallel GA optimization procedure is


developed for the multi objective
optimization of airfoil shape. It uses real
number coding for the representation of
design space of 35 decision variables as
genes and 30 populations to go from
generation to generation. 4 modification
operators – Pass through, Random average
A
cross over, Perturbation mutation and
Mutation are utilized to advance from one
generation to another. The best solution for
each objective is a parato front. For each
case attempted global parato front optimum
is achieved by the convergence of GA
IJ

optimization algorithm. Over 300


generations are considered to study the
convergence efficiency. In some cases
Graph 8: Lift vs. Generation convergence was achieved quickly and in
other cases it was much slower. One value
Three parameters , R1 and R2 variation of caused early convergence and the other
effect on the GA convergence is analyzed. value caused late convergence. R1 has small
The effect of different M vector on GA effect on convergence and R2 has negligible
convergence for number of function effect on convergence for all the generation
evolutions is also analyzed. The comparative considered. The M vector has moderate
study of aerodynamic characteristics of effect on the convergence. The effect of
baseline, GA optimized aerofoil with number of chromosomes used in each
experimental analysis is shown in graph 9. generation and the effect of number of genes

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 86


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

used in each generation will be the further


scope of study. Thus the GA optimization [7] Dominico Quagliarella and A Vicini,
procedure implemented is very attractive for “Coupling Genetic Algorithms and
parallel computing with at least 45 GB Gradient Based Optimization
memory. Techniques in Genetic Algorithms
and Evolution Strategies in
References Engineering and Computer Science”,
John Wiley and Sons Ltd., pages 289-
[1] DE Goldberg, “Genetic Algorithms 309, 1997.
in Search, Optimization and Machine
Learning,” Addition -Wesley, [8] D Quagliarella and A Vicini, “
Reading, MA, 59-88, 1989. Inverse and Direct Airfoil Design
Using a Multi Objective Genetic

T
[2] D Quagliarella and A Della Cioppa, Algorithm”, AIAA Journal, Vol 35,
“Genetic Algorithms Applied to the Issue 9, pages 1499-1505, Sept,
Aerodynamic Design of Transonic 1997.
Airfoils”, Journal of aircraft, Volume
32, Pages 889-891, 1995. [9] AP Giotis, KC Giannakoglou and

[3] K Yamamoto and O Inoue,


ES
“Applications of genetic algorithms
to aerodynamic shape optimization”,
AIAA-95- 1650- CP, 1995.
Jacques Periaux, “ A Reduced Cost
Multi Objective Optimization
Method Based on the Pareto Front
Technique, Neural Networks and
PDM”, in the Proceedings of
ECCOMAS, 2000 Conference,
[4] DJ Doorly, J Peiro, T Kuan and JP Barcelona, Spain, 11-14, September
Oesterle ”Optimization of Airfoils 2000.
Using Parallel Genetic Algorithms”,
A
in Proceedings of 15th [10] R Smith, “A First Investigation into
International Conference on a Voxel Based Shape Presentation
Numerical Methods in Fluid Technical Report”, Manufacturing
Dynamics, Monterey, 1996. Planning Group, Dept. of
Mechanical Engineering, University
[5] CZ Janikow and Z Michalewicz, “An of Edinburgh, 1995.
IJ

Experimental Comparison of Binary


and Floating Point Representation in [11] K.Krishna Kumar, R.Swaminathan,
Genetic Algorithms”, in the S.Garg and S.Narayana swamy,
Proceedings of the 4th International “Solving Large Parameter
Conference on Genetic Algorithms, Optimization Problems Using
Pages 31-36, 1991. Genetic Algorithms”, proceedings of
the Guidance, Navigation and
[6] A Oyama, S Obayasha and K Control Conference, page 449-460,
Nakahashi, “Wing Design Using 1995
Real Coded Adaptive Range Genetic
Algorithm”, in the Proceedings of [12] S.Mulgund, K H Arper, K K Krishna
IEEE International Conference on Kumar and G Zacharias, “Air
Systems, Man and Cybernetics, 1999. Compact Tactics Optimization Using

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 87


G MANIKANDAN et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 3, Issue No. 2, 078 - 088

Stochastic Genetic Algorithms” G. Manikandan was born


Proceedings of IEEE International on 12th January 1969 from
Conference on Systems , Man and the famous big temple city
Cyber tics, pages 3136-3141, 1998 Thanjavur, Tamil Nadu. He
obtained his Engineering
[13] M Arakawa, I Hagiwara, Graduation (Mech) in the
“Development of Adaptive Real year 1994 from Institution of
Range Genetic Algorithm”, JSME Engineers (India), Calcutta and M.Tech
International Journal Series C, Vol (CAD/CAM) in the year 2002 from JNTU,
41, Issue 4, pages 969-977, 1988 Hyderabad. He put up 16 years of colorful
service in Indian Air Force. In his credit, he
[14] A Oyama, S Obayashi and K overhauled 365 Rolls Royce Viper Turbojet
Nakahashi, “Wing Design Using Real Engine fitted on Kiran Aircraft and Carried

T
Coded Adaptive Range Genetic out Structural Repairs and maintenance of
Algorithm” , Proceedings of IEEE Cheetah and Chetak helicopters and Kiran
International Conference on Systems, aircraft. He was team leader for several
Man and Cybertics, 1999 Structural re-fabrications of Ardhra and
Rohini Gliders. He developed many Un-
[15]
ES
B Naujoks, L Willmes, W Haase, T
Back and M Schurtz, “Multi point
Airfoil Optimization using Evoluation
Strategies”, European Congress on
Computational Methods in Applied
manned Aerial Vehicles (UAV). Presently,
his contributions are in the area of aerofoil
shape optimization and flutter analysis. He
was awarded best in trade and all-rounder
for Kiran Aircraft in the year 2000.
Sciences and Engineering,
ECCOMASS 2000, Barcelona, Spain, M. Ananda Rao obtained
Sept 2000 B.E (Mech) in 1968, M.Tech
(Machine Design) in 1970
and M.Tech (Industrial
A
[16] M Anderson, J Burkhalter and R
Jenkins, “Missile Aerodynamic Shape Engg) in 1984. He was
Optimization Using Genetic awarded PhD from IIT,
Algorithm”, Journal of Space Craft Madras in the area of “Machine Dynamics”.
and Rockets, Vol 37, issue 5, pages He worked over 33 years in Andhra
663-669, September 2000 University at various capacities. He worked
IJ

in the Link Interchange Program with UK


[17] M Anderson and G Gebert, “Using Universities for about 03 years by British
Parato Genetic Algorithms for Council and Government of India. He was
Preliminary Subsonic Wing Design” , awarded three times “The Best Researcher
AIAA paper no.96-4023-CP, 1996 Award” in the year 1992, 1999 and 2001. He
worked as a technical adviser for Altair
Company for the development of software in
the domain of solvers. He is one of the
renowned researchers in the area of
Vibration and Condition Monitoring in the
World. He was the nucleus in the starting of
Condition Monitoring Society of India.

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 88

También podría gustarte