OUTLINE

About Grey Wolf

Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
.c
ss
re
dp
or
Grey Wolf Optimizer(GWO)

.w
ar
m
ku
Dr. Rajesh Kumar
sh
Associate Professor
je
ra

Electrical Engineering
dr

MNIT Jaipur
//

April 17, 2015

s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
OUTLINE

.c
ss
re
dp
1 About Grey Wolf

or
.w
2 Developers of Algorithm

ar
m
3 Wolf behaviour in nature
ku
sh
4 Algorithm development
je
ra

Example
dr

5
//

6 Advantages over other techniques

s:
tp
ht

7 Application on Unit commitment problem

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
About Grey Wolf

.c
ss
re
dp
or
Wolf is characterised by power full teeth, bushy

.w
ar
tail and lives and hunts in packs. The average

m
group size is 5-12.

ku
Their natural habitats are found in the mountains,
sh
je
forests, plains of North America, Asia and Europe.
ra

Grey wolf (Canis lupus) belongs to Canidae family.

dr
//

Grey wolves are considered as apex predators,

s:

meaning that they are at the top of the food

tp
ht

chain.

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Developers of Algorithm

.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr

Andrew Lewis
//
s:

Seyed Mohammad
tp

Mirjalili
ht

Seyedali Mirjalili

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature Social behaviour
Algorithm development Hunting behaviour
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Wolf behaviour in nature

.c
ss
re
dp
or
.w
ar
Social behaviour

m
ku
Hierarchy exits in pack sh
is the leader and decision maker.
je
ra

and assist in decision making.

dr

Rest of the wolves () are followers.

//
s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature Social behaviour
Algorithm development Hunting behaviour
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Wolf behaviour in nature

.c
ss
re
dp
or
.w
Hunting behaviour

ar
Group hunting behaviour is of equal interest in studying

m
ku
optimization. sh
Tracking, chasing, and approaching the prey.
je
ra

Pursuing, encircling, and harassing the prey until it stops

dr

moving.
//

Attacking the prey.

s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature Social behaviour
Algorithm development Hunting behaviour
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp
ht

Approach, track and pursuit

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature Social behaviour
Algorithm development Hunting behaviour
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp
ht

Pursuit

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature Social behaviour
Algorithm development Hunting behaviour
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp

Harass
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature Social behaviour
Algorithm development Hunting behaviour
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp
ht

Encircling

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature Social behaviour
Algorithm development Hunting behaviour
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp

At the end, when the prey stops, wolves make a approximate

ht

regular polygon around it and lay down

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Algorithm development

.c
ss
re
dp
or
.w
Social hierarchy

ar
In order to mathematically model the social hierarchy of wolves when

m
ku
designing GWO, we consider the fittest solution as the alpha ().
sh
Consequently, the second and third best solutions are named beta
je

() and delta () respectively. The rest of the candidate solutions

ra

are assumed to be omega (). In the GWO algorithm the hunting

dr
//

(optimization) is guided by , , and . The wolves follow these

s:

three wolves.
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Encircling prey

.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Encircling Prey: Mathematical Modeling

.c
ss
re
dp
or
.w
ar
The mathematical model of the encircling behaviour is represented

m
ku
by the equations: sh
D = |CXp AX (t)| (1)
je
ra

X (t + 1) = Xp (t) AD (2)
dr
//
s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Encircling Prey: Mathematical Modeling

.c
ss
re
dp
A and C are coefficient vectors given by:

or
.w
ar
A = 2ar1 a (3)

m
ku
sh C = 2r 2 (4)
je

t is the current iteration

ra

X is the position vector of a wolf

dr
//

r1 and r2 are random vectors [0, 1] and a linearly varies from

s:

2 to 0
tp
ht

More description in later slides

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Hunting

.c
ss
re
dp
or
Grey wolves have the ability to recognize the location of prey

.w
and encircle them.

ar
m
The hunt is usually guided by the alpha. The beta and delta

ku
might also participate in hunting occasionally.
sh
However, in an abstract search space we have no idea about
je
ra

the location of the optimum (prey).

dr

In order to mathematically simulate the hunting behaviour, we

//
s:

suppose that the alpha, beta and delta have better knowledge
tp

about the potential location of prey.

ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Hunting

.c
ss
re
dp








D = | C 1 . X (t) X (t)|, D = | C 2 . X (t) X (t)|, D = | C 3 . X (t) X (t)| (5)

or

.w
X 1 = X (t) A 1 .( D ), X 2 = X (t) A 2 .( D ), X 3 = X (t) A 3 .( D ) (6)

ar

X1 + X2 + X 3

m
X (t + 1) = (7)
3

ku




where t indicates the current iteration, X (t), X (t) and X (t) are the position of the gray wolves , and
sh


at t h iteration, X (t) presents the position of the gray wolf at t th iteration.
t
je
ra

A (.) = 2

a .rand(0, 1)

a (8)
dr
//

C (.) = 2.rand(0, 1) (9)
s:
tp

Where

a is the linear value varies from 2 to 0 according to iteration. A (.) and C (.) are the coefficient vector of
ht

, and wolfs.

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Attacking prey & Search for prey

.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Example

.c
ss
re
dp
or
.w
ar
m
minimization of Korn function

ku
sh
je

f(x1 ,x2 ) = min{(x1 5)2 + (x2 2)2 }

ra
dr
//
s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Iteration 1

.c
ss
re
dp
x1 x2 f (x) x1 x2 f (x)

or
1 6.1686 4.4100 7.1739 4.7372 3.3048 1.7717

.w
2 6.2104 4.0935 5.8479 4.8148 3.4931 2.2637

ar
3 7.4231 8.3880 46.6773 5.9444 3.4433 2.9751

m
ku
4 2.8950 0.8703 5.7074sh
5 6.1062 3.7275 4.2079
je
6 6.3458 3.2158 3.2896
ra

7 7.5690 6.1457 23.7866

dr

8 6.2471 4.0456 5.7397

//

9 6.9965 4.5846 10.6663

s:

10 4.7372 3.3048 1.7717

tp
ht

11 4.8148 3.4931 2.2637

12 5.9444 3.4433 2.9751

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Update process

.c
ss
re
dp
or

itr

.w
D = |2.rand().[4.7372, 3.3048] [6.1686, 4.4100]| a = 2 2.( maxitr )

ar
X1 = [4.7372, 3.3048] (2
a .rand(0, 1)

a )D

a = 2 2.( 13 )

ku
D = |2.rand().[4.8148, 3.4931] [6.1686, 4.4100]|
sh


a = 1.3333
X2 = [4.8148, 3.4931] (2
a .rand(0, 1)

a )D
je

ra

D = |2.rand().[5.9444, 3.4433] [6.1686, 4.4100]|

dr

X = [5.9444, 3.4433] (2
a .rand(0, 1)

a )D
//

X1 + X2 + X3
s:

X (1, :) = = [4.0487, 2.6051]

tp

3
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Iteration 2

.c
ss
re
dp
x1 x2 f (x) x1 x2 f (x)

or
1 4.0487 2.6051 1.2710 4.2452 2.6600 1.0054

.w
2 4.6492 3.0427 1.2103 4.1136 2.5382 1.0754

ar
3 5.4633 3.6633 2.9813 5.0927 3.1546 1.3418

m
ku
4 5.6096 3.5901 2.9001sh
5 4.6582 3.0302 1.1781
je
6 4.7476 3.3369 1.8509
ra

7 4.2452 2.6600 1.0054

dr

8 4.9026 3.2497 1.5712

//

9 4.5202 2.9588 1.1495

s:

10 5.3971 3.5432 2.5392

tp
ht

11 4.1136 2.5382 1.0754

12 5.0927 3.1546 1.3418

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Iteration 3

.c
ss
re
dp
x1 x2 f (x) x1 x2 f (x)

or
1 4.4838 2.7843 0.8816 4.5634 2.8257 0.8725

.w
2 4.5634 2.8257 0.8725 4.5696 2.8291 0.8727

ar
3 4.5899 2.8395 0.8730 4.5750 2.8321 0.8730

m
ku
4 4.7486 2.9400 0.9467sh
5 4.6340 2.8684 0.8881
je
6 4.5957 2.8445 0.8767
ra

7 4.5830 2.8366 0.8738

dr

8 4.5787 2.8339 0.8729

//

9 4.5750 2.8321 0.8730

s:

10 4.5724 2.8306 0.8727

tp
ht

11 4.5703 2.8295 0.8727

12 4.5696 2.8291 0.8727

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Flow chart

.c
ss
re
dp
or
.w
ar
m
ku
sh
je
ra
dr
//
s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Advantages over other techniques

.c
ss
re
dp
or
.w
Easy to implement due to simple structure.

ar
m
Less storage requirement than the other techniques.

ku
Convergence is faster due to continuous reduction of search
sh
space and Decision variables are very less (, and ).
je
ra

It avoids local optima when applied to composite functions also.

dr
//

only two main parameters to be adjusted (a and C ).

s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Unit Commitment Problem

.c
ss
re
dp
Unit Commitment (UC) is a very significant optimization task,

or
.w
which plays an important role in the operation planning of

ar
power systems.

m
ku
UCP is considered as two linked optimization decision pro-
sh
cesses, namely the unit-scheduled problem that determines on/off
je

status of generating units in each time period of planning hori-

ra
dr

zon, and the economic load dispatch problem.

//

UCP is a complex nonlinear, mixed-integer combinational opti-

s:
tp

mization problem with 01 variables that represents on/off sta-

ht

tus.

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Unit commitment problem

.c
ss
re
dp
or
Table : Total costs of the BGWO method for test systems

.w
ar
No. of Best Average Worst Std. De- CPU

m
ku
Unit Cost ($) Cost ($) Cost ($) viation Time
(Sec)
sh
je

10 563937.3 563976.6 564017.7 40.2 31.3

ra

20 1124687.9 1124837.7 1124941.1 128.7 58.7

dr

40 2248280.0 2248397.6 2248614.0 174.2 124.6

//

60 3367893.4 3367881.1 3367933.4 37.9 216.9

s:

80 4492399.4 4492608.1 4492672.5 154.4 347.5

tp
ht

100 5612309.4 5612377.2 5612496.3 96.9 505.6

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
Performance Comparison

.c
ss
re
dp
or
.w
Table : Comparison With Other Algorithms

ar
m
ku
10 20 40 60 80 100
LR 565825 1130660 sh 2258503 3394066 4526022 5657277
GA 565825 1126243 2251911 3376625 4504933 5627437
EP 564551 1125494 2249093 3371611 4498479 5623885
je

MA 565827 1128192 2249589 3370820 4494214 5616314

ra

GRASP 565825 1128160 2259340 3383184 4525934 5668870

dr

LRPSO 565869 1128072 2251116 3376407 4496717 5623607

PSO 564212 1125983 2250012 3374174 4501538 5625376
//

IBPSO 563977 1125216 2248581 3367865 4491083 5610293

s:

BFWA 563977 1124858 2248372 3367912 4492680 5612619

BGWO 563937.3 1124684.8 2248280.0 3367893.4 4492399.4 5612309.4
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)

 OUTLINE
About Grey Wolf
Developers of Algorithm
Wolf behaviour in nature
Algorithm development
Example
Advantages over other techniques

/
m
Application on Unit commitment problem

o
.c
ss
re
dp
or
.w
ar
Thank You . . .

m
ku
sh
Mail: rkumar [email protected]
je
ra
dr
//
s:
tp
ht

Dr. Rajesh Kumar Grey Wolf Optimizer(GWO)