Sie sind auf Seite 1von 23

OPTIMISATION

6 January 2011

PSO
PSO
PSO is a robust stochastic
optimization technique based
on the movement and
intelligence of swarms.
It
It was developed in 1995 by
James Kennedy (social(socialsychologist) and Russell
Eberhart (electrical
engineer).
6 January
2011

PSO
Steer Towards the Center
Match Neighbors Velocity
Avoid Collision
Seek Roost

6 January 2011

PSO



6 January 2011

Similarities
with
computation techniques

evolutionary

Searches for optima by updating


generations from population of random
solutions.
solutions.
The potential solutions, called particles,
fly through the problem space by
following the current optimum
particles.
particles.
Few parameters to adjust compared
with GA.
GA.

Presentation
 Artificial Life
 The Algorithm
 Genetic Algorithm
Vs PSO

6 January 2011

Artificial Life


Studies how computational


techniques can help in
studying biological
phenomena

Studies
Studies how biological
techniques can help out with
computational problems
6 January 2011

Popular Swarm Methods In Computational


Intelligence
Ant Colony Optimization (ACO)
ACO was inspired by the behaviors of
ants and has many successful
applications in discrete optimization
problems.
Particle Swarm Optimization (PSO)
The concept originated as a simulation of
simplified social system.
The original intent was to graphically
simulate the choreography of bird of a bird
block
or fish
6 January
2011school.

The Algorithm
simulating the behaviors of bird
flocking.

Example scenario
Objective
A group of birds are randomly searching
food in an area

Constrain
There is only one piece of food in the area
being searched
The birds are unaware where the food is.
6 January 2011

But they know how far the food is in each


iteration.

The Strategy
Follow the bird which is
nearest to the food
6 January 2011

Bird Flock Strategy to


Computational Techniques
 Every Solution is Bird

is

Particle

 Particle Optimization depends Fitness


Function and Velocity
 Each particle follow current optimum
particle

6 January 2011

Bird Flock Strategy to


Computational Techniques
 Each particle is updated by following two
"best" values PBEST & GBEST
 PBEST Best Solution achieved so far
 GBEST - Best value obtained so far by
any particle in the population during
PSO
6 January 2011

Flow Chart
General PSO Algorithm
Start

For each particles position (p)


evaluate fitness
If fitness(p) better than
fitness(pbest) then pbest= p
Set best of pBests as gBest
Update particles velocity (eq. 1) and
position (eq. 3)

6 January 2011

Stop: giving gBest, optimal solution.

Loop until max iter

Loop until all


particles exhaust

Initialize particles with random position


and velocity vectors.

Equation to find
Particle Velocity and
Position
v[] = v[] + c1 * rand() * (pbest [] - present[]) + c2 * rand() * (gbest [] - present[]) (a)
present[] = present [] + v[] (b)

6 January 2011

v[] is the particle velocity, present [] is


the current particle (solution). pbest[]
and gbest[] are defined as stated
before. rand () is a random number
between (0,1). c1, c2 are learning
factors. usually c1 = c2 = 2.

6 January 2011

PSO Model
vi (t +1) = wvi (t) +c1 rand()(pi xi (t))+c2 rand()(pg xi (t))

xi (t + 1) = xi (t ) + vi (t )
My best
perf.
Here I
am! x

pi
pg
v

6 January 2011

The best perf.


of team

PSO Vs GA
Similarity
 Starts with randomly generated population
 Fitness Values
 Updates and Searches with random technique
 No guaranteed success

Dissimilarity

Advantages

No evolution operators

 A few parameters

Particle follow the current optimum particles

 Easy to implement

Updates with internal velocity

 Finds the best solution fast

Availability of Memory
6 January 2011

Case Study
Overrunning Clutch
The contact angle Y is expressed as
Y= f(x1, x2, x3) =

a.cos ((x1+x2) / (x3-x2))

a - constant
The nominal values of xi (i=1,2,3) are 2.17706,
0.90000 and 4.00000 inches respectively. The
nominal value and tolerance of angle Y are
0.122 + 0.035 rad.

6 January 2011

The tolerance Boundaries


t1 & t3 = 1 to 120 (in 10 4 inches)
t2 = 1 to 5 (in 10 4 inches)
where,

t1= tolerance of Hub


t2= tolerance of Roller
t3= tolerance of Cage

Objective Function
The following are the manufacturing cost functions.

M (t1 ) = 0.731 +
M (t 2 ) = 8.3884 +

0.058
t1

0.688

5.7807
t 2 0.0784

M (t 3 ) = 0.978 +

6 January 2011

0.0018
t3

where, t1, t2, t3 are the


single side tolerances
value of hub, roller and
cage

Total manufacturing cost = M(t1) + M(t2) + M(t3)

The combined objective function is,

Minimize, y (t i ) = [M (t i )]
i =1

= 33.3066 +

6 January 2011

0.058
t1

0.688

23.1228
t2

0.0784

0.0018
t3

Table 1 Cost tolerance data for the clutch (tolerance in 10 4 inches, cost in dollars)

Hub
Tol.

Roller

Cost

Tol.

Cage

Cost

Tol.

Cost

19.38

3.513

18.637

13.22

2.48

12.025

5.99

1.24

5.732

16

4.505

1.24

2.686

30

2.065

16

1.20

16

1.984

60

1.24

30

0.413

30

1.447

120

0.825

60

0.413

60

1.200

120

0.372

120

1.033

6 January 2011

The Inertia Factor


When the inertia factor is used, the equation for updating velocities is changed
to:

Vid = Vid + 1 rnd () ( Pid X id ) + 2 rnd () ( Pgd X id )


Where is initialized to 1.0 and is gradually reduced over time (measured by
cycles through the algorithm).

The Constriction Coefficient


In 1999, Maurice Clerc developed a constriction Coefficient for PSO.

Vid = K [ Vid + 1 rnd () ( Pid X id ) + 2 rnd () ( Pgd X id )]


Where K = 2/|2 - - sqrt(2 - 4)|,
= 1 + 2, and
> 4.
6 January 2011

Table 2 Comparison of results for different values of A

Tolerance by PSO
A

t1

t2

Y PSO

0.00239

11.640

11.600

t3

0.00495

0.00484

0.0005

0.00246

11.610

11.601

52

0.00444

0.0005

0.00225

11.793

11.725

100

0.00480

0.0005

0.00208

11.911

11.910

300

0.00465

0.0005

0.00265

12.460

12.334

520

6 January 2011

0.00425

0.0005

Y SIP

0.0005

0.00267

13.047

13.011

6 January 2011

Das könnte Ihnen auch gefallen