Particle Swarm Optimisation

Question 1

What is Particle swarm optimisation

Answer 1

behaviour of flocks or swarms

Question 2

What does PSO do?

Answer 2

Performs optimization
Its population based with multipoint search
It is a metaheuristc
Each candidate solution is a particle

Question 3

What does each particle have:

Answer 3

Position
velocity
fitness value

Question 4

Where to particles move and what is the movement based on?

Answer 4

They move through search space

and the movement is based on particle position and velocity and
local + global best positions

Question 5

PSO Alg:

Answer 5

init all xi,vi and pbesti values

while(termination criteria is not met){
for i <- 1, N {
calculate F(xi)
if(F(xi) < F(pbesti) then pbesti = xi
if F(xi) < F(g best) then g best = xi
update all vi and xi using exuation 1 and 2

Question 6

What are the update equations?

Answer 6

1 : vi(t+1) = w*vi(t) + l1 * r1[pbesti(t) -xi(t)] + l2 * r2[g best -xi(t)]

2: xi(t+1) = xi(t) + vi(t+1)

Question 7

What are the PSO Parameters?

Answer 7

Swarm size (number of particles)
Vmax (max velocity permitted (explore and exploit)
inertia weight w
learning factors l1 and l2
termination criteria