Hi Dear all,
I am trying to implement the Bound Handling techniques for a particle velocity and position. I am using Deterministic Back and Nearest technique for velocity and position respectively. I have implemented these two techniques in two different methods (in the program). I am getting slightly different results from these methods. I am not sure which one is the right method. Please let me comment. If both of the following methods are incorrect then please let me suggest the correct way of implementation.
Method-1
for (i = 1 to SWARM_SIZE)
{
for (j=1 to PROBLEM_SIZE)
{
Compute V(t+1) using standard PSO velocity equation
If (V(t+1) < V_min || V(t+1) > V_max)
V(t+1) = - 0.5 * V(t+1) // this deterministic back technique
Compute X(t+1) using standard PSO position equation
If (X(t+1) < X_min)
X(t+1) = X_min
Else if (X(t+1) > X_max)
X(t+1) = X_max
}
}
Method-2
for (i = 1 to SWARM_SIZE)
{
for (j=1 to PROBLEM_SIZE)
{
Compute V(t+1) using standard PSO velocity equation
Compute X(t+1) using standard PSO position equation
If (X(t+1) < X_min)
{
X(t+1) = X_min
V(t+1) = - 0.5 * V(t+1) // this deterministic back technique
}
Else if (X(t+1) > X_max
{
X(t+1) = X_max
V(t+1) = - 0.5 * V(t+1) // this deterministic back technique
}
}
}
Many thanks
Best regards,
M Khan
An important work on PSO was done in the following paper :
"COMPARISON OF PARALLEL PARTICLE SWARM OPTIMIZERS FOR GRAPHICAL PROCESSING UNITS AND MULTICORE PROCESSORS"
on International Journal of Computational Intelligence and Applications
Vol. 12, No. 1 (2013) 1350006 (20 pages)
#.c Imperial College Press
DOI: 10.1142/S1469026813500065
by : VINCENT ROBERGE* and MOHAMMED TARBOUCHI.
it may help you to stand the fundamental expression of the PSO algorithm and implement it in any computational model.
good luck;
It is even better to use the following;
for i=1:n
for j=1:m
if x(i,j)UB(j)
x(i,j)=gbest(j);
end
end
end
- Badges
- Science method