Cite

2 Recommendations

Mohammadhossein Homaei

University of Extremadura

Hello there, 

You can use of this source code in Matlab for Accelerated Particle Swarm Optimization. I think is useful for you.

Description

This is an accelerated PSO (APSO). APSO does not use velocities or any inertia parameter.

A full program for solving nonlinear constrained optimization problem (welded beam design as an example) is provided, which can be extended to solve other continuous optimization problems.

MATLAB release

MATLAB 7 (R14)

• APSO.z

  ip

  5.49 kB

Cite

2 Recommendations

Mahamad Nabab Alam

National Institute of Technology Warangal

PSO Codes:

https://www.researchgate.net/publication/296636431_Codes_in_MATLAB_for_Particle_Swarm_Optimization?ev=prf_pub

Code Codes in MATLAB for Particle Swarm Optimization

Cite

Mahamad Nabab Alam

National Institute of Technology Warangal

useful link

https://www.researchgate.net/publication/280244004_A_comparative_study_of_metaheuristic_optimization_approaches_for_directional_overcurrent_relays_coordination

Article A comparative study of metaheuristic optimization approaches...

Cite

Mahamad Nabab Alam

National Institute of Technology Warangal

https://www.researchgate.net/publication/297245624_Particle_Swarm_Optimization_Algorithm_and_its_Codes_in_MATLAB

Research Particle Swarm Optimization: Algorithm and its Codes in MATLAB

Cite

Similar questions and discussions

How should I optimize the value of an array using particle swarm? can anyone help me how to write code in python for this?

Question

2 answers

• Asked 1 July 2019

• Saurabh Kumar

I am working to minimize the error of the autoencoder predicted value and actual value error. How to use fitness function mse in PSO using python.

View

How to use PSO-BP NN solve two-spiral classification ?

Question

3 answers

• Asked 10 January 2019

• Deleted profile

I want to use PSO algorithm to replace the GD algorithm to update the weight of 2-layer NN, solve the problem of two-spiral classification, there are two inputs and one output, the right result is 0.1 or 0.9 ,but the result I got is about 0.4-0.6. It means my method is useless.

I have tried different parameters value of PSO ,and the hidden nodes number but the best result is same and the calculation speed is too slow compare with NN or PSO only.

I have read some papers which they get good result in accuracy and speed and my method is same as their method in logically.

in the end I attach my coding(by matlab).the result of accuracy is about (152-60)/152 = 60%

I hope I can get some hints from you. thank you.

net = newff([-0.5 0.5;0 1],[20,1],{'tansig','tansig'},'trainlm');

swarmCount = 200;

swarmLength = 81;

vMax = 10;

pMax = 2;

ac = 0;

swarm = rand(swarmCount,swarmLength);

v = rand(swarmCount,swarmLength);

swarmfitness = zeros(swarmCount,1,'double');

pBest = rand(swarmCount,swarmLength);

pBestfitness = zeros(swarmCount,1,'double');

pBestfitness(:,:) = 100;

gBest = rand(1,swarmLength);

gBestfitness = 100;

c1 = 2;

c2 = 2;

w = 0.5;

maxEpoch = 500;

errGoal = 0.01;

epoch = 1;

while(epoch < maxEpoch && gBestfitness > errGoal)

for i = 1:swarmCount

net.iw{1,1} = [swarm(i,1:20);swarm(i,21:40)]';

net.b{1} = swarm(i,41:60)';

net.lw{2,1} = swarm(i,61:80);

net.b{2} = swarm(i,81:81);

tout = sim(net,input');

err = sum((tout - target').^2)/length(target');

swarmfitness(i,1) = err;

if(pBestfitness(i,1) > err)

pBestfitness(i,1) = err;

pBest(i,:) = swarm(i,:);

if(gBestfitness > err)

gBestfitness = err;

gBest(1,:) = swarm(i,:);

end

end

end

for i = 1:swarmCount

v(i, :) = w * v(i, :) + c1 * rand(1, 1) * (pBest(i, :) - swarm(i, :)) + c2 * rand(1, 1) * (gBest(1, :) - swarm(i, :));

tmp = v(i, :) > vMax;%change

v(i, tmp) = vMax;

swarm(i, :) = swarm(i, :) + v(i, :);

tmp = find(swarm(i, :) > pMax);

swarm(i, tmp) = pMax;

end

epoch = epoch + 1;

end

net.iw{1,1} = [gBest(1,1:20);gBest(1,21:40)]';

net.b{1} = gBest(1,41:60)';

net.lw{2,1} = gBest(1,61:80);

net.b{2} = gBest(1,81:81);

tout = sim(net,input');

figure,plot(input,target,'k-');

hold on;

toutt = abs(tout-target');

for ii = 1 : 152

if( toutt(1,ii) >= 0.4)

ac = ac + 1 ;

end

end

plot(input,tout,'b-');