Constrained Particle Swarm Optimization

Hi,

I have 'N' particles with one-dimensional variable a(=accelration). However, each particle has a different range (lower limit, upper limit). If I want to use PSO to optimize the cost function which includes v (=valocity) and p (=postion) and inequality constraint of [position(N) - position (N-1) >= some value], then how should I do it?

Answer 1

[Walter Roberson]

See https://www.mathworks.com/matlabcentral/fileexchange/25986-constrained-particle-swarm-optimization

[Mudasser Seraj]

Thank you

[Lalit Upadhyaya]

Dear walter,

The above code does not give correct resuts for LB & UB problems.

[Danish Ali]

Dear Walter,@Walter Roberson

Thank you. However, this code doesn't run well as compare to genetic algorithm. Is there any way, we can improve the results (I have tried Hybrid function too) ? Any suggestions will be appreciated.

[Walter Roberson]

@Danish Ali

Is there mathematical reason to expect that particle swarm algorithms will run about as well as genetic algorithm in all cases? Or is there a restricted set of situations in which particle swarm is expected to be as good or better, and if so then does your particular situation fit within those cases?

@Lalit Upadhyaya

When the only constraints are LB and UB, then particleswarm() handles that. https://www.mathworks.com/help/gads/particleswarm.html