Number of particle ( hybrid Particle Swarm and Pattern Search Algorithm)+genopt
I'm trying to optimize a building design, by coupling EnergyPlus and an optimization tool GenOpt.
For that , I'm using a Hybrid Generalized Pattern Search Algorithm with Particle Swarm Optimization Algorithm (GPSPSOCCHJ) for discrete and continuous variable, but I can't understand how to determine the number of the particle for this optimization.
Please, give me an idea about the mean of this parameter(NumberOfParticle).
The command file is :
* GenOpt example command file
MWetter 06/18/2003
*/
Vary{
Parameter{ // azimuthangle
Name = x1;
Min = -180.0;
Ini = 0.0;
Max = 180.0;
Step = 90.0;
}
Parameter{ // high windows size
Name = x2;
Min = 0.8;
Ini = 1.6;
Max = 1.6;
Step = 0.4;
}
Parameter{ //infiltrationrate
Name = x4;
Min = 0.5;
Ini = 0.5;
Max = 2.0;
Step = 0.5;
}
Parameter{ //ventilationrate
Name = x5;
Min = 0.0;
Ini = 0.0;
Max = 2.0;
Step = 0.5;
}
Parameter{ // windowstype
Name = x6;
Ini = 1;
Values = "100, 101" ;
Type = SET;
}
Parameter{ // wall-exterior-type
Name = x8;
Ini = 1;
Values="EWQ23M91,EWQ23M51,EWQ21M51,EWQ21M91,EWQ25M51,EWQ25M91,EWQ27M51,EWQ27M91,EWQ29M51,EWQ29M91" ;
Type = SET;
}
Parameter{ // roof-exterior-type
Name = x10;
Ini = 1;
Values = "ERQ23M91,ERQ23M51,ERQ21M51,ERQ21M91,ERQ25M51,ERQ25M91,ERQ27M51,ERQ27M91,ERQ29M51,
ERQ29M91" ;
Type = SET;
}
}
OptimizationSettings{
MaxIte = 10000;
MaxEqualResults = 200;
WriteStepNumber = false;
UnitsOfExecution = 4;
}
Algorithm{
Main = GPSPSOCCHJ;
NeighborhoodTopology = vonNeumann;
NeighborhoodSize = 5;
NumberOfParticle = ;
NumberOfGeneration = 50;
Seed = 1;
CognitiveAcceleration = 2.05;
SocialAcceleration = 2.05;
MaxVelocityGainContinuous = 0.2;
MaxVelocityDiscrete = 4;
ConstrictionGain = 1;
MeshSizeDivider = 2;
InitialMeshSizeExponent = 0;
MeshSizeExponentIncrement = 1;
NumberOfStepReduction = 4;
}
I think your question isn't about GenOpt or energy modeling, it's a math question: it's about what Particle Swarm algorithm actually is, and there's plenty of literature on the subject out there, just google it.