Hello all,
I am dealing minimization problem for sizing of PV and ESS . I am using particle swarm optimization algorithm as solver (matlab).
Initially i set the lb and ub to very large number and then started trying out the different values finally arrived at lb:[20 4];Ub:[35 8].
It worked fine. Now when i change the variables like EV battery capacity , no of people , the simulation doesn't end. It keeps going on and on . a)I have some important variables like no of people to decide the load consumption. please find the attached file to have a look at the functions modeledb)Ev capacity, etc
1) What is the logic behind setting the bounds in PSO ?
2) If the variables are changed are the logics too to be changed accordingly ? i mean every time i change the ev battery capacity should the crate , etc needed to be changed?
3) How to assess the authenticity of results ?
thanks in advance !
Dear Chamanth Kandalam ,
I would answer your first question.
The bound settings in PSO is problem dependent. And the logic is that variables cannot take an infinite range, they have to be bounded.
This depends on the problema. each decisión variable maintains a diferent value
Abubakar Bala so on what basis u decide the bounds ? is it that we choose some random bounds and keep limiting them until the optimum value repeats ?
Metaheuristic search is like a treasure hunt. You can't search the whole planet for the treasure. You must have some set boundaries in which to search.
Who sets the boundaries?
You the designer, based on your knowledge of the problem you want to solve.
Imagine there is a treasure hunt in location A. So your search boundary has to be the perimeter of location A. You don't need to look into location B, C. Why ? Because there is no treasure there. So why look there?
I hope my rough analogy makes sense.
Suppose you need to optimize 10 variables in order to maximize a function. Now before beginning the search process you have to provide a range to these variables. The logic is to reduce the effort of your agents by directing them to search within the respective bounds. Without the bounds it will become nearly impossible for the birds to search the entire multidimensional (10D) search space . Now what upper and lower limits you will choose for the variables will depend on the application which you are trying to optimize.
Abubakar Bala Tuhin Deshamukhya your explanations were on point . U also mean to say that designer needs to the search space where probable solution might be in .
Thanks for the insights!
- Badges
- Science topic