Mixed Integer Linear Programming using GA for solution generation of network topology
조회 수: 3 (최근 30일)
이전 댓글 표시
Hello,
I'm trying to use mixed integer linear programming (MILP) as recommended by Alan Weiss to solve an optimization network topology problem.
I've set all of the parameters, as documentation about MILP states (adapted to my problem), but when I run the GA function some generated solutions do not respect what I initially had defined.
I have 41 binary variables corresponding to all branches of a network composed by 30 bus with 6 generators. For this network to be radial, I need in all generated solutions to be valid ones, having exactly 24 (30-6) active branches, corresopnding to 24 1's in the solution vector x.
The problem is that I am having many solutions being evaluated by ff that do not verify this restriction.
I don't know if my text is clear, but if you can help me I really appreciate.
Best regards, Vitor Ribeiro.
댓글 수: 0
채택된 답변
Matt J
2014년 8월 26일
편집: Matt J
2014년 8월 26일
If it's a MILP, it might be more robust, as well as easier, to use intlinprog, as opposed to ga. Beyond that, we cannot say what's wrong without seeing your code.
댓글 수: 5
Matt J
2014년 9월 10일
편집: Matt J
2014년 9월 10일
I can see that when it finds a feasible solution GA rapidly learn and apparently converge because it is evaluating more feasible solutions around the 1st one it reach.
You need to verify, though, that it converges to the right thing. I.e., check that the final solution and objective value it gives agree with intlinprog. You also need to check that it does so 99% of the time, and not say 80%.
I do not have the knowledge for that. I don't know if BINTPROG or even INTLINPROG works better. I am interested in see documentation explaining how it works before seeing how it's used. If you can provide some link it would be really helpful.
Google's your friend,
추가 답변 (0개)
참고 항목
제품
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!