GA population too homogenous

3 views (last 30 days)
Mathias Tholey
Mathias Tholey on 23 Jul 2015
Dear community,
I try to optimise the dimensioning of a heat supply system consisting of a combined heat and power unit, a boiler and a heat storage tank.
Linprog and fmincom get to the same results, however, when I run the problem with a GA, it comes up with a completely different, but worse solution. I have found out, that this is because the initial population is not heterogenous enough meaning that the individuals are too similar to each other. I provided the GA with the optimised data of linprog as an initial solution. And starting from there it gets to the same results. However without this initial solution in ends up in a local minimum which has a worse value than the results from linprog or fmincon.
I use lb, ub as well as linear constraits. So actually the default function @gacreationlinearfeasible should create a diverse population as it does in this example (Scroll down). There you can see that it chooses its initial population near the boundaries of ub,lb and the linear constraints. However in my GA it does not do the same. It is literally stuck at one side of the solution space and this is my question now. How can I make the GA come up with more diverse solutions? There must be a mistake somewhere in my code, which I cannot find. Maybe there is something wrong with the formulations of my constraints...I have NO IDEA.
I have uploaded all the files you need in order to run the code. Please run the file GA_OptimisationScript within the given folder which you can download here .
Thank you so much for your help.
Best wishes,
Mathias

Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!