Hello everyone,
I have an MINLP optimization problem, and I can split it into two sub problems in forms of MILP and NLP. So, I was wondering if there is any merit to do that in terms of simulation time, global optimum finding, etc.
I am using GAMS software, by the way.
Thank you very much in advance for your responses.
Hi Alireza,
You can use Benders decomposition and convert it to a master and a sub problem. For more detail you can read my recent paper (10.1109/TPWRS.2019.2911576).
Let me know if you need more help.
Thank you Ehsan for your response,
Isn't Benders Decomposition useful just for MILP problems?
If so, my optimization problem is MINLP; I wonder to know, if it would be beneficial to convert my MINLP to an MILP + an NLP? To do so, I can calculate the binary variables in the first stage (MILP) and use them in the second stage (NLP); is it beneficial in terms of simulation time and global optimum finding?
Thank you.
Exactly I did the same. I convert a MINLP to MILP + NLP. Next, I solved the NLP through the proposed method which is based on the Newton method.
If you can get rid of the NLP your computation time will reduce drastically and you can find the global optimum solution. Generally there is no guarantee for solving a MINLP optimally.
- Badges
- Science topic
- Similar topics
- Political Science
- International History and Politics
- Globalization