I am using fuzzy logic to help decision-makers to get "quickly" and reliable decisions. I have collected many projects and I am interested on 12 parameters. So Finally my system has 12 inputs, each one has 3 membership functions(MFs), and one output with 4 MFs.
But I have a problem regarding the Inference system, i.e. If-Then-Rules. My parameters have a solid Relationship between them, so I think that I Don't have to consider all possible combination ( 3^12) but I have to make sure that my system got a good training and then I can trust him.
Before talking about the testing the system with a sall dataset, let's talk about the way I can consider all possible If then rules. should I come bach to each project and according to each one I see on which membership it belongs, or I use Simply the parameters ranges that I already realised and build my If-then-rules according to that?
One more question relating to unsuccessful projects. how could I consider these information (example: when a was low and b was Med and ….. Then the project X was insuccessful, Can I use it like this:
If a is Low and b is Med and …. Then output is not X ?)
I will more than happy f you share with me your experience so I can adapt it to my problem and act accordingly.
Thank you, and I am wainting for your interactions.
King Regards,
Here are some examples of the design and optimization of FLC:
>should I come bach to each project and according to each one I see on >which membership it belongs, or I use Simply the parameters ranges that I >already realised and build my If-then-rules according to that?
If you are using MATLAB then when you view the output. It already highlights the execution of rules and values on membership functions which depend on the data value passed by you and the way your rules have executed. The design of rules and membership functions depends on the
range of your data, results obtained from existing techniques, Experts opinion...
- Badges
- Science topic