Rule Extraction

[Unknown]

Hi,

I recently finished my masters project which involved controlling traffic lights using fuzzy logic and wireless sensor networks. I used qtFuzzylite to create a mini application that gets an output variable (PD – Priority Degree) from two inputs (TQ-Traffic Quantity and WT – Waiting time). I then imported it into my java application which was where i created my system’s interface.

My question is, how does qtFuzzylite extract rules? This is because, when I created the variables, qtFuzzylite automatically generated a set of rules (9 rules to be specific) based on my linguistic terms (Low, Medium, High) for each input variable. I was asked this question during my defense and i was unable to explain it. I would really appreciate a response from you. Thanks.

Regards Roxanne Hawi

[Juan Rada-Vilela (admin)]

Hi Roxanne,

thank you for your post and for using fuzzylite.

The rules automatically generated in QtFuzzyLite utilise the Operation::increment() method. It finds all the possible combinations of rules based on your linguistic variables. Given a vector x=<0,0,0> and upper limits u=<3,3,3>, the algorithm basically increments the vector x by the unit. The index in the vector refers to the proposition index in the antecedent, and the value refers to the index of the term in the linguistic variable.

Hope this answers your question.

Cheers.

[Unknown]

Hi,

Thank you so much for the reply. It has shed some light. However, is it possible for you to share the increment() method. I would highly appreciated it.

Thanks in advance.

Regards Roxanne Hawi

[Juan Rada-Vilela (admin)]

Hi Roxanne.

fuzzylite is free and open-source. You can find the method here:

https://github.com/fuzzylite/fuzzylite/blob/master/fuzzylite/fl/Operation.h

Cheers.

[Unknown]

Thank you so much for all the help and the link to the code. Now I can answer the questions my supervisor had asked.

[Author]

Posts