Fuzzy Logic:Centroid Defuzzification

rick1138

Does anyone know how to calculate the centroid of a fuzzy rule set,I am looking for specific formulae for both triangular and Gaussian shaped rules.I've found a lot of explanations on what is being done but no clear explanations on exactly how to do it.Any information would be appreciated.

ghost_user_name

<img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" /> ummm....



::whooosh!::

(sound of talk flying way over my head)

sinewave

Yeah I am glad I am not the only one heh.. BTW like the sig starfeelt

rick1138

It's not as obscure as it may sound,fuzzy logic deals with sets of which the members may belong to partially,wheras in classical logic members either do or do not belong to a set.In fuzzy logic a grey object can be described as belonging to a certain degree to the set white and a certain degree to the set black.For example,a neutral grey can be defined as belonging 50% to the set black and 50% to the set white.As the grey becomes darker it belongs to a greater degree to the black set and a lesser to the white.The degree to which it belongs to the black and the degree to which it belongs to the white can be represented graphically by a triangle.A Gaussian,or bell-shaped, curve can also be used,but triangles are more common because the calculations are faster to compute.The centroid is just the center of mass of a fuzzy rule,such as the degree to which a certain grey belongs to both black or white,and when a fuzzy set is defuzzified the centroids of all the rules are added together,the problem is that I don't know specifically how to do this,all the literature on the subject I've found tonight is either obscure or self-contradictory.