Tour code would consist of three stages. Of course you have to tune the controller beforehand, i.e. choose the number of membership functions, their shapes and parameters.

1. Fuzzification. You simply evaluate the membership degrees for every membersip function on every input. Something like

mu_x1_NL=exp(-x1^2)

for gauss membership function. If you are planning to use triangular functions, the code would be somewhat more complicated.

2. Inference. This is where you put your rule table to use. For the rule:

IF x1=NL AND x2=NS THEN y=PS

you would write

mu_y_PS=min(mu_x1_NL,mu_x2_NS).

If there are seeral rules, that evaluate the same output membership degree, you would basically take the largest of them using max command.

3.  Defuzzification. This would be the most complicated part, if you are using centroid defuzzification. But for simple applications singleton defuzzification provides an acceptable result. That is, the output membership functions are strictly constant, i.e. y_PS=10, y_NL=-10 etc. In such a case the total system output would be equal to weigted average of these values:

y=(mu_y_NL*y_NL+mu_y_NS*y_NS...)/(mu_y_NS+mu_y_NL+...)

These are the basic thing you may need to implement fuzzy controller on any MC. 

Hi Mohsen Shahraki,

This might be useful link.

Thank you.

http://volt5.persiangig.com/.rWOwtxbN7d/document/IEEE_06339355.pdf

Hi Mr. Senior Lecturer

thank you so much because of your useful hint

Hi Prof.Naveen Balaji G

thank you so much because of your useful paper

thank you So much Prof.Kiruthikamani Govardhanaraj

they are so helpful