it is suggested to write a code for this purpose.

If you prefer to use the MATLAB function, you can create a FIS as follows,

newfis:for creating a new FIS

a = newfis('mymodel','sugeno','min','max','prod')

defining first input

a = addvar(a,'input','input1',[-30 30]);

a = addmf(a,'input',1,'Negative','gaussmf',[x1 x2]);

a = addmf(a,'input',1,'Zero','gaussmf',[x3 x4]);

a = addmf(a,'input',1,'Posetive','gaussmf',[x5 x6]);

% defining output

a = addvar(a,'output','myFn',[-20 20]);

a = addmf(a,'output',1,'Vriable','constant',x10);

rulelist=[ 1 0 0 1];

a=addrule(a,rulelist);

then you can adjust the parameters, using :

[fis,err_r,stp_sz,chkfis, chk_err] = anfis([input_data output_data],fismat,trnoption,dispopt,chk_data)

But it is trained all parameters consisting the membership and fuzzy rules parameters.

If you want to adjust just the fuzzy rules, it is better to extract rule matrix and then tune them.

The alternative way is creating FIS without using the MATLAB fuzzy codes.

% (Adaptive Neuro-Fuzzy Inference System) file

x = (0:0.2:10)';

y = sin(2*x)./exp(x/5) + randn(size(x))/30;

TrainData = [x y];

NumMfs = 5;

MfType = 'gbellmf';

NumEpochs = 20;

StepSize = 0.1;

InputFismat = genfis1(TrainData, NumMfs, MfType);

OutputFismat = anfis(TrainData, InputFismat, [NumEpochs nan StepSize]);

yy = evalfis(x, OutputFismat);

plot(x, y, 'o', x, yy, 'x', x, y, 'y', x, yy, 'm');

legend('Training Data', 'ANFIS Output');

% ------- ----------- -------

[xx, yy] = meshgrid(-5:5);

input = [xx(:) yy(:)];

fismat = readfis('mam21');

out = evalfis(input, fismat);

surf(xx, yy, reshape(out, 11, 11))

title('evalfis')

plz where is file data1 at last program?