Fitting curve : difficultu to do the modelized

Hello

I use for the first time the trial version of Maltlab to test it before buy it. I thought that it will be more simple for modelize.

I collected data with some pressure recorders and I want to know the equation about the all data.

At this time , for each data recorder, I used Excel to put my data and I drew. After, I used the modelisation (poly, linear, exp, ...) and I asked the equation => It's very simple.

Because I have a result with my data recorder than I cannot have the equation with Excel, I decide to test Matlab.

I sucessed to download my data (collected directly by Excel), I successed to plot but how can I have the equation ?. I Itried to test fitting curv but impossible to understand.

Thanks

Regards,

RM

[darova]

Can you please show your curve and attach the data?

[Rémi MOREAU]

Thanks alot for your attention.

In attach, data (the data is in french format) and curve :)

Regards,

RM

[Star Strider]

Your curves look like the step response of a control system. Your best option is likely to identify it as such with the System Identification Toolbox.

[Rémi MOREAU]

Thanks a lot Star Strider.

I will try to test with System Identification Toolbox. .

Regards, RM

Answer 1

[darova]

I changed your time span

% A = xlsread('data.xls','B2:E162');
% A = flipud(A);            % flip up-down matrix
rec1 = A(:,1);
rec2 = A(:,2);
rec3 = A(:,3);
rec4 = A(:,4);
t = (1:length(rec1))/15 + 0.5;  % time span

y = rec4;
ix = ~isnan(y);                 % extract only numbers
f1 = @(a,b,x) a+b./exp(x);      % function to fit
F1 = fit(t(ix)',y(ix),f1)
plot(t,y,t,F1(t),'r')
legend('record 4')

Here is what this script produces

Also look in help for fit, you can boundaries of coefficients and start points

[Rémi MOREAU]

THANKS A LOT :).

I will try to understand (my Matlab experience is very poor).

I will contact another time on monday if I not success in what your write.

Regards and good WE.

RM