None linear least square fit for complex model

Hi,

I am trying to fit my data in log-log space. my model is 10 log (P(f))=10 log (P_0) + 10 log [1+(f/F_0)^k]- 10k log (f). the predector is the frequency (f), and the response is the estimated spectral density P(f). unknown coeficients are P_0, F_0 and K.

I have tried different matlab functions, but unfortunately there is no one able to fit my model. Please any help.?

Thank you in advance

Sat

Answer 1

[Stephan]

Hi,

have a read here:

https://de.mathworks.com/help/releases/R2015a/curvefit/fit.html?searchHighlight=fit#bto2vuv-11

There is an example how to fit a custom equation. This should solve your problem. If you do not have Curve Fitting Toolbox, you can do it very similar with lsqcurvefit from the Optimization Toolbox.

Best regards

Stephan

[Sattar Isawi]

Hi Stephan,

Thanks for your help, actually I tried this function, but unfortunatelly it doesnt solve my problem. Thanks again

[Stephan]

Maybe if you provide what you have tried so far and attach your data, someone can help.