Complex values using nlinfit function

Hi All

I am a PhD Student using Matlab2017a and I would like to ask you a question. Having a real model, with real data I am trying to fit them to an superellipoid. Making a previous study, I ended up with a model of 18 Coefficients. As everything is real, it is supposed to achieve real Coefficients when I am doing the fitting, but the actual outcome is the following;

My goal is to basically try to use instead of "nlinfit" to fit, to use "fitnlm". As by definition in the outcome, I have complex values for the Coefficients, is not possible to achieve the statistics associated to the Coefficients (SE or PValues per Coeff) using "fitnlm", despite the imaginary part of them its really small. Do you know any possibility of get rid of the imaginary part of the values (its really small) to be able to calculate the rest of the statistics of the coefficients using fitnlm?

Thank you in advance, I have added my script that runs everything

Jesús