The thermal response to a step in heating power is shown in the first picture. Thermal equivalence networks, Foster Network, can be obtained from the thermal response curve. The impedance, a(s), of a Foster-type RC network with n stages is a sum of impedances of each stage, which is a(s)=∑Ri/(1+sRi·Ci ).
Many pieces of literature used the structure function to get the time constant spectrum of the thermal resistance, which was proposed by Szekely in 1988. The process of structure function is shown in the third picture and the literature about thermal.
But the thermal response curve also can be curve fitted by MATLAB based on the function a(s)=∑Ri/(1+sRi·Ci ), which is much easier than doing deconvolution in the structure function. The number, i, can be set as higher I want, in order to reach the accuracy I need.
So, compared to the curve fitting method, what's the advantage of the thermal structure function?