acceleration auto spectrum fitting

acceleration auto spectrum fitting

I have accelerometer auto spectrums on a system with a fan. The fan rpm is
varied. I'm trying to curve fit the spectrum. The pseudo code to sweep
through the rpm data to locate possible structural peaks
% Find the indices of all significant structural peaks all_peaks = []; for
irpm=1:nrpm [peakmag, peakindex]=findpeaks(currentRPMPSD) sortPeakMag =
sort(peakmag); possibleStructuralPeakAmp = 0.1*sortedPeakMag(2);
[strcutPeakMag, indexStrcutPeakMag]
=findpeaks(currentRPMPSD,'magnitude',possibleStructuralPeakAmp )
all_peaks = [all_peak_indices; indexStrcutPeakMag]; end
% Select structural peaks that showed up at least 3 times [peakIndex,
peakCount]=hist(all_peaks);
iPeak = peakCount >= 3;
I could use the above peaks as starting points for estimates of the
natural frequencies. But, I'm not sure how to write the curve function to
fit for 1 / rev forcing.
I'm trying
h(w) = sum_modes( modal_amp(i) /
(-w^2+2*modal_zeta(i)*modal_wn(i)*i+modal_wn(i)^2) + sum_forcing(
forcing_amp(j)
/ (-w^2+2*forcing_zeta(i)*modal_wn(j)*i+forcing_freq(j)^2)