A strong source, for the purpose of this document, has a flux higher than 10000 Jy around 100um. We assume that you have available the complete set of files pertaining to your ISO LWS observation as received by ESA (through the CD-rom or the ISO data archive (http://pma.iso.vilspa.esa.es/obscat/app_autostart.html).

The peculiar problems affecting the spectrum of a strong source generally appear for detectors LW2, LW3 and LW4. The spectra should appear curved ('saggy') due to severe non linearities in the detectors response for these strong fluxes. A brief discussion about this effect is available in the ISO Data User Manual (http://isowww.estec.esa.nl/manuals/lws_idum5/node10.html#SECTION001040000000000000000).
Take your data, make a straight average and plot detectors LW2, LW3 and LW4. Do you see something like this (figure taken from the IDUM) ?

Fig. 1
 

If yes, your spectra are saturated. There is nothing available to you to correct for this. The data have to be reprocessed through a special version of the automatic Data Reduction Pipeline available at the LWS PI Institute (Rutherford Appleton Laboratories) in UK. Contact Sergio Molinari or Steve Lord at iso@ipac.caltech.edu (US) or Sunil Sidher at S.D.Sidher@rl.ac.uk to have your data reprocessed, corrected and
delivered to you.

Once you receive the reprocessed data corrected ofr non linearities, the analysis proceeds mostly as in the case of a medium brightness source (see http://www.ipac.caltech.edu/iso/lia/lws01_medium.html). Only consider that in this strong flux regime the detectors affected by memory effects (appearing as a systematic differences between different scan directions) are not the same as in the medium flux case. In the latter case the affected detectors were mostly LW2 and LW3; for a strong source these two detectors generally behave normally (in terms of differences between scan directions) and it may be possible to see these effects clearly on detectors like SW3.