When trying to interpret the DFDR traces give a thought that the sampling rate of its data trace could be much coarser than the detection rate of the computer systems that automatically control its functions.
the way to get beyond this is to trial model some engine performance graphs and then see which ones best match the actual DFDR traces.