I have a large number of data files containing values for acoustic and
vibration data from machinery. I am trying to look for distinguishing
features in particular groups of files (covering certain dates) but
there is a lot of random noise contained within the signals so this is
proving difficult. I am plotting time domain and f domain of the
individual files and the f domain does show up some peaks. However if
I try to average the FFT (which I believe is a valid signal processing
technique for a set of similar signals), the peaks just get lost in
the noise. Therefore, I'm looking for a way to remove the noise across
all the frequencies I'm measuring (0-25kHz). I'm a beginner at this,
so if it sounds like I'm on completely the wrong track, feel free to
set me straight!
