A quick comment from my perspective. If we are looking at machinery vibration we usually look at spectra. Occasionally we refer to a "raised noise floor". How high is the noise floor on a spectra depends on the bin width.... so not too easy to quantify from a normal spectra. PSD would be more helpful for quantifying broadband energy typically referred to as "raised noise floor" since it is energy per frequency bandwidht.... therefore independent of binwidth selected.
=====================================
Eng-tips forums: The best place on the web for engineering discussions.
The nomenclature "Power" Spectral Density is a misnomer. It really has nothing to do with power. A better description is Acceleration Spectral Density.
That is if it is accelerations you are after. It could just as well be velocities, displacements or even Force for that matter. Yes, it is important not get confused by "Power".