Analysis of gear failure by refined spectrum analysis

Due to the complexity of gear fault symptoms and signals, it is necessary to extract clear fault feature information on the premise of eliminating noise interference and improving signal-to-noise ratio as much as possible. There are several common methods.

The refined spectrum analysis method is to increase the frequency resolution of some frequency bands in the spectrum by using the frequency refined technology, which is called “local frequency expansion” method.

In the gear fault signal, the edge frequency obtained after modulation contains rich fault information, but it is often unable to find clear and specific edge frequency on the general spectrum diagram. The reason is that the frequency resolution of the spectrum diagram is too low. The frequency resolution on the spectrum chart is determined by the spectrum line and the highest analysis frequency, which is regulated in the industry. The specific relationship is as follows:

∆ f=fc/n= fs/N

Where, Δ f ~ frequency interval, i.e. frequency resolution;

FC ~ analysis frequency range, i.e. the highest analysis frequency;

FS ~ sampling frequency, in order to avoid frequency confusion, FS = (2.56 ~ 4) FC, generally FS = 2.56 FC;

N ~ the number of spectral lines, which is a fixed value, is divided into four grades: 100 lines, 200 lines, 400 lines and 800 lines;

N ~ sampling points, n = 2.56n, divided into 256 points, 512 points, 1024 points and 2048 points.

Because of the high meshing frequency and harmonic frequency of the gear, the analysis frequency range FC has to be very high, which results in a large frequency interval ∆ F, that is, the frequency resolution is very low, so it is difficult to show and distinguish the side frequency. The refined spectrum analysis method is just to enlarge some frequency bands along the frequency axis, just like a magnifying glass, to enlarge some local areas of interest on the spectrum map, so as to obtain the refined spectrum with high frequency resolution, as shown in the right figure. In this way, we can find the fault feature information by observing the refined sideband structure.