A gear rattle metric based on the wavelet multi-resolution analysis: Experimental investigation

In the article an investigation about the feasibility of a wavelet analysis for gear rattle metric in transmission gears, due to tooth impacts under unloaded conditions, is conducted. The technique adopts the discrete wavelet transform (DWT), following the Multi-resolution analysis, to decompose an experimental signal of the relative angular motion of gears into an approximation and in some detail vectors. The described procedure, previously developed by the authors, permits the qualitative evaluation of the impacts occurring between the teeth by examining in particular the detail vectors coming out from the wavelet decomposition. The technique enables discriminating between the impacts occurring on the two different sides of tooth. This situation is typical of the double-sided gear rattle produced in the automotive gear boxes. This paper considers the influence of oil lubricant, inserted between the teeth, in reducing the impacts. Analysis is performed by comparing three different lubrication conditions, and some of the classical wavelet functions adopted in literature are tested as "mother" wavelet. Moreover, comparisons with a metric based on the harmonic analysis by means of the Fast Fourier Transform (FFT), often adopted in this field, are conducted to put in evidence the advantages of the Wavelet technique with reference to the influence of some fundamental operative parameters. The experimental signals of the relative angular rotation of gear are acquired by two high resolution incremental encoders on a specific test rig for lightly loaded gears. The results of the proposed method appear optimistic also in the detection of defects that could produce little variations in the dynamic behavior of unloaded gears.