Comparação de representações conjuntas tempo-frequência aplicadas na análise de falhas em sistemas engrenados.

Abstract

The knowledgement of the faults development phenomenon in geared systems has been seen as an important task in the design and maintenance phases. Usually, the vibration signals originated are very representative of the physical operating system, and it can be investigated through the computational models or by real measurement of the signature vibrations in several situations or fault conditions. In general, these signals are multi-components and not stationary due to the transient gear forces that appeai-s starting from the superficial contact between the teeth with some type of imperfection (wear, fatigue crack, etc.) developed in the form of vibrations that excite several resonance frequencies of the system. The classic spectral analysis applied to the fault detection has show limitations, especially concerning the appearance of spurious components that can contaminate the spectrum, and difficult the interpretation of the analysts. To minimize these problems, this work to investigate some time-frequency representations, which can supply a new discernment about the interpretation of these signals, evidencing the physical phenomenon caused by transient vibrations and allowing to separate the components of the signal caused by different sources. The objective of this work is also to establish some parameters that allows to compare the time-frequency representations, Spectrogram and Pseudo Wigner-Ville applied for the faults analysis in gears systems. Some tests are performed with experimental signals of gears subject to three different conditions: toothless gear scratched gear and normal gear. The two methods above mentioned are applied to study these gears. It is show in this work the use of residual signals, and was verified that these our can help the analysis of the time-frequency maps. At the end of this work it is verified that the Pseudo Wigner-Ville distribution is more robust than the Spectrogram for analysis of geared system with faults.