Abstract: The main objective of the work carried out in this paper was to study the response of messing tooth wear on vibration signal and suggest a suitable signal processing technique which can help in detecting the defects easily. Experiments were conducted on meshing gear assembly fabricated for the purpose. The signals were acquired using both the accelerometer and acoustic sensors. It has been observed that the wavelet de-noising is more effective for filtering of defective gear vibration signal compared to standard FIR and IIR filter. Part of the signal representing defect becomes more prominent after wavelet de-noising. It results in the kurtosis parameter improved twice with respect to raw signal after filtering. For the vibration signal processing, the noise reduces by 42 % which justifies the selection of wavelet filtering. In the wavelet de-noising, the vibration burst was not getting distorted during filtering for vibration. The interpretation of spikes from enveloped signal with driver gear rotation was easier to understand and correlate for the cases, where vibration burst was present in vibration signal due to defect. In the case when the vibration burst is suppressed by other noise or is having low amplitude, undecimated wavelet transform implementation was more suitable. It shows that UDT also acts as a de-noising tool for gear vibration signal analysis. In vibration signal processing, the approximation signal was having higher Kurtosis value and lesser noise. For the acoustic signal processing, Kurtosis parameter improved 70 % with respect to raw signal after filtering. The approximation signal having- little higher kurtosis and SNR values than raw signal, but detail coefficients were having significant higher kurtosis value and lower SNR value.

Keywords: MATLAB, LABVIEW, Data acquisition system, Accelerometer.