Comparison of Wavelet Coefficients for Condition Monitoring of Ball Bearings using Kolmogorov-Smirnov (KS) Test

摘要

The paper presents application of Discrete Wavelet Analysis of the vibration signals to study the effect of damage in ball bearing- Signal to Noise ratio (SNR) and Retained Signal Energy (RSE) of wavelet coefficients of simulated signal is determined using Daubechies wavelet for four thresholding rules. viz., Rigrsure (SURE), Sqtwolog(S), Heusure (H) and Minimax(M). SNR and RSE values obtained using simulated signals is used as a tool to select Rigrsure as the thresholding rule and Db8 wavelet as the mother wavelet. This paper explores the possibility of applying the Kolmogorov-Smirnov test (KS test) for comparing the Detail coefficients of the bearing signatures obtained using bearings having different defect sizes. Defects are created artificially on the raceways using Electric-Discharge Machine (EDM). Experiments were performed on single row deep groove ball bearing. The feasibility of using this technique is checked by comparing the outcome of the KS-test method with that of the other statistical methods, ie rms, crest factor and kurtosis. The method hus been tested successfully. Hence it can be used for fault detection and to check the progress of the defect in bearings.