Fault Size Identification of Rolling Element Bearings using Vibration Signal Analysis

Abstract

Prognostics is the estimation of a machine's remaining useful life (RUL) based on relative and future operating conditions. One of the most common causes of machine failure is rolling element bearing failure. When a machine is running and a fault develops in the bearing, determining the size of the fault is extremely without stopping the machine. As a result, before the fault size grows to a significant level, it is necessary to change bearings for machine safety. Fracture mechanics based approach can help the maintenance personnel to estimate the RUL if the initial fault size is known in advance. The present work focuses on the use of vibration signal as a medium to estimate the size of fault without stopping and disassembling the machine. Decomposition of vibration signals using discrete wavelet transforms and Singular spectrum analysis, as well as the autoregressive approach, minimal entropy deconvolution are used to estimate fault size. The decomposed signal is searched for two events, one for the ball entering the fault and the other for the ball exiting the fault. Singular spectrum analysis and Wavelet Analyze is carried out on the experimental data of various multiple size defect in outer part race and inner part race and the data available from Paderborn University. Varies sizes of defects and analyzed in Discrete Wavelet Transform in Matlab software and for Singular spectrum analysis (SSA) in R software. SSA method is done in two stages 1.Decomposition 2.Reconstruction. Waves in decompositions reveal the presence of impulses in vibration signals caused by bearing failures. It is discovered that the impulses emerge on a regular basis, with a period corresponding to the typical defect frequencies. It is discovered that the impulses emerge on a regular basis, with a period corresponding to the typical defect frequencies. The proposed technique's estimated defect size is found to be very close to the real defect size.