Fault Diagnosis of Ball Bearing using Time Domain Analysis and Fast Fourier Transformation

In this study Fault diagnosis of Ball bearings is done by statistical analysis under various time domain parameters. The objective of this study is to investigate the correlation between time domain and frequency domain analysis of vibration signature to judge and find the fault in bearing. This is achieved by vibration analysis and investigating different time domain parameter like Kurtosis, Skewness, Crest factor, RMS Value. For this purpose the bearing is coupled to the motor and observation were taken at 810 rpm. Vibration of the bearing are converted in voltage signal (milivolt) using an accelerometer/piezoelectric transducer. The bearing is taken under two different conditions viz Healthy (normal bearing) and Faulty (defected outer race bearing) with the aim of fault detection. Vibration data of healthy bearing are used as a standard for the analysis of vibration spectra of faulty bearing. Vibration signals are analyzed through different operations performed in MATLAB software. The result shows that the statistical analysis through different time domain parameters and its fast Fourier transformation provides efficient representation of fault detection in rolling element bearings. So as an initial stage if we find kurtosis and skewness values it can predict a fault. And if we get higher values of time domain parameters then only it needs to go for its frequency domain analysis. In this paper we also get exact fault position for defective bearing by its frequency domain analysis