Detection of rolling element bearing faults via vibration analysis

Rolling element bearings are the main components in rotary machines due to their advantageous friction characteristics. It is very important to keep the rolling element bearings in good condition in terms of general machine health. In this study, the condition of deep groove rolling element bearings is monitored by means of vibration measurement. Experimental vibration signals are collected from a test rig, including two rolling element bearings, which carry a shaft having an unbalanced mass on it. A local defect on the inner or outer race of one of the rolling element bearings is introduced artificially and vibration measurements are performed using a portable vibration analyzer in terms of displacement, velocity and acceleration. Some statistical indices such as rms, peak to peak and kurtosis values of the vibration signals are calculated for healthy and faulty cases, in order to obtain the change in the time domain parameters due to bearing deterioration. The vibration measurements are performed for a broad range of shaft speed. Signs of the bearing deterioration are also investigated in frequency domain by the Fast Fourier Transform (FFT) and Short Time Fourier Transform (STFT). As the main part of this study, a nonlinear time domain transform, named as Curve Length Transform (CLT) is applied to the vibration signals for diagnostic purposes. The statistical indices of the CLT signals for healthy and faulty cases are calculated and compared, to make a decision about the condition of the rolling element bearing. The statistical indices of the CLT signals are also compared with the statistical indices calculated for raw vibration signals, in order to show the efficiency of the Curve Length Transform. The experimental results show that the CLT can be used successfully to enrich some time domain parameters, which give useful information in capturing the local bearing failures.