There are many systems that monitor and analyze machinery conditions, but there is a lack on monitoring techniques that are able to identify early faults. This is critical since faulty roller bearings cause most of the problems in rotating machinery. Two major difficulties limit the early prediction of roller bearing failures: The acquisition of high frequency data, and the low energy levels emitted by surfaces cracks on rollers or tracks. Early fault detection requires adequate instrumentation and signal analysis. Therefore, it is important to identify its nonlinear response, and to be able to detect low energy signals in order to detect early faults. The vibration characteristics of the bearing depend on the rotational speed, clearance, radial load, rolling element stiffness, and the surface waviness. Additionally, they generate transient vibrations due to stiffness nonlinearities and structural defects. We present in this work a technique for the detection of early faults. The technique is based on the derivation of the rolling bearing nonlinear stiffness as a function of the roller translation and roller deformation. For the deformation, we determine the relative displacement of the roller tracks and we apply Hertz’s contact stress function. With this formulation we identify the nonlinear response to the external excitation forces. Then, the results are analyzed with a time-frequency map and the phase diagram; the two analyses are compared in order to determine the best signal procedure to identify early failures.

This content is only available via PDF.
You do not currently have access to this content.