An autonomous method for diagnosing raceway defects and misalignment in a self-aligning rolling-element bearing

Abstract

This paper presents an autonomous method to diagnose a double-row self-aligning ball bearing for any dynamic misalignment or localized defects, using vibration signals. The frequency spectra from the experimental investigation revealed that irrespective of the health condition of the bearing i.e. either healthy or defective, the frequency spectrum had a clear cage frequency peak whenever the system was under dynamic-misalignment. An improved version of Harmonic Product Spectrum (HPS) called sideband product spectrum (SPS) was used to identify the cage frequency peaks which was preprocessed using kurtosis-based band-pass filter and parameter-optimized variational mode decomposition (VMD) algorithm. A mathematical model is also presented to analyze the dynamic behavior of double-row self-aligning ball bearing under such conditions. From the model, it was realized that during misalignment, the contact load shared by the rolling elements, generates some moment which has a tendency to rotate the bearing’s inner-race about the radial axis and thus resulting in a wobbly rotary motion. This wobbly motion of the inner race in fact characterizes the bearing for dynamic misalignment. The simulated results were in accordance with the experimentally obtained results.