Flexible Simulator for the Vibration Analysis of Rolling Element Bearings
Keywords: Rolling element bearings, Early fault detection, Vibration measurements, Spectral analysis, Simulation model
AbstractThe detection of incipient faults as early as possible has great economic value in the monitoring of the rolling element bearings in industrial applications. In the early stage, a local fault in the bearing element produces a series of weak impacts at a rate dependent on the bearing geometry. These impacts in turn excite a special type of vibration that is, in principle, possible to detect using various condition monitoring methods. However, the analysis of vibration measurements taken from a real industrial environment can be challenging because measurements are noisy and periodic phenomena from other external, known and unknown sources may overlap the known behavior of interest, for example, in the spectral representation. In this paper, we present a flexible simulator for advancing early bearing fault detection. In the simulator, the vibration signals are generated by parametric models of the impulse responses for different vibrating components, with adjustable noise and jitter effects included. The possibility to adjust models and parameters during the simulation allows a more realistic exploration of changes, for example, in different operating conditions. This includes the generation of non-stationary components into the vibration signals, which are not suitable per se for frequency domain methods. It is possible to load measured vibration signals to verify the simulation model. The main purpose of the versatile simulation environment is to enable the rigorous testing of analysis tools consisting of digital filters, frequency domain methods (FFT, HFRT), time-frequency methods (STFT, WT) etc. In turn, improved knowledge on the behavior of analysis methods and approaches can be reflected back to the simulation when searching the limits of the early fault detection methods.
Number of References cited: 20