DIAGNOSIS OF BEARING FAULTS IN A CENTRIFUGAL COMPRESSOR USING STOCHASTIC PROCESSES: GAMMA AND LÉVY

Abstract

Predicting the failures of certain industrial systems has become essential for improving reliability. This prediction is based mainly on the analysis of the evolution of the level of degradation of the system. For systems whose state of deterioration is not directly observable, when the latter is subjected, during its use, to measurable degradation phenomena such as wear, vibration, temperature rise

We propose an approach based on stochastic processes which represent a mathematical structure for modeling, mainly models of continuous degradation and more particularly the Gamma process and the Levy process.

In this article, we present the monitoring of sliding bearing degradation at the level of a compressor, the purpose of which is to evaluate the operating time, as well as the evolution over time of the change, the physical quantities present in this study are vibration and bearing temperature.

The measurements forming the sample of the temperatures and vibrations recorded on a sliding bearing are tested by using an Easy-fit 5.4 statistical software, the overall idea of which is to compare the distribution of the data collected (measurements carried out) , with respect to the theoretical distribution function. Followed by simulation of the evolution of the physical quantities studied have been proposed.

The simulation results of the evolution of the vibration confirm that the latter properly follows a Gamma process, while the temperature variation is described by a Levy process.