Imitation model of unbalanced rotor on fluid-film bearings

Abstract

Diagnostics of rotating machines is taking a new step forward with the development of intellectual technologies based on predictive and machine learning tools. Despite having a range of advantages compared to humans in terms of big data processing powers, correlation and various feature extraction possibilities, and swiftness of operation, such systems are limited by measurement system elements in terms of their parameters: sensors and ADCs with their sensitivity properties and accuracy restrictions, microprocessors with limitation of processing powers, etc. Experimental data is used to recreate experimental environment in simulation of induced unbalance. The imitation model is based on rotor dynamics equations of rotor motion, Reynolds equation to estimate reaction forces of a fluid-film bearing and takes into account sensor parameters and position, gear coupling effect and other measurement system elements parameters. The results show that under certain conditions it becomes impossible to successfully track unbalance which in real conditions could lead to malfunction or failure of a rotor machine.