VIBRATION CONTROL OF PUMPING UNITS USING ARTIFICIAL INTELLIGENCE AND SENSOR TECHNOLOGIES

Abstract

The paper presents the issue of effectiveness of diagnosing imperfections of high performance industrial pumps. At present, the most commonly applied diagnostic method for such machines is based on an analysis of the rotor vibration spectrum. Measurements of vibrations over time are not often performed on a continuous basis, they are rather carried out periodically, according to a defined inspection schedule. The reliability of this diagnostic method is based on an assumption that damage to a rotor does not occur rapidly, but it is a process stretched over a longer period of time resulting in a change of rotor movement dynamics. The most frequently occurring type of pump rotor damage is a fatigue crack of a shaft, the propagation of which actually proceeds in a finite period of time. However, there is also damage caused by a rapid pump load increase leading to stresses in rotor or disc sections, which exceed permissible values resulting from mechanical properties of material used to manufacture the rotor. Some imperfections are not related to rotor damage but result from the nature of cooperation between the pump and a pipeline connected thereto.