Balanceamento de máquinas rotativas com eixos flexíveis utilizando metamodelos Kriging

Abstract

The present work proposes a new approach to balancing rotating machines using techniques based on vibration signals. The proposed methodology uses Kriging surrogate models to perform the balancing of rotating machines. To create the surrogate, it is necessary to create input and output samples, which are the unbalance conditions with their associated phases and the correction mass(es), with their respective position(s) angle(s), respectively. These data are provided by an existing traditional balancing technique, which in this case was the Coefficient of Influence Method. The results obtained with the obtained surrogate model were compared with the balancing via the CI method, for validation purposes. This validation was performed numerically and experimentally for one and two balancing planes. In this way, the proposed technique proved to be efficient for the balancing of rotating machines, leading to a satisfactory reduction in the vibration amplitudes of the considered rotor. In addition to this new approach, it is presented an analysis devoted to determine the best statistical distribution for creating samples of uncertainties for balancing a rotor in the methodology proposed by Carvalho (2017).