Comparison of vibration values of rotating discs with variable parameters obtained by finite element analysis modeling with different machine learning algorithms

Abstract

PurposeRotating functionally graded (FG) disks of variable thickness generates vibration. This study aims to analyze the vibration generated by the rotating disks using a finite element program and compare the results obtained with the regression methods.Design/methodology/approachTransverse vibration values of rotating FG disks with variable thickness were modeled using different regression methods. The accuracies of the obtained models are compared. In the context of comparing regression methods, multiple linear regression (MLR), extreme learning machine (ELM), artificial neural networks (ANNs) and radial basis function (RBF) were used in this study. The error graph between the observed value and the predicted value of each regression method was obtained. The error values of the regression methods used with scientific error measures were calculated.FindingsThe analysis of the transverse vibration of rotating FG disks with the finite element program is consistent with the studies in the literature. When the variables and vibration value determined on the disk are modeled with ELM, MLR, ANN and RBF regression methods, it is concluded that the most accurate model order is RBF, ANN, MLR and ELM.Originality/valueThere are studies on the vibration value of rotating discs in the literature, but there are very few studies on modeling. This study shows that ELM, MLR, ANN and RBF, which are machine learning methods, can be used in modeling the vibration value of rotating discs.