Damped Vibration Responses of Functionally Graded Rotating Discs With Variable Geometry and Modeling With Deep Neural Networks

Abstract

PurposeThe discs used in many fields of engineering can have dangerous consequences in terms of vibration value and temporal wear. This research aims to model the lateral vibrations of disks with different boundary conditions and Forward (FW) and Backward (BW) frequencies using a deep neural network (DNN).MethodsThe transverse vibration analysis of rotating (Functionally graded) FG discs with different geometries was carried out using finite element software. The material properties of FG discs were calculated by the combination of the mixing rule and power law function obtained from the mixture of metallic and ceramic components. The material properties of FG discs are calculated by the combination of the rule of mixtures and power law function obtained from the mixture of metallic and ceramic components. This approach provides an important reference for material design in engineering applications. Modelling is provided by DNN tool with the determined hyperparameters.ResultsThe vibration responses of annular discs are obtained for Free-free (F-F), (Clumped-free) C-F and Clumped-Clumped (C-C) boundary conditions on the inner and outer surfaces and Campbell plots are drawn to show the effect of velocities on frequencies. The velocity dependent frequencies of FG discs are obtained for FW and BW waves. Different modelling results are obtained with DNN as the forward and backward wave boundary conditions.ConclusionIt is shown that the damped vibration responses of functionally graded rotating discs with variable geometry can be modelled in the range of hyperparameters determined by DNN.