Machine learning aided uncertainty analysis on nonlinear vibration of cracked FG-GNPRC dielectric beam

Functionally graded graphene nanoplatelet reinforced composite (FG-GNPRC) have shown significant potential for the development of high-performance and multifunctional materials and structures. Conducting an accurate analysis of the nonlinear vibration of FG-GNPRC dielectric beams can accelerate their application in various engineering fields. To investigate the nonlinear vibration of FG-GNPRC dielectric beams, the effective material properties of a cracked composite beams are evaluated using effective medium theory (EMT) and the rule of mixture. Three ML models adopted in this work, i.e., artificial neural network (ANN), random forest (RF), and AutoGluon (AG), are used to capture the complex relationship between the systematic inputs (i.e., parameters for materials and structure, the attributes of electrical field, etc.) and frequency ratio of the cracked FG-GNPRC beams. Moreover, shapley additive explanations (SHAP) is employed for the analysis of the nonlinear vibration of FG-GNPRC dielectric beams considering the influence of multi-factor coupling and uncertainties. The SHAP analysis suggests that the uncertainty of structural parameters has the greatest impact on the nonlinear vibration of FG-GNPRC dielectric beams, particularly when the structure is subjected to a voltage of 40 V.