Coupled machine-learning approach and thermoelasticity for analyzing thermal stability of the nanocomposite sandwich system

Abstract

To offer actionable strategies for mitigating the negative impacts of thermal shock loading and enhancing structural design, this study emphasizes the analysis of the transient coupled thermo-elastic behavior of a sandwich cylindrical panel. This panel is subjected to thermal and heat-flux shocks and features face-layers reinforced by functionally graded graphene platelets (FG-GPLRC) alongside a polymeric core. Leveraging compatibility conditions, a three-layered sandwich structure model is developed. Ensuring calculation precision, the governing equations are formulated based on the comprehensive three-dimensional elasticity theory. A numerical approach is adopted for the analytical solution, targeting the response of both fully simple and clamped-supported setups. The Dubner and Abate method aids in inverting the Laplace transform to ascertain the system’s time evolution. In a subsequent section, the derived results are integrated into a novel machine learning algorithm, the Forest Regression for LAyer-wise Structures (FRAS), to anticipate deflection and stress within the layers. The machine learning outcomes align reasonably well with the multi-layered nature of the structure and diverse regression across layers. This suggests that, for swift preliminary insights in engineering applications, machine learning techniques might be a more efficient alternative to traditional numerical computations.