Machine Learning-Assisted Heat and Mass Transfer Analysis in Orthotropic Piezoelectric Thermoelastic Half-Space Under Elliptic Motion Using Green–Naghdi III and Three-Phase-Lag Models

Abstract

This study develops a computational framework for Rayleigh-type surface wave propagation in an orthotropic piezoelectric thermoelastic half-space, emphasising thermoelastic coupling, heat transfer effects, and the influence of electrical and thermal boundary conditions. Generalised thermoelastic models based on the Green–Naghdi type III (GN-III) and three-phase-lag (TPL) theories are employed to overcome the limitations of Fourier heat conduction, and secular relations for the phase velocity, attenuation, and energy dissipation are derived. Analytical simulations are carried out in MATLAB, while machine-learning surrogates in Python enable fast prediction and boundary classification. A regression-based model accurately predicts wave characteristics with reduced computation time, and a confusion-matrix classifier effectively identifies boundary states. Results show that phase velocity increases with propagation angle and saturates with wave number, while attenuation and specific loss are strongly boundary-dependent. The work is limited to homogeneous half-spaces with idealised interfaces, suggesting scope for multiscale modelling and experimental validation. The integrated GN-III/TPL and ML-assisted framework offers a powerful tool for thermoelastic wave analysis with potential applications in surface acoustic wave sensing, non-destructive testing, aerospace structures, and energy harvesting systems.