A Hybrid Direct Probability Integral Assisted Machine Learning Framework for Reliability Analysis of Imperfect Advanced Sandwich Structures Under Nonlinear Thermal Loading


Sharma N., Kumar P., Aimmanee S., Singh R., YAYLACI M.

Journal of Vibration Engineering and Technologies, cilt.14, sa.8, 2026 (SCI-Expanded, Scopus)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 14 Sayı: 8
  • Basım Tarihi: 2026
  • Doi Numarası: 10.1007/s42417-026-02824-x
  • Dergi Adı: Journal of Vibration Engineering and Technologies
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Anahtar Kelimeler: Direct probability integral method, Extreme gradient boosting, Layerwise theory, Monte Carlo simulation, Nonlinear thermal loading, Reliability assessment, Surrogate modelling, Uncertain frequency response, Voronoi cell
  • Recep Tayyip Erdoğan Üniversitesi Adresli: Evet

Özet

Purpose: Reliability assessment of functionally graded (FG) sandwich structures under severe thermal environments requires efficient quantification of material uncertainties, geometric imperfections, and coupled thermomechanical effects. Methods: This study proposes a computationally efficient stochastic framework to evaluate the vibration reliability of imperfect FG porous sandwich plates subjected to nonlinear thermal loading. The methodology integrates a Direct Probability Integral (DPI) formulation with an Extreme Gradient Boosting (XGBoost) surrogate model for rapid and accurate prediction of stochastic dynamic response. Two sandwich configurations are considered: (i) graded face sheets with a homogeneous core, and (ii) a graded core with homogeneous face sheets, enabling systematic evaluation of material gradation and structural imperfections. A higher-order layerwise finite element formulation is used to capture through-thickness kinematics and temperature-dependent material behavior under nonlinear thermal gradients. Uncertainties in material properties and thermal fields are incorporated within the DPI-based probabilistic framework. Results: Compared to conventional Monte Carlo simulation, the proposed DPI–XGBoost approach achieves similar accuracy with significantly reduced computational cost. Results indicate that nonlinear thermal loading decreases mean natural frequency and increases failure probability. Reliability is strongly affected by porosity distribution, gradation index, and boundary conditions, demonstrating the effectiveness of the proposed framework. Conclusion: The study demonstrates the effectiveness of the proposed DPI–XGBoost framework for vibration reliability assessment of imperfect FG porous sandwich plates under nonlinear thermal loading.