Structural integrity of composite laminates can be significantly affected by damage resulting from lightning strikes. Accurately quantifying the residual strength and stiffness post-lightning strike, while accounting for inevitable compound uncertainties in temperature-dependent material properties due to manufacturing irregularities, defects such as random voids and stochastic lightning current parameters, is crucial for ensuring the operational safety of key composite structural components in aircraft. Here, we introduce a Bayesian inference-driven stochastic framework that integrates finite element-based hybrid thermal–electrical–mechanical simulations for uncertainty quantification in residual mechanical properties of composite laminates, wherein the parameters are estimated based on Markov chain Monte Carlo approach along with the Gibbs sampling algorithm. The inherent disadvantages concerning over-fitting and dealing with extraordinarily high-dimensional input parameter space in traditional surrogate-based Monte Carlo simulation methods for uncertainty quantification can be averted through the current approach. To obtain adequate confidence in the presented uncertainty quantification results, the probabilistic descriptions and B-basis design allowable obtained using the current Bayesian approach are compared with full-scale Monte Carlo simulations and classical non-parametric Bootstrap method. The maximum likelihood estimation-based machine learning model is further exploited for global sensitivity analysis to assess the relative influence of various governing parameters on residual mechanical properties post-lightning strike.
扫码关注我们
求助内容:
应助结果提醒方式:
