Scholarlink

<p>This study presents an efficient surrogate-based reliability analysis framework for evaluating the structural integrity of a composite panel used in marine and naval applications under varying elevated temperatures. A new limit state function based on maximum deflection was defined, and tensile tests were conducted at multiple temperatures to capture degradation in mechanical properties. An updated finite element (FE) model, calibrated through experimental modal analysis, was developed to predict panel behavior under elevated temperatures. Using datasets generated via updated FE model and response surface methodology (RSM), artificial neural networks (ANNs) were trained to construct surrogate models. These models were then combined with Monte Carlo Simulation (MCS), First-Order Reliability Method (FORM), and Second-Order Reliability Method (SORM) to estimate failure probabilities. The results revealed a clear increase in failure probability with rising temperature, primarily due to the degradation of tensile properties. The proposed surrogate-based approach achieved high predictive accuracy while reducing computational cost by up to 80% compared with direct FE-based reliability analysis. Sensitivity analysis further indicated that the reduction of elastic modulus with temperature was the most influential parameter affecting failure probability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Efficient Reliability and Sensitivity Assessment of a Composite Panel Subjected To Elevated Temperatures Based on Surrogate Modeling

  • 期刊论文

  • 发表日期：

    2026年2月9日

查看全文
  • Mohsen Kouhi,
  • Alireza Mojtahedi

摘要

This study presents an efficient surrogate-based reliability analysis framework for evaluating the structural integrity of a composite panel used in marine and naval applications under varying elevated temperatures. A new limit state function based on maximum deflection was defined, and tensile tests were conducted at multiple temperatures to capture degradation in mechanical properties. An updated finite element (FE) model, calibrated through experimental modal analysis, was developed to predict panel behavior under elevated temperatures. Using datasets generated via updated FE model and response surface methodology (RSM), artificial neural networks (ANNs) were trained to construct surrogate models. These models were then combined with Monte Carlo Simulation (MCS), First-Order Reliability Method (FORM), and Second-Order Reliability Method (SORM) to estimate failure probabilities. The results revealed a clear increase in failure probability with rising temperature, primarily due to the degradation of tensile properties. The proposed surrogate-based approach achieved high predictive accuracy while reducing computational cost by up to 80% compared with direct FE-based reliability analysis. Sensitivity analysis further indicated that the reduction of elastic modulus with temperature was the most influential parameter affecting failure probability.