首页 >

基于RBF增强直接概率积分法的板壳结构随机屈曲分析

王超凡,周焕林,王选   

  1. 合肥工业大学
  • 收稿日期:2025-05-09 修回日期:2025-09-12 出版日期:2025-09-18 发布日期:2025-09-18
  • 通讯作者: 王选
  • 基金资助:
    国家自然科学基金;工业装备结构分析优化与CAE软件全国重点实验室开放课题

RBF-enhanced Direct Probability Integral Method for Stochastic buckling analysis of plate and shell structures

  • Received:2025-05-09 Revised:2025-09-12 Online:2025-09-18 Published:2025-09-18
  • Contact: Xuan WANG

摘要: 板壳结构作为航空航天、船舶、建筑等工程领域的关键承载组件,其屈曲稳定性直接决定整体结构的安全性与可靠性。然而,材料属性离散性等因素显著影响屈曲临界载荷的分布特性,传统确定性分析方法难以准确量化此类随机影响。为此,本文提出一种径向基函数(Radial Basis Function, RBF)增强的直接概率积分法(Direct Probability Integration Method, DPIM),用于高效求解板壳结构在多重随机变量作用下的屈曲临界载荷概率特性,为随机屈曲不确定性量化评估提供理论依据。通过RBF构建屈曲临界载荷与随机变量之间的高精度显式代理模型,有效减少原始直接概率积分法中实施耗时的屈曲有限元分析的计算代价。数值算例中将径向基函数增强的直接概率积分法与原始直接概率积分法、蒙特卡洛模拟方法进行对比,结果表明本文提出的方法在保持精度损失可控的同时,能显著提升计算效率。

关键词: 板壳结构, 屈曲分析, 直接概率积分法, 径向基函数, 随机力学

Abstract: As critical load-bearing components in aerospace, marine, and construction engineering fields, the buckling stability of plate and shell structures directly determines the safety and reliability of entire systems. However, factors such as material property variability significantly influence the distribution characteristics of critical buckling loads, which traditional deterministic analysis methods struggle to accurately quantify. This paper proposes a Radial Basis Function (RBF)-enhanced Direct Probability Integration Method (DPIM) to efficiently determine the probabilistic characteristics of buckling critical loads in plate-shell structures under multiple random variables, establishing theoretical foundations for stochastic buckling uncertainty quantification. By constructing a high-precision explicit surrogate model between critical buckling loads and random variables using RBF, this approach effectively reduces the computational cost of implementing the time-consuming buckling finite element analysis in the original direct probability integration method. Comparative analyses with traditional DPIM and Monte Carlo simulation methods in numerical examples demonstrate that the proposed RBF-enhanced DPIM achieves remarkable computational efficiency improvements while maintaining controlled accuracy loss.

Key words: Plate and shell structures, Buckling analysis, Direct probability Integral method, Radial basis functions, Stochastic mechanics

中图分类号: