摘要
针对传统复杂水工结构可靠度分析方法的局限性,提出了结构可靠度分析的高斯过程重要抽样方法。通过高斯过程机器学习方法实现结构功能函数及其偏导数的显式表达,进而结合传统重要抽样法计算结构的失效概率与可靠指标。实例验证了方法的可行性,与传统方法相比较,高斯过程重要抽样方法具有计算精度高、计算效率高和参数自适应化等优点。
For dealing with the limitations of conventional methods on analyzing hydraulic structural reliability,a new method combining the Gaussian Process(GP) and importance sampling method(ISM) was proposed.The implicit performance and its derivatives in structure stability analysis were approximated by GP model with explicit formulation,and importance sampling method was applied to get failure probability and reliability index of structure.The case studies show that the method is feasible and has advantages of higher accuracy,higher efficiency and adaptive parameter determination.
出处
《水力发电》
北大核心
2010年第12期31-33,50,共4页
Water Power
基金
国家自然科学基金资助项目(50809017)
中国博士后科学基金(20080440812)
广西自然科学基金(0832023)
