基于ANN响应面法和Monte Carlo法的海洋平台可靠度分析

Reliability Analysis of Offshore Platform Based on Response Surface Method of ANN and Monte Carlo Simulation

利用人工神经网络(ANN)响应面法和Monte Carlo法对波流联合作用下固定式海洋平台进行了可靠度分析。综合考虑各种随机变量对海洋平台失效的影响,通过有限元方法获得一组随机变量和结构响应数据,利用BP神经网络模型构建平台结构的隐式功能函数。基于Monte Carlo方法由MATLAB软件产生大量的随机变量组合,通过统计网络映射结果确定结构的可靠性。

Reliability analysis of fixed offshore platform subjected to wave and current is studied by response surface method of artificial neural network （ANN） and Monte Carlo simu- lation. The effects of different random variables to the invalidation of platform are synthetical-ly considered, the structural response data and random variables are obtained by finite element methods. The BP neural networks are built to the implicit performance function of platform structure. Based on the Monte Carlo simulation, MATLAB software generates lots of combinations of stochastic variables, by counting the mapping results of BP networks; the reliability of structure is obtained.