基于PSO的结构可靠度及随机变量敏感性分析

Structural reliability and sensitivity analysis of random variables based on PSO

从Hasofer-Lind可靠指标的几何意义入手,建立可靠指标的优化模型,提出采用计算效率高、并能获得全局最优解的粒子群优化算法(PSO)对可靠指标优化模型进行求解,并根据优化收敛过程中随机变量候选解的统计特性,提出了可靠度计算中随机变量敏感性分析的新方法.结果表明,采用PSO求解可靠指标优化模型,方法简单,无需求解偏导数且无需给出初始迭代点,提高了计算效率,对多变量极限状态方程复杂的情况求解可靠指标尤为适用,在优化过程中同时得到随机变量敏感性分析结果.

On the basis of Hasofer-Lind structural reliability index, an optimum model for structural reliability analysis under arbitrary random variables was established and PSO with the excellent efficiency and global optimization was offered to settle the optimum model. Furthermore, a new method of sensitivity analysis was proposed based on the statistic characteristic of the particles of random variables in the convergence course. The results indicate that PSO is an efficient zero order algorithm without the partial derivatives and the initial design point for settling the optimum model of structural reliability, and the sensitivity analysis of random variables can be obtained from course of optimization.