区间模型非概率可靠性指标的一维优化算法

A ONE-DIMENSIONAL OPTIMIZATION ALGORITHM FOR NON-PROBABILISTIC RELIABILITY INDEX

  • 摘要: 证明了基于区间模型的非概率可靠性指标只可能存在于:标准化区间向量张成的凸域及其扩展空间中,通过原点和凸域顶点的有限条超射线与标准化失效面的某个交点处,从而极大地缩小了优化方法求取非概率可靠性指标的可行域,使放弃计算结果极易扩张的区间运算法则和计算量太大的高维寻优方法成为可能。提出了求取非概率可靠性指标的一维优化方法,讨论了一维优化方法搜索区间关于0对称且具有半径递减的特性,利用该特性明显提高一维优化算法的效率。

     

    Abstract: The following points are proved in this paper: (1) Non-probabilistic reliability index of an arbitrary structure merely exists at one of the crossing points at which standard failure surface of the structure intersects the straight lines passing through both origin of a standard infinite space and vertices of holohedric convex polyhedron centered at the origin, and (2) The non-probabilistic reliability index equals to the absolute value of the coordinate of one of the crossing points. Based on the reduction of feasible region from the standard infinite space to finite crossing points within the standard infinite space, a one-dimensional optimization algorithm for non-probabilistic reliability index is developed to replace the interval algorithm and multi-variable optimization approach. The algorithm is reliable and efficient in arithmetic operations.

     

/

返回文章
返回