• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>4th International Conference on Structural Dynamics Vol.1, Sep 2-5, 2002, Munich, Germany >Improvement of Outcrossing Rates by Importance Sampling
【24h】

Improvement of Outcrossing Rates by Importance Sampling

机译：通过重要抽样提高异型率

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Time-dependent reliabilities are usually evaluated using the outcrossing approach. The required multidimensional integrations over probability spaces and time are performed using approximation methods (first and second order). These approximate solutions can be corrected by simulation and the paper reviews the methods available for two different types of stochastic processes: rectangular wave renewal processes and differentiable normal process. The simulation method using the axis-parallel technique and importance sampling is the most efficient in each case. The presented methods are tested on several examples, dealing with both stationary and non-stationary cases. Another solution consists in updating an existing solution using a correction factor computed with simulation techniques. This method seems to be more efficient that classical importance sampling, but can suffer from numerical problems in higher dimensions. It has been successfully tested at several examples, dealing with both stationary and non-stationary cases. In all cases, the numerical effort implied is considerable. Therefore, starting from first-order solutions is preferable. However, the corresponding correction is practically not that important, at least not for high reliability problems.
机译：与时间有关的可靠性通常使用异形交叠法进行评估。使用近似方法（一阶和二阶）对概率空间和时间进行所需的多维积分。这些近似解可以通过仿真进行校正，并且本文回顾了可用于两种不同类型的随机过程的方法：矩形波更新过程和可微正态过程。在每种情况下，使用轴平行技术和重要性采样的模拟方法是最有效的。所介绍的方法在涉及平稳和非平稳情况的几个示例上进行了测试。另一解决方案在于使用通过仿真技术计算出的校正因子来更新现有解决方案。这种方法似乎比传统的重要性抽样方法更有效，但可能会遇到更高维度的数值问题。它已在固定和非固定情况下的多个示例中成功进行了测试。在所有情况下，暗示的数字努力都是相当大的。因此，从一阶解开始是优选的。然而，实际上，至少对于高可靠性问题而言，相应的校正并不那么重要。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号