...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Quality Control and Applied Statistics >Importance sampling for reliability evaluation with stochastic simulation models
【24h】

Importance sampling for reliability evaluation with stochastic simulation models

机译：随机仿真模型可靠性评估的重要性抽样

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

With the advent of high power computing systems, simulation studies to evaluate large-scale, complex system have become common. This article studies reliability assessment using simulation studies when estimation of reliability is challenging when the system is complex. It extends the theory of importance sampling (IS) aimed at variance reduction to compute reliability using simulations. The underlying concept is to control the sampling of input variables. Various IS methods have been proposed for the cases where a simulation output is deterministic at a fixed input in which the randomness resides only in input variables. These simulation models are called deterministic simulation models. Other simulation models, called stochastic simulation models, have one or more stochastic elements inside simulations. A unique input generates different outputs in multiple simulation replications. Stochastic simulation models can represent complex stochastic systems more realistically. However, the conventional IS method used in deterministic simulation models is not applicable to the stochastic simulation model. The study develops a new approach, called stochastic importance sampling (SIS), which efficiently uses stochastic simulations with unknown output distribution. Two methods to estimate a failure probability are proposed: using a failure probability estimator that allows multiple simulation replications at each input and an estimator that allows one simulation replication at each sampled input and derives the optimal IS density. Both methods use variance decomposition to account for different sources of output variability and functional minimization. (32 refs.)
机译：随着高功率计算系统的出现，仿真研究可以评估大规模，复杂的系统变得普遍。本文研究了当系统复杂时可靠性估算时使用仿真研究的可靠性评估。它扩展了重要性采样理论（AS），旨在减少方差以计算使用模拟来计算可靠性。潜在的概念是控制输入变量的采样。已经提出了针对模拟输出在固定输入确定的情况下的情况下提出了方法，其中随机性仅驻留在输入变量中。这些仿真模型称为确定性模拟模型。其他仿真模型称为随机仿真模型，具有内部模拟中的一个或多个随机元素。唯一输入在多种模拟复制中产生不同的输出。随机仿真模型可以更现实地代表复杂的随机系统。然而，传统的是在确定性模拟模型中使用的方法不适用于随机仿真模型。该研究开发了一种新的方法，称为随机重要性采样（SIS），其有效地使用具有未知输出分布的随机仿真。提出了两种来估计失败概率的方法：使用失败概率估计器，该估计器允许在每个输入处的多个模拟复制和允许在每个采样的输入处进行一个模拟复制的估计器，并导出最佳是密度。两种方法都使用方差分解来解释不同的输出变异性和功能最小化。 （32 refs。）

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