基于贝叶斯和Bootstrap方法的传感器网络节点可靠性评估  

Research of sensor networks nodes reliability evaluati on based on Bayesian and Bootstrap methods

作　　者：何永强[1] 董金超[2] 朱子牮 

机构地区：[1]河南工程学院计算机学院,河南郑州451191 [2]河南工程学院理学院,河南郑州451191 [3]中原工学院软件学院,河南郑州450007

出　　处：《河南工程学院学报（自然科学版）》2016年第4期53-56,共4页Journal of Henan University of Engineering：Natural Science Edition

基　　金：河南省科技计划课题(152102210027);河南省高等学校重点科研项目(15A520054)

摘　　要：针对传感器节点故障样本数据较少和传统可靠性评估方法无法有效进行评估的问题,提出了基于贝叶斯理论在小样本数据下采用Bootstrap方法依据原生样本平均无故障间隔时间进行多次抽样生成再生样本的分布,选取Weibull分布为验前分布,推导确定可靠性的验后分布.通过仿真分析,验证了小样本数据下传感器节点可靠性评估结果的准确性,为小样本数据下的可靠性评估提供了新的思路和方法.In order to solve the problem of little fault date of sensor networks nodes,and that traditional reliability evaluation method cannot be used to assess. A method on Bayesian of sensor networks nodes was proposed under the conditions of small sample. Distribution of regeneration sample is obtained based on mean time between failure of native sample by conducting Bootstrap sampling repeatedly. Weibull distribution is chosen as the prior distribution,calculation model were deduced to obtain reliability posteriori distribution. Through simulation analysis,the accuracy of sensor nodes reliability evaluation results under small sample data is verified,providing news ideas and methods for reliability evaluation results under small sample data.

关 键 词：贝叶斯 BOOTSTRAP 传感器节点 可靠性评估 小样本数据 WEIBULL分布 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]