基于减函数的多层贝叶斯离散型软件可靠性验证测试方案

发布时间：2018-06-06 10:05

本文选题：软件可靠性 + 验证测试　；参考：《计算机应用研究》2017年03期

【摘要】：为有效降低软件可靠性验证测试工作量,满足高可靠软件的指标验证要求,在减函数理论及贝叶斯理论基础上,提出一种基于减函数的多层贝叶斯离散型软件可靠性验证测试方案(MBDFDSRDT)。选取面向失效概率的减函数作为第一层先验分布密度函数,然后选择第一层先验分布密度函数超参数的均匀分布函数来构造其第二层先验分布密度函数,进而给出相应的多层后验分布密度函数。选取若干组失效数据,对MBDFDSRDT方案进行实践应用,并与基于贝叶斯理论的软件可靠性验证测试方案以及基于减函数的单层贝叶斯软件可靠性验证测试方案进行对比。应用结果表明,MBDFDSRDT方案可显著地降低所需的测试工作量,特别适用于高可靠安全关键软件。
[Abstract]:In order to effectively reduce the workload of software reliability verification test and meet the requirements of high reliability software index verification, based on the subtraction function theory and Bayesian theory, This paper presents a reliability verification and testing scheme for multilayer Bayesian discrete software based on subtraction function. The reduction function oriented to failure probability is selected as the first prior distribution density function, and then the uniform distribution function of the superparameter of the first prior distribution density function is selected to construct its second prior distribution density function. Then the corresponding multilayer posterior distribution density function is given. The MBDFDSRDT scheme is applied to several groups of failure data and compared with the software reliability verification test scheme based on Bayesian theory and the single-layer Bayesian software reliability verification test scheme based on subtraction function. The application results show that the MBDFDSRDT scheme can significantly reduce the required testing workload and is especially suitable for high reliability security critical software.
【作者单位】：中国船舶工业系统工程研究院;北京航空航天大学可靠性与系统工程学院;
【分类号】：TP311.53

【参考文献】