摘要
文章针对现有软件可靠性组合模型评估方法在模型选择、评估精度和评估效率方面的不足,从软件可靠性模型特征的形式化定义出发,借助特征量化获取的特征矩阵,给出了构建软件可靠性模型基本集的统一方法。依据软件可靠性评估的特点,对传统遗传算法的适应度函数和爬山操作进行了改进,并将其用于基模型权值的求解。在此基础上,给出了基于加权组合基模型的软件可靠性评估方法。实例验证表明,该方法在全局评估精度和评估效率上均有显著提升。
To overcome deficiencies of the existing software reliability assessment approaches with combination models such as model selection,assessment accuracy and efficiency,a unified approach to build the base set of software reliability models is presented with the help of the characteristic matrix from the formal definition of characteristics of software reliability models.An improvement is made for the fitness function and climbing operation of the traditional genetic algorithm according to characteristics of software reliability assessment.It is used to solve the weights of base models.On this basis,a software reliability assessment approach based on weighted combination base models is presented.The example shows this approach has a significant improvement in the global assessment accuracy and efficiency.
作者
毕翔
吴家伟
陆阳
BI Xiang;WU Jiawei;LU Yang(School of Computer and Information,Hefei University of Technology,Hefei 230601,China;Post-doctoral Research Station,Wuhu Token Sciences Co.,Ltd.,Wuhu 241009,China)
出处
《合肥工业大学学报（自然科学版）》
CAS
北大核心
2019年第11期1498-1504,共7页
Journal of Hefei University of Technology：Natural Science
基金
国家重点研发计划专项资助项目(2016YFC0801804)
国家自然科学基金资助项目(61572167)
安徽省博士后研究人员科研活动经费资助项目(2017B144)
工业安全与应急技术安徽省重点实验室开放课题资助项目(ISET701804)
关键词
软件可靠性
组合模型
遗传算法
陷入度系数
模型基本集
software reliability
combination model
genetic algorithm
trapped index
model base set
