软件可靠性增长G-O模型的改进  

Improvement of Software Reliability Growth G-O Model

作　　者：顾爱华[1] 张春阳[1] 胡庆新[1] 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《内蒙古大学学报（自然科学版）》2014年第2期192-195,共4页Journal of Inner Mongolia University：Natural Science Edition

基　　金：教育部高等学校博士学科点专项科研基金,No.20120111110001;国家自然科学基金,No.61070220

摘　　要：在传统的软件可靠性增长G-O模型中,故障检测率被视为常数,或被视为增函数,或被视为减函数.考虑到测试人员的学习过程和剩余错误数的不断减少,故障检测率应该是一个随时间先增后减的函数,于是提出改进的G-O模型.利用公开发表的两组失效数据对改进模型进行了测试,并与已有模型进行了较,实验结果表明改进G-O模型的拟合能力和预测能力都很好.In the traditional software reliability growth G-O model, fault detection rate is regar- ded as a constant, or an increasing function,or a reduction function. Considering the learning process of the testers and the reductions of the residual error number, fault detection rate should be a first increased and then decreased function with time. An improved G-O model is proposed in this paper. The improved model is tested by using two public failure data sets,and is compared with other ex- isting models. The results show that the improved model＇s fitting ability and prediction ability are very good.

关 键 词：非齐次泊松过程 故障检测率 软件可靠性增长模型 G-O模型 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]