Journal of Systems Engineering and Electronics ›› 2009, Vol. 31 ›› Issue (8): 2024-2028.

• 可靠性 • 上一篇    下一篇

一种新的软件可靠性增长模型

吴彩华1,2, 朱小冬2, 刘俊涛3, 王毅刚2   

  1. 1. 空军雷达学院信息对抗系信息作战指挥教研室, 湖北, 武汉, 430019;
    2. 军械工程学院装备指挥与管理系维修工程研究所, 河北, 石家庄, 050003;
    3. 军械工程学院计算机工程系软件工程教研室, 河北, 石家庄, 050003
  • 收稿日期:2008-04-28 修回日期:2009-03-05 出版日期:2009-08-20 发布日期:2010-01-03
  • 作者简介:吴彩华(1980- ),女,博士研究生,主要研究方向为软件保障、软件测试、软件可靠性评估.E-mail:wucaihua_1999@yahoo.com.cn
  • 基金资助:
    “十一五”国防预先研究基金(513270104)资助课题

New software reliability growth model

WU Cai-hua1,2, ZHU Xiao-dong2, LIU Jun-tao3, WANG Yi-gang2   

  1. 1. Section of Information Combat Command, Dept. of Information Countermeasure,Radar Academy of Air Force, Wuhan 430019, China;
    2. Maintenance Engineering Inst., Dept. of Equipment Command and Management,Mechanism Engineering Coll., Shijiazhuang 050003, China;
    3. Section of Software Engineering, Dept. of Computer Engineering, Mechanism Engineering Coll., Shijiazhuang 050003, China
  • Received:2008-04-28 Revised:2009-03-05 Online:2009-08-20 Published:2010-01-03

摘要: 在软件测试阶段,由于加速测试的影响,用软件可靠性增长模型测得的运行阶段可靠性很难真实反映实际运行时的可靠性.因此,修正了软件测试环境与运行环境相同的假设,针对测试阶段排错过程的延迟性,提出在测试阶段要把软件检错过程和排错过程结合起来建模.通过采用不同的故障检错率函数减少了测试与运行环境的差别,并结合移动点技术,提出了一个新的考虑测试环境和运行环境不同的可靠性增长模型(TDO-SRGM).在两个公开发表的数据集上进行的拟合试验和预测试验证明,该模型具有很好的拟合效果和预测能力.

Abstract: In the test phase,the reliability obtained from software reliability growth model(SRGM) can not represent the real one in the operation phase because of the effect of accelerated testing.Therefore,the hypothesis that the test environment is same as the operation environment is corrected.Aiming at the delay of the correction process in the test phase,a model integrating both detection process and correction process is presented.By using different rates of fault detection,the difference between test and operation environments is reduced.Integrating the change-point technique,a new SRGM model with considering of the difference between test and operation environments(TDO-SRGM) is presented.The experiments on two open data sets prove that the proposed model has good fitting and prediction ability.

中图分类号: