基于聚类思想的软件可靠性模型选择

软件可靠性模型应用的不一致性一直是困扰软件可靠性研究人员的主要问题。模型选择和组合策略作为主要解决方法已成为模型应用的研究重点。该文主要探讨运用聚类思想进行软件可靠性模型选择的方法。根据针对实际失效数据的可靠性模型评价准则值编码,该文采用高斯混合模型进行聚类分析,使用EM算法估计高斯混合模型的参数,贝叶斯准则进行模型选择。为验证此方法的有效性和可行性使用了多个实际项目中的失效数据进行试验。结果表明,此模型选择方法简单有效,有利于软件可靠性模型应用不一致性问题的解决。

Software Reliability Growth Model Selection and Combination: A New Approach

In this paper,the use of the statistical technique of Gaussian mixture models as a tool for Software Reliability Growth Model(SRGM)selection is investigated.The Expectation-Maximum likelihood(EM)algorithm is used to estimate parameters of the mixture model.Using the encoded vectors of the SRGMs applicability criteria for failure data,the approach can be used to select the most appropriate SRGM for the failure data.More than twenty failure data sets are used to demonstrate the approach.The result indicated that it is effective on SRGM selection and combination.