摘要
提出了多模型加权平均的软件可靠性综合模型,通过权值的动态调整实现模型的选择与混合.选择的目的是要找出最适合待评估软件的模型,混合目的是要综合各模型的评估结果.所以,关键是要进行权值的动态调整.为此提出用贝叶斯分析理论进行权值计算的方法,将权值作为贝叶斯超参数,根据实际数据进行超参数更新,从而达到权值计算的目的.
An integrated software reliability model, which is based on weighted average, is proposed. With the weights′ dynamical adjustment, the model selection and mixing are implemented. The goal of selection is to find out the model most suitable for assessment of the software of concern. And the goal for model mixing is to take into account of the results of various models. To attain the weight values, this paper suggests the Bayesian analysis methodology. By treating the weights as Bayes super_parameters, the weight values are updated iteratively based on the actual incoming failure time data.
出处
《武汉大学学报（工学版）》
CAS
CSCD
北大核心
2003年第1期86-88,共3页
Engineering Journal of Wuhan University
关键词
软件工程
可靠性
贝叶斯分析
先验分布
software engineering
software reliability
Bayesian analysis
prior distribution
