软件可靠性多模型综合评估

Multi-model Assessment of Software Reliability

软件可靠性工程领域一直存在模型很难应用于实际的问题 .不同的软件可靠性模型对同一个软件工程项目的估计差别极大 .为此 ,探讨了软件可靠性估计的稳健方法 ,研究了多模型的贝叶斯自动混合与选择的理论 ,其重点是要探讨贝叶斯先验表达及多模型的混合机制 ,以解决上述难题 .

There has existed for a long time the difficulty in applying reliability models in practice of reliability engineering of software.Which model should we trust among about 100 available reliability models with different predictions for a single piece of software under consideration?With the estimates of a model,how much confidence do we have in believing that the model predicts the truth?Indeed,software practitioners are in a dilemma in making decisions.In this paper,an effort is made in trying to provide a solution by exploring the robust inference methodology,and by investigating into the machinery for model mixing and selection,and the Bayes prior structure.