摘要
对于软件分层结构故障的检测,能够更好的提升软件系统的运行质量。在软件体系结构测试中对软件分层结构故障的检测,需要计算故障数据挖掘特征集,将软件故障样本数据的均值定义为软件故障特征量的均值,完成对软件分层结构故障的检测。传统方法先对故障特征进行数据分类处理,给出软件故障特征向量集,但忽略了故障特征量均值的定义,导致检测精度偏低。提出基于决策同态理论的软件体系结构测试中软件分层结构故障挖掘方法。融合于非同态信息增益惩罚因子,构建软件故障的同态区间,对区间内冗余关联软件故障数据进行关联约束,融合于深度网络学习理论获取特征输入向量,计算故障数据挖掘特征集,将软件故障样本数据的均值定义为软件故障特征量的均值,以此为依据完成对软件故障挖掘。仿真证明,所提方法挖掘精度高,可以为提升软件系统的运行质量奠定结实的基础。
A mining method for fault detection of software hierarchical structure in test of software architecture is proposed based on decision homomorphism theory. The homomorphism section of software fault is built integrated with penalty factor for information gain that is not homomorphism, and relevance constraint for fault data of redundancy rel- evance software in section is carried out. The feature input vector integrated with theory of deep network learning is acquired to calculate feature set of fault data mining, and mean value of sample data is defined as its mean value of feature quantity. Thus, the excavation of software fault is completed. Simulation shows show that the proposed method has high excavation accuracy and can promote riding quality of software system and establish stout foundation.
作者
吕莉
LV Li(Sichuan Institute of Industrial Science and Technology, School of Electronic and Information Engineering Sichuan Chengdu 610500, China)
出处
《计算机仿真》
北大核心
2017年第10期371-374,共4页
Computer Simulation
关键词
软件体系结构
分层结构
故障检测
Software architecture
Hierarchical structure
Fault detection