摘要
软件静态缺陷预测系统经过常规的优化设计后,泛化能力较差,容易将低缺陷率软件模块误判为高缺陷率软件模块。为此,提出微型嵌入式软件静态缺陷预测系统优化设计。在优化设计中硬件部分使用原有的硬件结构,分析软件静态缺陷预测中的问题,建立软件静态缺陷预测模型,得到的学习目标用来训练样本,利用分类函数和训练样本通过多次迭代实现缺陷预测,完成缺陷系统的优化设计。测试结果表明,在数据集和模块数相同的情况下,与经过常规优化设计的系统相比,所提系统测试结果未发生误判情况,与实际数据相符,泛化能力较强,适合应用在实际项目中。
The software static defect prediction system achieved by the conventional optimization design has poor generalization ability,and is easy to misjudge the software module with low defect rate as the one with high defect rate.Therefore, an optimization design of static defect prediction system for miniature embedded software is proposed. In the optimization design,the original hardware structure in the hardware part is still used,the problems in the software static defect prediction are analyzed,the software static defect prediction model is established,the obtained learning objectives are used to training sample,and the classification function and training samples are adopted to realize the defect prediction by multiple iterations. In this way,the optimization design of the defect prediction system is completed. The testing results show that,under the condition of the same number of data sets and modules,in comparison with the system designed by the conventional optimization,the proposed system has no misjudgment,is consistent with the actual data,and has strong generalization ability,which is suitable for the application in the actual project.
作者
白首华
胡天彤
BAI Shouhua;HU Tiantong(Zhengzhou University of Aeronautics,Zhengzhou 450046,China)
出处
《现代电子技术》
北大核心
2020年第10期97-99,共3页
Modern Electronics Technique
基金
河南省科技发展计划科技攻关项目(172102210076)
河南省高校重点科研项目(13A510391)。
关键词
微型嵌入式软件
缺陷预测
预测模型
优化设计
样本训练
系统测试
miniature embedded software
defect prediction
prediction model
optimization design
sample training
system testing
