摘要
软件可靠性评估性能直接影响软件测试的工作量,本文针对软件测试工作中的故障检测和校正处理问题,提出一种基于Logistic增长神经网络的软件测试方法。该方法考虑到软件工程的多样性,利用Logistic增长曲线构建神经网络模型完成故障检测,并结合指数分布校正时间完成故障校正过程。通过两组真实失效数据集(Ohba与Wood)的试验,将所提方法与现有的软件可靠性增长模型(software reliability growth model,SRGM)进行了比较。结果显示Logistic增长神经网络模型的模型拟合效果最优,表现出了更好的软件可靠性评估性能及模型适应性。
Evaluation of software reliability performance directly affects the course of software testing.In this paper,we investigate fault detection in software testing to improve its performance.A software testing methods based on neural network of the logistic growth was proposed.Considering the diversity of software engineering,the proposed method using the logistic growth curve to construct the neural network model in order to complete fault detection.The proposed method combines the exponential distribution correction time in order to complete the fault correction process.Through the test of two sets of real failure data sets(Ohba and Wood),the proposed method is compared with the existing software reliability growth model(software reliability growth model,SRGM).The results confirm that the model fitting effect of the proposed logistic growth neural network model is optimal,demonstrating the adaption and better performance of the software reliability assessment model.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2017年第4期646-651,共6页
Journal of Harbin Engineering University
基金
国家自然科学基金项目(61063028
31560378)
江苏省自然科学基金青年基金项目(BK20150784)
中国博士后面上项目(2015M581800)
甘肃省科技支撑计划(1604WKCA011)
陇原青年创新创业人才项目(2016-47)
2016年度甘肃省高校重大软科学(战略)研究项目(2016F-10)