摘要
软件可靠性增长模型(SRGM)是软件可靠性工程中一项重要的研究内容.在可靠性增长模型应用的过程中,常常因为模型假设与实际软件开发和调试过程有差异,导致可靠性预测的准确性不高.至今尚没有一种能适应各种软件开发环境的SRGM.为此,某些国外文献采用遗传(GA)算法,提出了模型组合方法,以期提高SRGM的预测能力.本文采用GM DH神经网络,提出一种非线性的SRGM模型组合方法.通过对比基于GA算法的模型组合方法,实验结果表明,基于GMDH神经网络的组合方法能有效提高模型预测能力.
Software reliability growth model（ SRGM ） is a key point of software reliability engineering. However in the development process of software, the prediction ability of SRGM is not very good because of the improper reliability prediction, while no single SRGM can be trusted according to various applications. Consequently some foreign references have proposed the unified scheme of SRGM and according SRGM combining method so as to improve the prediction performance of SRGM and combined SRGM. This paper proposed a nolinear SRGM model combining method based on GMDH neuro network. The prdiction ability of GA combining method and GMDH was compared in experiment. The experiment result shows that the GMDH method could improve the predictive ability of SRGMs effectively.
出处
《小型微型计算机系统》
CSCD
北大核心
2016年第6期1164-1167,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61073163)资助
上海市引进技术的吸收与创新年度计划项目(12CH-19)资助
