一种基于改进的K-means聚类和PSO的软件可靠性模型选择  

Model selection of software reliability based on improved K-means clustering and PSO

作　　者：黄为勇[1] 

机构地区：[1]徐州工程学院信息与电子工程学院,江苏徐州221008

出　　处：《计算机应用研究》2011年第10期3778-3781,共4页Application Research of Computers

基　　金：国家自然科学基金重点资助项目(50534050);江苏省高校自然科学研究计划资助项目(06KJD460174);徐州工程学院校科研基金资助项目(XKY2009117)

摘　　要：针对软件可靠性选择主要依靠人的主观经验进行判断、缺乏客观性和准确性的问题,提出了一种基于改进的K-means聚类和粒子群优化(PSO)算法的软件可靠性模型选择方法。该方法采用多评价标准编码,选定一种新的规则化距离作为元素间的相似性度量,应用K-means聚类和PSO分析实现了软件可靠性模型的选择。实验结果验证了该方法的有效性,为软件可靠性模型选择提供了一条新途径。Addressing at the problem that software reliability selection depends mainly on the individual＇s subjective expe-rience,and lacks objectivity and accuracy,this paper proposed a new method for model selection of software reliability based on improved K-means clustering and particle swarm optimization（PSO）.The method used standard codes with multi-evaluation,employed a new regularized distance as similarity measurement between elements,and applied K-means clustering and PSO analysis to realize model selection of software reliability.The experimental results validate the effectiveness of the proposed method,which provides a new approach to model selection of software reliability.

关 键 词：软件可靠性模型 K-MEANS聚类 粒子群优化 规则化距离 多评价标准编码 模型选择 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论] TP181[自动化与计算机技术—计算机科学与技术]