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.