基于蚁群算法的软件可靠性模型参数估计方法

由于软件可靠性模型大多是非线性模型，导致其参数难于估计。总结了常用的软件可靠性模型的参数估计方法，提出一种基于蚁群算法的可靠性模型参数估计方法。通过对Musa软件可靠性模型分类方案中三个不同类型模型（G-O模型、Weibull模型以及M-O模型）的实验，发现本算法对不同模型具有很好的适应性，解决了应用传统数值计算方法时的无法收敛问题；与粒子群算法相比，本算法的收敛速度比粒子群算法快一倍以上，且对于部分实验对象的拟合结果精度比粒子群算法高一个数量级以上。

Estimating parameters of software reliability models by ant colony algorithm

It is difficult to estimate the parameters of software reliability models,since most of them are non-linear models.The most widely used methods for parameters estimating of software reliability models have been summarized,and a new approach based on ant colony algorithm was proposed.The experiments with three typical models,G-O model,Weibull model and M-O model,show that this algorithm demonstrates good applicability.And the results demonstrate that the proposed method has solved the nonconvergent problem that resulted from traditional methods.Compared with Particle Swarm Optimization(PSO),the method given in this paper shows up to two times faster convergence rate,and for some subjects,the new method shows ten times higher precision.