基于Copula函数的故障相依软件系统可靠性模型  

Failure-Dependent Software System Reliability Model Based on Copula Function

作　　者：郑周桃 杨剑锋 贺孟兰 ZHENG Zhou-tao;YANG Jian-feng;HE Meng-lan(School of Mathematics and Statistics,Guizhou University,Guizhou Province,Guiyang 550025,China;School of Data Science,Guizhou Institute of Technology,Guizhou Province,Guiyang 550025,China)

机构地区：[1]贵州大学数学与统计学院,贵州贵阳550025 [2]贵州理工学院大数据学院,贵州贵阳550025

出　　处：《计算机仿真》2024年第8期386-393,共8页Computer Simulation

基　　金：国家自然科学基金资助项目(71901078);贵州省电力大数据重点实验室(黔科合计Z字[2015]4001)。

摘　　要：传统的软件系统可靠性模型常常容易忽略子系统之间的故障相依性从而不能很好的反映对整个软件系统的可靠性建模分析。故根据实际情况考虑了屏蔽数据的存在且基于Copula函数考虑了子系统之间的失效相依性,进而提出了具有故障相依的软件系统可靠性叠加模型来解决此类相依性问题。又可靠性叠加模型的参数估计是难点,所提模型参数复杂,涉及到的似然函数变量维度高难以求解,故从可靠性出发,利用期望最小二乘算法(ELS)和最小二乘估计(LSE)来求解模型参数的近似估计值。最后通过一组实际数据和一组仿真数据进行数值验证,从最后的结果分析发现所提模型结果更加贴合于实际,且能很好的拟合累计故障数,从而验证了所提模型的有效性。The traditional software system reliability model often overlook the fault dependence between subsystems,and cannot reflect the reliability modeling and analysis of the entire software system well.Therefore,according to the actual situation,the existence of shielded data and the failure dependence between subsystems are considered based on Copula function,and then a software system reliability superposition model with fault dependence is proposed to solve this kind of dependency problem.In addition,parameter estimation of reliability superposition model is difficult.The proposed model parameters are complex and the likelihood function variables involved too high dimension to solve.Therefore,from the perspective of reliability,expected least squares algorithm(ELS)and least squares estimation(LS)are used to solve the approximate estimates of model parameters.Finally,a set of actual data and a set of simulation data are used for numerical verification.From the analysis of the final results,it is found that the results of the proposed model are more consistent with the reality,and can well fit the cumulative fault number,thus verifying the effectiveness of the proposed model.

关 键 词：可靠性 故障相依 屏蔽数据 期望最小二乘算法 仿真数据 

分 类 号：TP311.53[自动化与计算机技术—计算机软件与理论]