软件可靠性专家系统(SRES)中经验模型的奇异性问题与参数估计方法

SINGULARITY AND PARAMETER ESTIMATION METHODS OF EXPERIMENTAL MODELS IN SOFTWARE RELIABILITY EXPERT SYSTEM (SRES)

本文作者以“一切让数据说明问题”为原则，采用人工智能技术，开发出软件可靠性专家系统SRES（SoftwareReliabilityExpertSystem），根据统一的标准，通过对一组软件可靠性模型SRM（SoftwareReliabilityModels）的拟合结果，经过推理，得出一个“最适”模型，推荐给用户，作为估测该软件系统的一个“标准模型”，同时为软件开发方和使用方所接受，以解决软件可靠性模型应用中的不一致性难题．在开发SRES的过程中，我们选用的经验模型有三个经分析具有奇异性，增大了对它们的参数进行估计的难度．针对这些，本文研究并报告软件可靠性专家系统中经验模型的奇异性问题与参数估计方法．

According to the principle, 'Let failure data explain every thing', theauthors built a software reliability expert system (SRES) by adopting the artificialintelligence technology. By reasoning out the conclusion from the fitting results offailure data of a software project, the SRES can recommend users 'the suitablemodel' as a unique software reliability measurement model. The SRES canovercome the inconsistency in applications of software reliability models well. In thispaper, investigation results of singularity and parameter estimation methods ofexperimental models in SRES are reported. The experimental model base consists offourteen software reliability models, and in the experimental model base of SRESthere are three software reliability models which are singular. They are non-homogeneous Poisson process model with three parameters (Nhp3ad), Littlewood'sBayesian reliability growth models(LVLM and LVQM). The singularity forms areanalyzed in the paper. Further, the parameter estimation methods designedspecifically for these three software reliability models are reported. At last, the estimationresults by using complete software failure data, NTDS, and incomplete softwarefailure data, D. inc, are given. The calculation results demonstrate that theestimation methods are very effective to overcome the singularity.