基于蒙特卡洛模拟方法的软件质量预测研究  

Software Quality Prediction Based on Monte-Carlo Simulation

作　　者：陈鑫[1,2] 周敏刚 王闯[3] 郭强 CHEN Xin;ZHOU Min-gang;WANG Chuang;GUO Qiang(Xi′an Aeronautics Computing Technique Research Institute,AVIC,Xi′an 710000,China;Xi′an Kai-Xiang Computer Software Co.,Ltd.,Xi′an 710000,China;Generic Technology Co.,Ltd.,AVICAS,Yangzhou 225000,China)

机构地区：[1]航空工业西安航空计算技术研究所,陕西西安710000 [2]西安凯翔计算机软件有限责任公司,陕西西安710000 [3]航空工业机载系统共性技术有限公司,江苏扬州225000

出　　处：《航空计算技术》2022年第2期35-39,共5页Aeronautical Computing Technique

基　　金：装备技术基础科研项目资助(202ZX31006)。

摘　　要：对装备软件的质量评价方法大多在软件研制后期开展,发现软件质量缺陷可能为时已晚。针对软件质量难以及早预测的问题,采用了蒙特卡洛模拟方法,讨论了获取度量元统计分布及参数的2个前提和3个步骤、生成大量符合统计分布随机数的方法、以及当采样值统计特征不明显时通过频数和频率确定推荐值的做法,详细说明了用蒙特卡洛方法预测软件质量的5个步骤。应用实例表明,通过蒙特卡洛方法,可以在软件研制前期或中期阶段,就预测出软件质量每个等级出现的概率和置信度,可以为及早采取软件质量纠偏措施提供决策量化依据。The current software quality evaluation methods for the equipment software are often carried out in the later development stage.It is too late when software quality defects are found.To solve the problem it is hard to predict the software quality as early as possible, the Monte-Carlo Simulation method is used.The two preconditions and three steps of getting the statistical distributions and parameters of metric units, the method of generating a lot of random numbers which meet the statistical distributions, the practice of setting the recommending value by the frequency and relative frequency when the statistical features of samples are not significant are discussed.The five steps of predicting software quality using the Monte-Carlo simulation method are explained in detail.It is demonstrated by an application instance that the probability and the confidence level for each degree occurrence of software quality can be predicted in the early or middle stage of software development by the Monte-Carlo simulation.These results can be offered as the quantitative reasons of decision which the software quality deviation correction measures need taking in time.

关 键 词：软件质量预测 蒙特卡洛模拟 统计分布 概率和置信度 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]