基于灰色变权搜索的软件故障预测方法  

Method for Software Fault Prognosis Based on Grey Variable Weight Search

作　　者：刘博宁[1] 张建业[2] 张鹏[1] 潘俊辰[3] 

机构地区：[1]空军工程大学工程学院 [2]空军工程大学科研部 [3]中国人民解放军96656部队

出　　处：《计算机测量与控制》2012年第8期2085-2088,共4页Computer Measurement &Control

摘　　要：针对软件故障数目历史数据的小样本特征,在传统灰色模型基础上,将预测模型的建模问题转换为灰色权因子的搜索问题,建立了变权搜索模型进行故障数目预测;提出用实测和预测序列的均方误差和斜率距离构建搜索指标集,用欧式距离构建计算各指标权重的最优准则函数,对模型的灰色权因子进行全局最优搜索,权因子随故障发展过程实时变化;经实例仿真,预测与实测序列相对误差较小,验证了所提方法的可行性和有效性,对装备软件风险评估提供了有效的决策支持。Aimed at the small sample size historical data of software fault, the model establishing problem is switched to the grey weight searching problem, the variable weight search model for fault prognosis is established based on traditional model. For the best result of searching the grey weight in the range, the method for establishing guide--lines＇ set by using mean square error and slope--distance of measure and prognosis series is proposed, meanwhile the method for establishing the best function to reckoning the weight of each guide- line by using Euclidean--distance is proposed, the grey weight is Real--time changed along with the development of fault. According to the field example examination , the fractional error of measure and prognosis series was very little, the result shows that such method is feasi- bility and validity, and the method is serviceably to the decision--making of risk assessment for equipment＇s software.

关 键 词：软件故障预测 变权搜索 灰色模型 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]