基于神经网络集成的软件可靠性预测研究

英文篇名：Research on software reliability prediction based on neural network ensemble
中文刊名：计算机工程与设计
英文刊名：Computer Engineering and Design
作者：韦大伟 ; 崔正斌
英文作者：WEI Da-wei ; CUI Zheng-bin(Institute of Electronic Technology ; PLA Information Engineering University ; Zhengzhou 450004 ; China)
中文关键词：软件可靠性 ; 可靠性预测 ; 神经网络集成 ; 遗传算法 ; 主成分分析
英文关键词：software reliability ; reliability prediction ; neural network ensemble ; genetic algorithm ; principal component analysis
出版日期：2010-10-16
机构：解放军信息工程大学电子技术学院;
年：2010
期：19
出版单位：计算机工程与设计

摘要

为解决软件可靠性预测精度差和泛化能力不强问题,提出一种遗传算法集成神经网络的软件可靠性预测模型。通过遗传算法对神经网络集成权重进行了优化,并用主成分分析方法对软件属性度量数据进行了预处理,降低数据维数,简化神经网络的结构,加快神经网络的运算速度。仿真实验结果表明,基于遗传算法集成神经网络的软件可靠性预测模型同BP网络、LVQ网络和PNN网络相比具有更好的预测精度和泛化能力。
To solve the problem of software reliability prediction effectively,a model of neural network ensemble based on genetic algorithm(GA-NNE) is proposed,by the genetic algorithm neural network ensemble weights are optimized,and the method of principal component analysis is conducted on the experimental data pre-processing to reduce the dimension of the experimental data to simplify the structure of neural network,speed up the neural network computational speed.The simulation results show that the software reliability prediction model based on GA-NNE is effective in the prediction of software reliability,and has better precision than BP network,LVQ network and PNN network.

引文

[1]Taghi M Khoshgoftaar,Naeem Seliya.Tree-based software qua-lity estimation models for fault prediction[C].METRICS'02,2002:123-128.
    [2]Elish K O,Elish M O.Predicting defect-prone software modulesusing support vector machine[J].Journal of Systems and Soft-ware,2007,81(5):649-660.
    [3]胡求索.基于人工神经网络的软件质量预测模型研究[D].桂林:广西大学,2007:44-48.
    [4]沈掌泉.神经网络集成技术及其在土壤学中应用的研究[D].杭州:浙江大学,2005:38-55.
    [5]刘悦.神经网络集成及其在地震预报中的应用研究[D].上海:上海大学,2005:28-39.
    [6]Salah Bouktif,Balazs Kegl,Houari Sahraoui.Combining soft-ware quality predictive models:an evolutionary approach[C].Proceedings of the International Conference on Software Main-tenance,2002:385-392.
    [7]巩文科,李心广,赵洁.基于神经网络集成的入侵检测方法的研究[J].计算机工程,2007,33(8):152-156.
    [8]NASA.Metrics Data program(MDP)database[OL].http://mdp.ivv.nasa.gov/.