粒子群算法优化的BP网络预测软件质量  

Software quality prediction of BP network based on PSO

作　　者：宫丽娜[1] 马怀志[1] 

机构地区：[1]枣庄学院信息科学与工程学院,山东枣庄277160

出　　处：《计算机工程与应用》2014年第23期65-68,共4页Computer Engineering and Applications

摘　　要：预测软件质量的技术中,软件建模技术是软件质量评价体系中的关键技术,它可以发现软件中度量数据和软件质量要素之间的非线性关系。BP神经网络能够很好地模拟度量数据和质量要素之间的非线性关系,但是BP网络存在易于陷入局部极小和收敛速度慢的问题,所以提出了用粒子群算法优化BP神经网络,通过优化的BP网络建立软件质量模型,这样能很好地解决BP网络收敛速度慢和局部极小的问题。在实现该进化BP神经网络的基础上,利用28组数据进行实验,并通过与BP模型的结果的比较,验证了该模型。The modeling technology of software which can find the nonlinear relationship between metric data and quality factors is the key technology in the software quality evaluation system. BP neural network is a kind of modeling method for the nonlinear relationship between metric data and quality factors, but there are some problems, such as slow conver-gence speed and easily getting into local minimum. So it proposes that using the optimized BP network based on PSO to establish the prediction model of software quality, which solves the problem of slow convergence speed and easily getting into local minimum well. In the basis of the evolutionary BP neural network, through the experiment with 28 groups of data, and by comparing with the result of BP model, the model is validated.

关 键 词：软件质量预测模型 软件度量 神经网络 粒子群算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]