基于神经网络的级联软件可靠性模型

针对一般经典软件可靠性模型适用范围的局限性问题和预测精度问题,提出了一种新的级联模型.将4个经典软件可靠性模型的输出作为误差背向传播(error back propagation,BP)神经网络的输入,级联组合成一个软件可靠性模型,称之为级联软件可靠性模型.通过对一组经典的实际软件故障数据SYS1进行实验,将级联软件可靠性模型与4个经典软件可靠性模型预测的结果进行对比,结果表明级联软件可靠性模型的预测精度要远远高于4个经典软件可靠性模型,而且具有更好的通用性.

Cascade software reliability model based on neural network

A new cascade software reliability model is proposed. The new model is to deal with the problem of narrow application range of classical software reliability model and to obtain higher prediction precision. The output of four classical software reliability models are applied as the input of BP network to establish a new cascade model, which is called cascade software reliability model. Through experiments based on real failure data SYS1, the cascade model is compared with four classical models. The results show that the pre-diction precision of cascade model is higher than that of four classical models. Furthermore, the cascade software reliability model has better generalization.