广义动态模糊神经网络对软件质量的预测

摘要

软件质量预测是软件质量评价体系中的关键技术,针对软件质量预测过程中难以建立精确数学模型的特点,提出了将广义动态模糊神经网络应用于软件质量预测模型中.以模糊ε的完备性作为高斯函数宽度的确定准则,避免了初始化过程中选择的随机性,同时,能对模糊规则和输入变量的重要性作出评价,从而使每条规则的输入变量的宽度可以根据它对系统性能贡献的大小实施在线自适应调整.通过对软件可靠性的仿真实验结果证明,广义动态模糊神经网络不仅适合模糊规则抽取也可用于系统建模,而且具有较高的辨识精度和效率.%The prediction model of software quality is the key technology in the software quality evluation system. According to the characteristic of establishing the precise mathematical model difficultly in the process of software quality prediction,this paper pre-dictes the prediction model of software quality by generalized dynamic fuzzy neural network. This algorithm uses fuzzy-complete as the norm of the width of the Gaussian function,which avoids the randomness of selecting in the initialization. At the same time,the algorithm can evaluate the fuzzy rules and the importance of input variables,which makes the width of the input variables of each rule online adjust adaptive according to its contribution to performance. By simulating the reliability of software,this paper turnes out that generalized dynamical fuzzy neural networks is not only suitable for fuzzy rule extraction but also used for system modeling,which also has a high recognition accuracy and efficiency.