摘要
针对应用服务器系统中存在的软件老化现象,监测系统资源消耗的性能参数,采用粗糙小波网络建立系统老化预测模型。该模型首先采用信息熵约简方法化简系统性能参数,从而确定粗糙小波网络的输入变量;然后采用自适应遗传算法对网络结构和参数进行优化。最后通过实验表明,该模型比传统的神经网络和小波网络模型具有更高的预测精度及更好的收敛性能。
Concerning the software aging in application sever the systematic performance parameters were observed and the aging forecast model was set up based on Rough Wavelet Network RWN.Then the dimensionality of input variables of RWN was reduced by information entropy reduction method and the structure and parameters of RWN were optimized with adaptive genetic algorithm.Finally the experiments were carried out to show that the aging forecast model based on RWN is superior to the wavelet network model and wavelet network model in convergence rate and forecasting precision.
出处
《计算机应用》
CSCD
北大核心
2010年第8期2024-2028,共5页
journal of Computer Applications
基金
陕西省教育厅科研计划项目(116-220915)
西安理工大学科技创新研究计划项目(116-21090)
关键词
应用服务器
软件老化
软件可靠性
粗糙小波网络
遗传算法
application server software aging software reliability Rough Wavelet Network RWN genetic algorithm