摘要
将软件可靠性测试阶段获得的失效数据作为时间序列进行多尺度分解,对分解到不同尺度上的数据分别利用不同的时序预测模型进行分析,得到软件可靠性多尺度预测模型。数据实验表明与单一时序预测模型相比,该模型逼近和预测效果良好,具有较高的预测精度和很好的模型适应性。
It is introduced that firstly the fault data sets coming from software reliability test phase is to be multi-scale discomposed as time series, and then different time series analysis models are built to analyze data of different scale, so the multi-scale prediction model is used in software reliability estimation. The experiment data indicate: compared with other single time-series prediction models, the model can provides better approximate and prediction effort, have higher precision of prediction and better adaptability.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第11期2520-2523,共4页
Computer Engineering and Design
基金
军队科研计划基金项目(2002装司字第775号)。
关键词
软件可靠性
时间序列
多尺度分析
失效数据
RBF神经网络
AR模型
software reliability
time series
multi-scale analysis
fault data
RBF neural network
automatic regression model