摘要
针对高校涉密项目风险因素多和保密环境复杂的特点,利用三层BP神经网络对能够逼近任意非线性函数的良好特性,突破传统上基于统计学方法进行预测的限制,综合了时间序列的计算简单,需要历史数据少的优点,设计了一种体现时序的多因素动态时间序列BP神经网络预测模型,并将模型运用于某高校涉密项目泄密风险的预测研究中。仿真实验表明,此方法切实可行,而且具有较好的预测精度。
Three-layer BP neural network can be approximated to the arbitrary nonlinear function,it breaks the prediction restrictions based on the traditional statistical method,integrates the advantages of simple calculation,less historical data of time series.A prediction model of multi-factor time series of BP network is designed and applied in risk prediction of college confidential project.The simulation result proves that the method is applicable and is of high precision.
出处
《信息安全与通信保密》
2011年第8期32-34,共3页
Information Security and Communications Privacy
关键词
时间序列
BP网络
风险预测
高校涉密项目
time series
BP neural network
risk prediction
college confidential project
