摘要
在网络态势感知系统中需要对影响网络性能的各项指标设置阈值,通过设置阈值和阈值检查可以在网络出现性能问题时及时向管理人员告警。本文提出了一种利用BP神经网络来确定告警阈值的方法:在采集到的大量性能数据中选取典型值作为训练样本训练BP神经网络,输出该值隶属于各模糊区间的隶属度,最后利用检验样本找到各区间的分界点即为阈值。文章还利用MATLAB对BP神经网络进行了仿真实验,验证了该方法的有效性。
In the Network Situation Awareness System, setting and checking the threshold for every tolerance which influence network's performance can alarm network administrators when the network has a performance problem. A method that imposing BP neural network to set threshold for network performance tolerance is provided. What select some typical values from the selected performance data as train swatch to train BP neural network and output the degree that the value belongs to every fuzzy region, at last, find the dividing line also called threshold using the test data. An experiment which to simulate BP neural network by MATLAB is provided. It validates the validity of the method.
出处
《世界科技研究与发展》
CSCD
2008年第4期443-445,共3页
World Sci-Tech R&D
基金
高等学校博士学科点专项科研基金项目(20050217007)
国防预研重点资助项目(413150702)
武备预研基金资助项目(51416060104CB0101)