摘要
为了避免单一节点宕机或其他网络相关系统的故障造成核电站整体仪控系统故障,构建了基于仪控系统的核电站网络安全系统防护体系模型,通过基于差值序列的超限学习机算法提取数据的周期性特征,采用空间卷积算法强化数据特征,经过模糊神经网络挖掘数据规律,使用对数神经网络实现数据整合,将数据优化,并进一步分析数据挖掘效果。通过仿真测试,发现使用该算法模型增大了数据预警提前量,延长了数据预警周期并且不会降低数据预警的敏感度和特异度;对预警系统的稳定性进行分析。比较发现,使用新设计的预警系统,可以降低数据延迟时间、丢包率和计算响应周期,同时宕机时间比也有所降低。实验证明该系统具有一定的可行性。
In order to avoid the failure of the overall instrument and control(I&C)system of the nuclear power plant caused by the downtime of a single node or the failure of other network related systems,a protection architecture model of the network security system of the nuclear power plant based on the I&C system is constructed.The periodic characteristics of data are extracted by the transfinite learning machine algorithm based on difference sequence,the data characteristics are strengthened by the spatial convolution algorithm,the data rules are mined by the fuzzy neural network,the data integration is realized by the logarithmic neural network,the data is optimized,and the data mining effect is further analyzed.Through simulation test,it is found that this algorithm model increases the data early warning lead,prolongs the data early warning cycle,and does not reduce the sensitivity and specificity of data early warning.By analyzing and comparing the stability of the early warning system,it is found that using the designed early warning system can reduce the data delay time,packet loss rate and calculation response cycle,and also reduce the downtime ratio.The experiment proved that the system has certain feasibility.
作者
谢磊
杨鑫
杨林
Xie Lei;Yang Xin;Yang Lin(Shandong Nuclear Power Company,Shandong Haiyang,265116,China;State Nuclear Power Automation System Engineering Co.,Ltd.,Shanghai,200241,China)
出处
《机械设计与制造工程》
2022年第10期131-134,共4页
Machine Design and Manufacturing Engineering
关键词
核电站
网络安全系统
仪控系统
防护体系模型
模糊神经网络
nuclear power plant
network security system
I&C system
protection system model
fuzzy neural network