摘要
工业控制系统安全是生产安全的保障,各类网络攻击给系统带来安全隐患,及时监控预警系统异常能有效避免生产事故的发生。针对工控系统中时序态势要素,为提高工况态势预测精度,提出基于GRU-SVM的态势预测模型。对采集的大量非线性时序数据进行预处理并提取态势要素;通过门控循环单元(Gated Recurrent Unit,GRU)融合注意力机制对多个单维要素分别训练,得到多维态势要素组;将要素组输入到支持向量机(Support Vector Machines,SVM)评估模型,获得未来安全态势值。实验结果表明,与其他算法相比,GRU-SVM预测模型有较高的精准度。
The safety of industrial control system is the guarantee of production safety.It is necessary to prevent various network attacks from bringing security risks to the system.Timely monitoring the abnormalities of the early warning system can effectively avoid production accidents.A situation prediction model based on GRU-SVM is proposed for the time series nonlinear situation elements in the industrial control system to improve the prediction accuracy of the working condition situation.A large number of nonlinear time series data are preprocessed and situation elements are extracted.The Gated Recurrent Unit(GRU)integrates the attention mechanism to train multiple single dimension elements respectively,and obtains multi-dimensional situation element groups,the predicted element group is input into the Support Vector Machines(SVM)evaluation model to obtain the future security situation evaluation value.Experiments show that GRU-SVM prediction model performs better than other algorithms.
作者
陈鸿韬
郭烁
洪悦
CHEN Hongtao;GUO Shuo;HONG Yue(College of Information Engineering,Shenyang University of Chemical Technology,Shenyang 110142,China)
出处
《电子设计工程》
2024年第5期7-12,共6页
Electronic Design Engineering
基金
辽宁省教育厅科学研究面上项目(LJKZ0460)。
关键词
工业态势预测
注意力机制
门控循环单元
支持向量机
industrial situation prediction
attention mechanism
Gated Recurrent Unit
Support Vector Machines
作者简介
陈鸿韬(1997-),男,广东梅州人,硕士。研究方向:工业控制系统信息安全。
