基于K-means和稀疏自编码器的工控网络安全态势评估研究  

Research on Industrial Control Network Security Situation Assessment Based on K-means and Sparse Autoencoder

作　　者：黄晓培 张春芳 HUANG Xiaopei;ZHANG Chunfang(China Software Testing Center(Software and Integrated Circuit Promotion Center of the Ministry of Industry and Information Technology),Beijing,100048 China)

机构地区：[1]中国软件评测中心(工业和信息化部软件与集成电路促进中心),北京100048

出　　处：《科技资讯》2025年第5期45-47,共3页Science & Technology Information

摘　　要：工控网络安全态势评估是保障工业控制系统安全的重要手段。为提高评估精度和效率,提出了一种基于改进K-means聚类算法和稀疏自编码器的评估模型。首先,模型使用稀疏自编码器对数据进行降维处理;其次,通过改进K-means聚类算法进行聚类。实验采用田纳西伊斯曼过程数据集,并与传统模型进行对比,结果表明,所提出的模型在准确率和评估效率上均优于其他对照模型,准确率达到98%,评估时间为87 ms。研究结果表明,该模型能够有效提高工控网络安全态势的评估能力。The assessment of industrial control network security situation is an important task to ensure the securityof industrial control systems.To improve the accuracy and efficiency of evaluation,a research proposes an evalua-tion model based on improved K-means clustering algorithm and sparse autoencoder.Firstly,the model uses asparse autoencoder to reduce the dimensionality of the data,and then clusters it using an improved K-means clus-tering algorithm.The experiment used the Tennessee Eastman process dataset and compared it with traditionalmodels.The results showed that the proposed model outperformed other control models in both accuracy andevaluation efficiency,with an accuracy of 98% and an evaluation time of 87 ms.The research results indicate thatthe model can effectively improve the assessment capability of industrial control network security situation.

关 键 词：改进K-MEANS 稀疏自编码器 态势评估 工控网络 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]