深度自编码网络的集成学习ICPS入侵检测模型  

Integrated Learning ICPS Intrusion Detection Model of Deep Auto-encoder Network

作　　者：孙子文 金浩[2] SUN Ziwen;JIN Hao(Engineering Research Center of Internet of Things Technology Applications Ministry of Education,Jiangnan University,Wuxi 214122,China;School of Internet of Things,Jiangnan University,Wuxi 214122,China)

机构地区：[1]江南大学物联网技术应用教育部工程研究中心,江苏无锡214122 [2]江南大学物联网工程学院,江苏无锡214122

出　　处：《信息与控制》2021年第5期591-601,共11页Information and Control

基　　金：国家自然科学基金资助项目(61373126);中央高校基本科研业务费专项资金资助项目(JUSRP51510);江苏省自然科学基金资助项目(BK20131107)。

摘　　要：当前工业信息物理系统网络环境下的网络数据呈现出比以往更为庞大和复杂的特性,传统采用机器学习方式提取特征的过程繁琐且计算量大,不利于实时检测工业网络数据流量.鉴于此,研究一种基于深度自编码网络的集成学习入侵检测模型.首先,通过栈式叠加多个正则降噪自编码网络构造深度自编码网络,以非线性方式降低数据特征维数,获取到新的低维特征数据集;然后,采用集成多个深度信念网络模型以投票方式对降维后数据进行分类识别.实验结果表明,通过比对传统的降维方式和入侵检测方法,本文方法在分类性能和检测效率方面均有较优的效果.Recent network data under the current industrial cyber-physical system presents more significant and complex characteristics.The traditional use of machine learning to extract features is cumbersome and computationally expensive,which is not conducive for real-time industrial network data traffic detection.Given this,an integrated learning intrusion detection model based on a deep autoencoder network is studied.First,a deep autoencoder network constructed by stacking multiple regular denoising autoencoder networks is used to reduce the data feature dimension non-linearly to obtain a new low-dimensional feature set.Furthermore,multiple deep belief network models are integrated to classify and identify the reduced feature using voting.Compared with traditional dimensionality reduction methods and intrusion detection methods,the experimental results showed that the proposed method performed better in classification and detection applications.

关 键 词：工业信息物理系统 正则降噪自编码网络 深度自编码网络 集成学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]