摘要
随着电力行业逐步跨入新型数字化全面互联时代,电力物联网在其中扮演着越来越重要的角色。在电力物联网快速发展的同时,负责承载数据流和信息流的电力信息网络存在遭受入侵的风险,而网络流量的异常检测将成为解决这一问题的重要手段。文章提出一种基于深度非对称稀疏自编码器(deep nonsymmetric sparse autoencoder,DNSAE)和随机森林(random forest,RF)的网络入侵检测模型,在保证准确率的同时,实现更快速高效的识别。首先由DNSAE对网络流量数据进行特征提取,再将得到的抽象特征数据训练随机森林。实验结果表明,与深度信念网络(deep belief network,DBN)和堆叠非对称自编码器(stacked nonsymmetric deep autoencoder,S-NDAE)相比,此模型具备更高的检测效率。
With the power industry gradually stepping into the new era of digital comprehensive interconnection,the power Internet of things plays an increasingly important role in it.However,with the rapid development of power Internet of things,the power information network,which is responsible for carrying data flow and information flow,has the risk of invasion,and the anomaly detection of network traffic will become an important means to solve this problem.In this paper,a network intrusion detection model based on deep nonsymmetric sparse autoencoder (DNSAE) and random forest (RF) is proposed to achieve faster and more efficient recognition while ensuring accuracy.Firstly,DNSAE is used to extract features from network traffic data,and then these abstracted feature data are used as input for RF training.The experimental results show that this model has higher detection efficiency than DBN and S-NDAE.
作者
潘羿
李彬
PAN Yi;LI Bin(School of Electrical and Electronic Engineering,North China Electric Power University,Beijing 102206,China)
出处
《电力信息与通信技术》
2022年第5期23-29,共7页
Electric Power Information and Communication Technology
基金
国家电网有限公司总部科技项目资助“客户侧柔性资源互动运营与交易支撑技术研究”(5400-202019491A-0-0-00)。
关键词
入侵检测
电力信息网络
深度非对称稀疏自编码器
随机森林
网络安全
intrusion detection
power information network
deep nonsymmetric sparse autoencoder
random forest
cyber security
