计算机科学
贝叶斯网络
工业控制系统
自动机
依赖关系(UML)
实时计算
人工智能
数据挖掘
控制(管理)
作者
Yinghao Sun,Gang Wang,Pei-zhi Yan,Lifang Zhang,Yong Xu
标识
DOI:10.1007/978-3-030-96282-1_6
摘要
The current industrial control system attack detection methods are single, the detection results are fuzzy and cannot be applied to the domestic industrial environment. In response to the above problems, an industrial control system attack detection model based on Bayesian network (BN) and Timed automata (TA) theory is proposed. First, collect the real industrial purification data of the aluminum factory, that is, the sensor and actuator signals, and preprocess the signals through time compression, segmentation, and queue division; secondly, establish Timed automata and Bayesian network models respectively, using probability time automatization The computer simulates the regular behavior of the time series, and at the same time uses the Bayesian network to build the dependency relationship between the sensor and the actuator; finally, the model’s detection result of the attack data is calculated. Theoretical analysis and experimental results show that compared with Deep Neural Network (DNN) and Support Vector Machine (SVM), the model in the article has improved time and accuracy.
科研通智能强力驱动
Strongly Powered by AbleSci AI