Anomaly detection method of sensor data under cyber-physical attack on industrial sensor network

Abstract

The invention discloses a sensor data abnormal detection method under information physical attack of an industrial sensor network, and belongs to the field of sensor data abnormal detection. The method standardizes the sensor data; introduces a virtual sensor to expand the standardized data; converts the one-dimensional sensor data into a two-dimensional data format; directly generates a grayscale image from the two-dimensional data; uses the convolutional neural network (CNN ) classification algorithm to classify the converted images; expand the number of sensors to thousands and tens of thousands of levels, and test the scalability of the anomaly detection model and its application in large-scale industrial sensor networks. Multiple simulation experiments were performed, and multiple evaluation indicators were used to compare anomaly detection models based on different algorithms. The method of the invention can solve the problem that the sensor data of the industrial sensor network has complex relationships, and meets the real-time and accuracy requirements of the industrial network, and has good expandability.

Description

technical field [0001] The invention relates to the field of sensor data anomaly detection, in particular to a sensor data anomaly detection method under information physical attack of an industrial sensor network. Background technique [0002] With the scale, informatization, and intelligent construction of the national industrial control system, the security issues of the industrial control system have become increasingly prominent. How to grasp the behavioral nature, protocol characteristics and data characteristics of industrial control systems has undoubtedly become a key issue in the research of industrial control systems in our country. At present, China is making every effort to build an intelligent and information-based industrial control system. A large number of smart sensors, actuator access and real-time data update requirements have expanded the scale of industrial sensor networks, resulting in a large increase in data flow in industrial control systems. The c...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the modeling problem of sudden industrial sensor network sensor data anomaly detection, and propose a kind of industrial sensor network information physical attack for the deficiency of existing industrial sensor network sensor data anomaly detection research The anomaly detection method of sensor data under the following guidelines has guiding significance for sensor configuration planning and anomaly detection of industrial control systems

Images