Research on Real-time Detection Method of Distributed Setpoint Anomaly in Nuclear Power Industrial Control System Based on Stream Computing
Hanjun Gao, Wei Chang Feng, Hang Tong · 2023
The nuclear industry as a national important basic support industry, its safety is more and more attention by the state, nuclear power control system as the core of the nuclear industry, its safe and stable operation is particularly important. With the development of attack technology, more and more forms of attack from a single attack to a distributed attack, the attack has a single flow of normal characteristics, so that the existing single-point attack detection methods can not meet the malicious attack behavior detection requirements. In order to cope with the challenge, this paper takes the distributed setpoint anomaly attack as the research object, and proposes a real-time detection method security-Setpointer based on stream computing. The method constructs an aggregation model for the setpoints in the time window according to the controlled objects for the sequence of operation aggregation, and at the same time, due to the large scale of the core monitoring point, in order to meet the timeliness and real-time detection. This paper implements a distributed DTW algorithm based on stream computing, trains on a large amount of historical a priori data, realizes the feature learning and characterization of normal setpoint sequences, and adopts different DTW real-time detection tasks for different monitoring points to realize the anomaly detection and analysis of different monitoring points. Finally, in order to verify the effectiveness of the method proposed in this paper, functional and performance tests are carried out, and the test results show that the method proposed in this paper is effective.