Anomaly detection method based on multi-criteria evaluation for energy data of steel industry
Hao Wu, Feng Jin, Jun Yan Zhao, Wei Wang · 2021 IEEE 10th Data Driven Control and Learning Systems Conference (DDCLS) · 2021
The stability and integrity of the monitoring data in the energy system of the iron and steel industry is of great significance for ensuring the safety of the system. Aiming at the data with periodic characteristics in the steel energy system, an abnormal data detection method based on multi-criteria evaluation is proposed in this study. The data time series is divided according to its periodic characteristics, and each sub-period sequence are evaluated through the quasi-measures established in this paper. Then the outlier detection of the periodic time series is realized through the AFCM (Adaptive fuzzy C-means) of these evaluation results. The simulation results based on the actual operating data of a steel enterprise show that this method can identify the local abnormal state in the original time series and improve the detection efficiency.