Name
Abstract | ||
|---|---|---|
With the popularity of Supervisory Information System (SIS), Supervisory Control and Data Acquisition (SCADA) system and Internet of Things (IoT) sensors, we can easily obtain abundant sensor data in manufacturing. We could save manufacturing maintenance costs and prevent further damages if we can accurately predict system anomalies from the sensor data. Yet learning from individual sensors often cannot directly determine whether the system will have anomaly because each sensor only measures a partial state of a big system. By detecting events across sensors collectively and their temporal dependencies, this paper proposes a new system-level anomaly prediction framework by mining anomaly dependency graph from sensor data. The advantages of the approach include explainability, collective prediction and temporal sensitivity. We applied our approach with a real-world power plant dataset to evaluate its feasibility. |
Year | DOI | Venue |
|---|---|---|
2018 | 10.1109/BigDataCongress.2018.00028 | 2018 IEEE International Congress on Big Data (BigData Congress) |
Keywords | Field | DocType |
Smart Manufacturing, Anomaly Prediction, Predictive Maintenance, Data Stream Mining, Sensor Data Driven | Information system,Time series,Data mining,Anomaly detection,Computer science,SCADA,Predictive maintenance,Dependency graph,Electricity generation,Power station | Conference |
ISSN | ISBN | Citations |
2379-7703 | 978-1-5386-7233-4 | 0 |
PageRank | References | Authors |
0.34 | 0 | 5 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Jianwu Wang | 1 | 215 | 26.72 |
| Chen Liu | 2 | 30 | 6.29 |
| Meiling Zhu | 3 | 16 | 3.18 |
| Pei Guo | 4 | 0 | 1.01 |
| Yapeng Hu | 5 | 0 | 0.34 |