PMU data has the potential of providing a wealth of information on power system operation, health, faults and anomalies. PMU tend to provide tens of measurements per second, therefore automated anomaly detection is required; especially for use in real or near-time applications by power system operators. This paper demonstrates a method of detecting local anomalies in PMU data utilizing multiple linear regression. A window of near-time data is employed to generate a regression function that predicts the live data that arrives. If the error between the observed and predicted values exceeds a threshold an exception is noted. The threshold is dynamically updated based on the error in the regression function, allowing the method to work equally well on data of varying regularity. This anomaly detection method is not tuned to particular events and should detect novel occurrences. The method is evaluated on two numerical case studies, a genuine power system event and a man in the middle cyber attack. Real data was collected from PMUs placed on the Irish power system.
|Title of host publication||Proceedings of Power and Energy Society General Meeting (PESGM), 2018|
|Publication status||Published - 24 Dec 2018|
|Event||IEEE Power and Energy Society General Meeting - Portland, United States|
Duration: 05 Aug 2018 → 09 Aug 2018
|Name||IEEE Power & Energy Society General Meeting: Proceedings|
|Conference||IEEE Power and Energy Society General Meeting|
|Abbreviated title||IEEE PES GM 2018|
|Period||05/08/2018 → 09/08/2018|
Bibliographical noteThe research is supported by a British Council Newton Institutional Links Programme grant with Helwan University, Egypt.
- Anomaly detection
- Machine Learning
- Linear regression
- Cyber Security