Power system anomaly detection using innovation reduction properties of iterated extended kalman filter

Abstract

The performance of forecasting-aided state estimation can be significantly affected by the presence of anomalies, such as sudden load changes, bad data, or line outage/topology errors. The existing robust alternatives may suppress the influences of some anomalies, but not all of them, while the anomaly detection methods have very low accuracy or even may not detect, e.g., line outages. To bridge this gap, this paper proposes a new anomaly detection framework using the properties of innovation reduction in iterated extended Kalman filter (IEKF) and the normalized residual of static state estimator. The proposed framework can detect and distinguish above-mentioned anomalies with a very high accuracy. It is also robust to different levels of noise, different degrees of measurement redundancies, and different sizes/complexities of the networks. Simulation results carried out on the IEEE 14-, 39-, 57-, and 118-bus test systems demonstrated the effectiveness and accuracy of the proposed method.