A robust LWS state estimation including anomaly detection and identification in power systems

The electrical power system measurements are transmitted to the control center through a communication network. These measurements may contain bad data due to communication errors, systematic errors, incorrect wiring or infrequency of instrument calibration. As stated in the statistical literature, the estimators with high breakdown point are robust enough to overcome the effect of bad data. This paper discusses the application of one such estimator, Least Winsorized Square (LWS) by applying it to Tracking State Estimation (TSE). The proposed estimator detects, identifies the anomalies such as the existence of bad data and sudden change in load if present in the power system. Discrimination between the anomalies has been accomplished by a test of asymmetry (skewness measure). The proposed estimator has an inbuilt bad data rejection property with an ability to operate at any operating point without undergoing re-analysis phase of the TSE. The state estimation problem has been solved using JADE-adaptive differential evolution algorithm as an optimization problem. The effectiveness of the LWS technique has been tested on three different IEEE standard test systems. The results of the proposed method are compared with conventional weighted least square, particle swarm optimization, and gravitational search algorithm based state estimation techniques. Simulation results demonstrate the efficacy of the proposed algorithm as state estimates of the proposed method are highly precise even when the anomalies are present in the power system.