State forecasting in electric power systems

The state vector of a power system varies with time owing to the dynamic nature of system loads. Therefore, it is necessary to establish a dynamic model for the time evolution of the state vector. The dynamic state estimation approach consists of predicting the state vector based on past estimations, followed by a filtering process performed when a new set of measurements is available. This paper presents a new algorithm for forecasting and filtering the state vector, using exponential smoothing and least-squares estimation techniques. The proposed algorithm is compared with another one based on standard Kalman filtering theory. Numerical results showing the performance for both dynamic estimators under different operational conditions are presented and discussed. Detection and identification of multiple bad data are also included. The new dynamic estimator exploiting state forecasting is extremely useful to real-time monitoring of power systems.