Pré-processamento de medidas para estimação de estado em condições adversas de monitoramento

Abstract

Static state estimators are still widely used for power systems monitoring and operation. This type of estimator makes use of data related only to the instant under estimation, ignoring past information. This paper proposes a measurement pre-processing method using past data and regression techniques for error detection and correction, providing improved data to static state estimators, specially under adverse situations. This approach requires no knowledge of the system dynamics, relying simply on data analysis, and is able to work with constrained datasets and high granularity. Unlike forecast aided state estimators, the proposed method operates over measurement data and does not output state variables, providing processed measurements instead. Results obtained from simulations with the 30 Bus IEEE test system indicates the proposed method can detect and correct errors in measurements, improve its accuracy and reduce effects of data losses by providing high quality pseudomeasurements.