Sliding-Priors for Bayesian Information Fusion in SCADA+PMU-based State Estimation

This paper describes a new development within the conceptual framework BAYSE (Bayesian State Estimation), which enables the full integration of the SCADA (Supervisory Control and Data Acquisition) data with PMU (phasor measurement units) data. It is based on Bayesian inference principles and extends the concept of the prior distributions to accommodate a broad set of past state conditions, under a sliding window approach. By choosing an appropriate window length, the method enhances accuracy under stationary conditions, with a reduced impact under system changes. The work also submits a rectangular coordinates transformation procedure, based on the Jacobian method, to consistently integrate polar coordinates estimations with the PMU linear model (in rectangular coordinates). The paper presents the new approach in proof-of concept mode over a didactic test-bed, using real PMU time series, to emphasize the enhanced accuracy and good asymptotic properties.