The State Estimator function in a control center today is a suite of three programs solved sequentially: topology processing, state estimation, and bad data detection-identification. The state estimation equations are nonlinear because the inputs are mostly real and reactive power measurements. A linear state estimator is possible if the inputs are only complex currents and voltages and if there are enough such measurements to meet observability and redundancy requirements. The main contribution in this paper is the suggestion that the topology processing function and the bad data detection-identification be done at each substation rather than at the control center. It is shown how this two-level processing is faster and more accurate leaving the control center level state estimator solution free of the bad data errors that are major problems today.

%B 2009 IEEE/PES Power Systems Conference and Exposition (PSCE) %I IEEE %C Seattle, WA, USA %P 1 - 6 %8 03/2009 %@ 978-1-4244-3810-5 %R 10.1109/PSCE.2009.4840218