Fast Decoupled State Estimation and Bad Data Processing

Abstract

This paper presents fast-decoupled state estimators, using also decoupled detection and identification of bad data. Bad data is eliminated by pseudo-measurement generation. This procedure avoids gain-matrix retriangulations or the use of modification techniques like Woodbury formula. In the identification process, the diagonal of the covariance matrix of the measurement residuals is calculated using the sparse inverse matrix method. Two main types of fast-decoupled estimators were tested : algorithm- decoupled and model-decoupled. The methods have been tested on IEEE 30-bus load-flow test system, and the FURNAS and CPFL systems that form part of the 835-bus interconnected 15 GW power system of the S.E. Brazil.