A Factorized Approach to WLS State Estimation

This paper presents a factorized state estimation methodology based on the solution of two successive WLS problems. In the proposed scheme, a minimal set of intermediate variables is first introduced so that the resulting measurement model is linear. Estimates of those variables are then used as pseudo-measurements of a subsequent nonlinear estimator, along with the associated covariance matrix. The aim of the preliminary step is to reduce the size of the raw measurement vector to the maximum extent, without losing any relevant statistical information. Simulation results show that the proposed approach converges faster, is computationally more efficient, and provides accurate estimates after the first linear stage.