A Variable Step Size Decoupled State Estimator

This paper describes an improved decoupled static-state estimator based on minimization of weighted least squares of the residuals (WLS). The basic idea of P-θ, Q-V decoupling of the fast decoupled load flow for node injections is extended to line flows. The solution is obtained alternately iterating the real and reactive power equations using constant submatrices of the 'information matrix'. Since no approximation is made in mismatch functions, the final solution is as accurate as the exact solution. This algorithm is compared with a 'constant gain algorithm' (CGA) using a full size gain matrix derived in terms of the Jacobian which is evaluated only at the first iteration. The 'decoupled state estimator' (DSE) requires considerably less storage and solution time but more iterations. However, by varying the step size during iterations, the convergence behavior of DSE approaches that of CGA. Thus the paper extends the research work reported in references [16, 17].