State estimation algorithm for ill-conditioned power systems by Newton's method

Abstract

A Newton's method for static-state estimation of power systems with weighted-least-square formulation is presented. This method uses the Brown-Dennis technique to update the analytically evaluated information matrix to approximate the Hessian matrix such that the linear convergence of Gauss-Newton algorithm is improved to nearly second-order convergence of Newton's method. The method is particularly effective in state estimation of ill-conditioned power systems faced with convergence difficulties. The additional price to be paid in implementing the new method into existing online programs with Gauss-Newton algorithm (standard WLS algorithm) is quite small, as the terms of the Hessian matrix are extracted from the Jacobian matrix.