Power system state estimation with continuous and discrete variables

Abstract

As a basic energy management system function that processes real-time measurements, power system state estimation deals with both continuous and discrete variables to estimate the state of a power system. One possible source for the discrete parameters is the transformer taps, whose positions should be estimated with high confidence. Given that the tap estimation error causes a network topological modeling inaccuracy, its range should be minimized. Motivated to accurately estimate the power system state vector including transformer taps, this thesis presents an mixed-integer quadratic programming (MIQP) based rounding estimator that processes the initial continuous vector solution obtained from a conventional weighted least squares optimization, and produces the estimated discrete values without any iteration. For comparison purposes, three recently proposed methods for power system state estimation are explained and tested against the proposed MIQP based rounding estimator: an ordinal optimization formulation based on a sensitivity analysis, a probabilistic technique, and an adaptive threshold technique. Experimental results on the IEEE 30-, 57- and 118-bus benchmarks reveal a slight superiority of the proposed MIQP based algorithm in terms of overall accuracy and motivate follow on research.