State Estimation in Power Distribution Systems

Abstract

State estimation in power distribution systems is a key component for increased reliability and optimal system performance. Well understood in transmission systems, state estimation is now an area of active research in distribution networks. While several snapshot-based approaches have been used to solve this problem, few solutions have been proposed in a dynamic framework.

In this thesis, a Past-Aware State Estimation (PASE) method is proposed for distribution systems that takes previous estimates into account to improve the accuracy of the current one, using an Ensemble Kalman Filter. Fewer phasor measurements units (PMU) are needed to achieve the same estimation error target than snapshot-based methods. Contrary to current methods, the proposed solution does not embed power flow equations into the estimator. A theoretical formulation is presented to compute a priori the advantages of the proposed method vis-a-vis the state-of-the-art. The proposed approach is validated considering the 33-bus distribution system and using power consumption traces from real households. Engineering insights are presented highlighting the major trade-offs in the choice of decision variables (number of PMUs, PMU accuracy, estimation time-step - i.e. elapsed time between two consecutive estimations) for the LDC: using a smaller time-step allows the LDC to relax the requirements on the PMU quality and their number.