Distribution line parameters estimation framework with correlated injections using smart meter measurements

Abstract

Accurate estimation of power system line parameters (resistance and reactance) is vital for operational and planning studies in power distribution systems. However, obtaining these estimates is challenging due to temperature and aging dependencies, limited measurements, correlated injections from renewable sources, and change/addition in network elements in due course of time. Most of the existing research for line parameter estimation (LPE) uses regression-based approaches, which are highly susceptible to noise in measurement data. In addition, these methods are unable to converge in the case of an ill-conditioned matrix, which often occurs in the active distribution system due to correlated injections. Furthermore, existing work requires μPMUs at each node of the distribution network, which may increase the overall cost. To address these challenges, this paper presents a data-driven error-in-variable model in the optimization-based framework for LPE in low and medium-voltage networks using smart-meter data. The proposed LPE method is specifically designed to handle correlated power injections and noise in measurement data. An equidistant loss sampling approach is also introduced for injection sample selection, reducing sample complexity. Simulation results on a real 6-Bus and IEEE 33-Bus system depict the ability of the proposed method to obtain line parameters with correlated and erroneous injections.