An L-1 Regularized Forecasting-Aided State Estimator for Active Distribution Networks

With the increased penetration of intermittent distributed energy resources (DERs) and the inclusion of complex loads, the states of an active distribution network (ADN) can experience sudden variations. Most distribution networks are only partially observable using available measuring devices. A distribution system state estimator (DSSE) uses pseudo, i.e., forecasted power injection measurements for the unobserved buses to perform state estimation. The pseudo measurements may fail to capture the sudden change in DERs and loads’ power. Consequently, the conventional DSSEs may fail to estimate the states when there is a sudden change in the loads and DERs connected to the unobserved buses. This paper proposes a forecasting-aided state estimator for ADNs, which can capture the sudden changes in states caused due to intermittent variations in DERs and loads’ power. The estimator requires measurements from a limited number of
$\mu $
PMUs, which are placed according to the depth-of-one unobservability criterion. The present work proposes to split the state vector into two parts. The first part captures the gradual Spatio-temporal variations in the states. The second part captures temporally uncorrelated variations in the states that may occur due to the sudden change in power injections. The proposed estimator tracks both state vector components using an l1-regularization-based estimation model. The algorithm is tested on an unbalanced IEEE benchmark test feeder using real load profiles.