Data-Driven Bayesian Framework to Mitigate Smart Meter Asynchrony in Forecasting Aided Distribution System State Estimation

Abstract

Smart meters (SMs) are pivotal in improving the accuracy of distribution system state estimation (DSSE) in the modern-day monitoring and control paradigm. Due to limited communication bandwidth or signal strength, the synchronized data conveyance from SMs to the control center remains an onerous task. Performing DSSE using nonsynchronized SMs introduces asynchrony errors, which affect estimation accuracy. Toward this objective, a gated recurrent unit (GRU)-based SM data forecasting is implemented to facilitate SM data imputation, thereby mitigating SM asynchrony. Simultaneously, the effect of out-of-date (OD) variance incurred due to SM asynchrony, given the node load changes, needs to be priorly detected and iteratively updated to improve estimation results. This burdensome task gets eliminated by adopting variational Bayesian (VB) adaptive cubature information filter (VB-ACIF), which can intrinsically adapt the unknown measurement uncertainties incurred by the SM OD variances. Numeric results from test systems reveal that forecasting aided-DSSE (FA-DSSE) using VB-ACIF delivers improved estimation accuracy and is suitable for practical distribution system operation. The proposed estimator’s robustness in dealing with multiple anomalies and correlated data communication failure (CF) is also validated for the proposed FA-DSSE framework.