Applicability and Enhancement of Matrix-Completion-Based Distribution System Monitoring With Limited Measurements

The transition to a sustainable energy future will result in greater dynamics of aggregated power generation and demand in distribution networks. A vast network with limited real-time measurements presents challenges in comprehensively monitoring the system status and responding promptly to localized disturbances. Sparse signal reconstruction techniques based on affine-constrained rank minimization have been proposed to estimate distribution feeder states with a few measurements, which are presumably well-suited for contexts with very limited observability. Although previous studies on matrix completion-based distribution system state estimation (MCDSSE) have demonstrated acceptable node voltage estimation performance, their effectiveness in estimating branch flow states, to monitor line and transformer loading, remains unproven in the literature. This study assesses the consistency between estimated system states obtained using a variant of the MCDSSE method and actual branch flows and node injection powers to judge the method’s applicability in distribution system monitoring with limited available measurements. The test results confirm that solving a low-rank minimization problem does not enable MCDSSE to recover a realistic AC solution in areas with insufficient measurements, and show that degenerate solutions and multiplicity appear to occur frequently. To leverage its advantage in providing complete solution of the system states, based on operator-defined visibility requirements, four fidelity-monitoring-driven MCDSSE meter placement methods were tested to enhance its acceptability for system monitoring with affordable additional measurements. Numerical examples highlighting practical implications are provided.