On the Imputation of Power System Measurement Streams with Imperfect Wireless Communication

Dependable measurement data are essential for the accuracy and integrity of power system state estimation and actuation. However, distribution grids empowered by wireless connectivity are subject to missing sensor observations due to channel stochasticity. In this paper, we leverage the potential of dynamical models to extract knowledge from the ambient measurement space under unfavorable and ideal network conditions. A rigorous assessment of various network configurations based on empirical evaluations reveals meaningful performance trends for model fitting and recovery of incomplete trajectories. In particular, while the imperfect connectivity may curtail the ability of dynamical models to exploit the intrinsic spatio-temporal structure, measurement imputation under ideal network configuration exhibits favorable performance. Our research outcomes highlight the impact of the underlying wireless connectivity on measurement data acquisition and handling in distribution grids.