Stochastic State Estimation for Smart Grids in the Presence of Intermittent Measurements

Future smart grids are envisioned to have significant distributed generation penetration. In this paper, we develop a dynamic nonlinear model for the power distribution networks, incorporating power flow equations along with load and distributed generation forecasts. As traditional state estimation approaches based on Weighted Least Squares (WLS) are inadequate in dynamic system models, we consider an extended Kalman filter (EKF) for state estimation. Unlike prior efforts, we analyze impact of communication network on state estimation process by considering intermittent measurements. The intermittent measurements denoted by packet drops are modeled as a Bernoulli random process. A stochastic analysis for boundedness of state estimation error is presented. The analysis establishes system conditions for which stochastic stability of state estimates can be assured. An upper bound on critical packet drop rate is derived. We also relate the bound on critical packet drop rate with randomness in load fluctuations. Finally, we verify our analysis by simulating a single phase radial distribution network model as an example.