Time Window Selection via Online Risk Assessment for Power Distribution System Analysis

Electric power system modernization has been driven by the need to integrate inherently stochastic distributed resources (e.g., photovoltatics) into the grid. New infrastructure including smart meters, distribution automation devices, and two-way communications networks allow operators to monitor and control power grids in real-time. Informed operational decisions, however, require nontrivial system analysis. This paper presents a model for effectively allocating computational power for the online analysis of electric power distribution systems. A time-varying risk measure is defined as a function of forecasted injection conditions. Nonlinear optimization is used to select specific analysis times (i.e., times when the state is computed) that occur more frequently when risk is relatively greater. The selected analysis times divide an operating horizon into non-uniform time windows. Simulation results are presented for a 2556-node multi-phase distribution system.