Performing a Virtual Field Test of a New Monitoring Method for Smart Power Grids

Abstract

This paper presents a virtual field test to evaluate the performance of a new grid monitoring method using artificial neural networks (ANN) under realistic operating conditions. The ANN monitoring method is able to accurately estimate voltage magnitudes of distribution grids with high penetration of distributed generators without the need for a redundant amount of measurements. The simulation framework OpSim allows a logical separation of a grid simulator and a simple distribution grid control center to create a realistic testing environment. A CIGRE benchmark grid with diverse distributed energy resources and corresponding time series is used in a grid simulator. Measurements are derived from the simulator and sent to the control center via the simulation message bus. The ANN monitoring method makes use of the measurements to estimate all bus voltage magnitudes. These can, in turn, be used by a transformer tap controller to control the overall voltage profile of the grid to stay within desired limits. Transformer tap set points are returned to the grid simulator if limits are violated. The performance of both the monitoring method and its impact on the tap controller are evaluated under normal operation, bad data and delayed measurement cases. Results show that the ANN monitoring method works reliably and accurately.