Using Artificial Intelligence To Detect Oscillations In Power Grids

Morgan Holzman*, Chase Holley*, Vishal Verma, and Jignesh Solanki
Lane Department of Computer Science and Electrical Engineering, West Virginia University, Morgantown WV 26505

Presentation No.: 57

Assigned Category (Presentation Format): Engineering (Poster Presentations)

Student’s Major: Mechanical Engineering

Power oscillations are a phenomenon that occur in power systems that are potentially disruptive. These oscillations cause disturbances in the delivery of power to consumers. The development of artificial intelligence provides a potential answer to the problem that plagues power systems. To accomplish this, it needs to be determined whether an AI neural network is a feasible answer to this problem and if so, what scale will this be possible to be applied to? The baseline test to prove this concept will be done using data from the 9-bus test case by utilizing the Matpower software. This data will then be used to train an AI created using TensorFlow. This AI will be tested for accuracy in determining which bus there is a disturbance at, by being able to recognize which voltage outputs match certain inputs. This research is ongoing, but an accuracy above 80% would be considered acceptable for a standard AI neural network. The goal of system like this is to prevent future large scale blackouts, like the one that occurred in Texas.

Funding:

Program/mechanism supporting research/creative efforts: WVU's SURE program (Rita Rio & Michelle Richards-Babb)