Carnegie Mellon Researchers Receive NVIDIA Academic Partnership Gift-Silicon Valley Campus - Carnegie Mellon University

Carnegie Mellon Researchers Receive NVIDIA Academic Partnership Gift-Silicon Valley Campus - Carnegie Mellon University

Carnegie Mellon Researchers Receive NVIDIA Academic Partnership Gift

Ole J. Mengshoel, a Senior Systems Scientist with Carnegie Mellon Silicon Valley at the NASA Ames Research Center, and Jike Chong, an adjunct professor at Carnegie Mellon Silicon Valley campus, have received an NVIDIA Academic Partnership gift of two Tesla C2070 supercomputing GPUs for research in “Smart Grid Fault Detection and Diagnosis Applications.” Together, the two GPUs provide over 2 Teraflops of computation capability and have a retail value of $8,000.

Currently, cars, trains and aerospace vehicles, as well as buildings, towns and cities, are undergoing major electrification transformations. Both the scope and nature of electrical power generation, storage, and distribution are rapidly changing. In particular, there is the rapid move towards smart grid based infrastructures. The successful deployment and robust operation of smart grids in vehicles and buildings will require scalable fault detection and diagnosis techniques, as a slow response to power-grid faults could lead to blackouts and other problems in mission-critical infrastructures. With the move towards distributed power generation, the unpredictability introduced by solar and wind based technologies creates additional failure modes, making power grid fault detection and diagnosis an essential component of smart grid deployment.

The Tesla C2070 supercomputing GPUs will be used to better understand fault detection and diagnostic techniques in smart-grids. Faults are detected by performing probabilistic inference based on Bayesian networks representing the state of the smart grid with observable and hidden variables.

For more on this research, please contact Mengshoel (ole.mengshoel@sv.cmu.edu) and Chong (jike.chong@sv.cmu.edu).