A power grid resilience system monitors the status of a power grid and the threats to its function. These machine learning monitoring systems manage all parts of the power grid: the generation centers, the substations that adjust voltage, the transmission lines, and the distribution centers.
Also called electrical grids, these systems can cover individual buildings, cities, nations, or even more than one nation.
A good power grid resilience system makes sure that people get the power they need. It can also prepare for adverse events that result in outages by warning citizens of expected outages, prioritizing energy to emergency services, alerting utilities crew to damaged equipment for swift repair, and more.
Important internal data includes a power grid map and data on the companies and personnel that support it. Smart meter technology may also be necessary in this case, as they provide real-time data and alerts to utilities companies. Scheduled or predicted repair data would also improve the power grid resilience system.
Essential external data for this system includes vegetation, land cover, and population data; additionally important are historical and current weather and natural disaster data.
Additional data may include GPS data on utility crew vehicles so the nearest team can be routed to damaged equipment. This data can also help crew avoid routes damaged by storms or other adverse weather events.
Another potentially useful data source comes from competitor research: after all, other energy companies or municipalities may be doing something that you can replicate.
There are many challenges to a power grid resilience system, mostly related to scale and to integration. Scale is an obvious challenge but integration more tricky, especially as every power grid faces its own integration problems. In other words, one municipal power grid may have multiple energy sources—say, solar and hydroelectric—via a multitude of private and public companies and possibly a mix of old and new infrastructure.
Finally, a resilient power grid must stay secure against malicious attacks from humans.
Cornell University arXiv: Improving Power Grid Resilience Through Predictive Outage Estimation
NREL: Predicting Storm Outages Through New Representations of Weather and Vegetation
Hewlett Packard: Predicting Service Outage Using Machine Learning Techniques
“COVID-19 presents a unique problem — a human resource problem for utilities trying to send their workers into the field to perform repairs, maintenance, and other activities,” said Kandaperumal. “Since the distribution grid serves critical infrastructure that is directly tied to the health and safety of the nation and its citizens, utility workers are essential workers.”
At the student competition, Kandaperumal presented his work on supplementing the decision-making tool with an algorithm that can ensure utility workers’ occupational safety in the field. The tool considers power grid damage due to extreme events, needed repairs, crew requirements, and COVID-19 information to determine the best route the crew can take to complete the repairs while keeping themselves clear of COVID-19 hotspots.
Enerdata Energy Research & Data services provided by Enerdata supplies energy information services to businesses.
IBM PAIRS Services provides queryable geospatial and temporal data in the form of maps, satellite images, weather data, drone data, and other data.
Genscape Energy Data provides real time natural gas production forecast, customer pricing model and more across several energy markets.
IBM The Weather Company’s Global High-Resolution Atmospheric Forecasting System (IBM GRAF) provides real-time weather forecast data.
The Weather Company has partnered with the National Center for Atmospheric Research (NCAR) organization to provide the latest and most accurate weather forecasts. Additionally, in parts of the world with less specialized weather-measuring equipment, IBM GRAF uses data from airplane sensors and smartphone users to cover these areas.