Falling trees are the main driver of power outages in major storms. For instance, more than 1 million people in New Jersey lost power during the historic 2012 Hurricane Sandy that knocked branches, limbs, and entire trunks atop delicate and crucial connections. Since then, trimming has become a major undertaking to prevent similar widespread outages.

But a Texas A&M team now says they have an even smarter way to keep the grid up and running: a big-data predictive risk model that can predict where the limbs will fall, and where the lights may go out as the next storm clouds approach. They present the methodology in the journal IEEE Transactions on Smart Grid.

The model was based off recordkeeping at the utility CenterPoint Energy, based in Houston, and which serves Texas, Arkansas, Louisiana, Minnesota, Mississippi, and Oklahoma.

The factors include the basics such as operational records and weather forecasts, but also altitude and vegetation estimates and maintenance schedules to provide a matrix of what the risks are, they report.

Also included in the details of the predictions are: smart meter data, GPS and GIS data, satellite imaging, customer call information, load forecasting, electricity market records, and asset management histories.

“The utility grids and related assets are mostly located outdoors and are exposed to all kinds of weather hazards,” said Mladen Kezunovic, a professor of electrical and computer engineering, and senior author. “Any kind of environmental data that has some relevance to the power system can be fed into this prediction framework.”

“By improving reliability, we can predict outages,” said Or-Chen Chen, a graduate student who developed the methodology with fellow graduate student Tatjana Dokic. “If we can prevent outages with historical and close-to-real-time data, we can save millions of dollars since the outages may be mitigated.”