Presented By O'Reilly and Cloudera
Make Data Work
September 26–27, 2016: Training
September 27–29, 2016: Tutorials & Conference
New York, NY

Using the explosion of data in the utility industry to prevent explosions in utility infrastructure

Kim Montgomery (GridCure)
4:35pm–5:15pm Thursday, 09/29/2016
Data-driven business
Location: 1 E 15/1 E 16 Level: Beginner
Tags: iot, energy
Average rating: **...
(2.80, 5 ratings)

What you'll learn

  • Understand how utilities can reduce the number and severity of outages by applying machine-learning methods to predict which equipment is most likely to fail
  • Description

    The electrical utility industry, an industry accustomed to gathering customer usage data on a monthly basis, now has access to a regular stream of data from smart meters and other smart sensors. Analyzing these new streams of data has given utilities the opportunity to understand their customer usage patterns, perform preventative maintenance, detect fraud, exercise demand management, and allocate resources more effectively.

    Outages cost US businesses up to $150 billion a year. Due to aging infrastructure, the number of outages in the US has increased 285% since 1984. Utilities need improved data-driven methods for determining which infrastructure is most critically in need of replacement. Improving maintenance procedures for key pieces of equipment such as transformers, feeder cables, and reclosers can substantially reduce the risk of an outage. Kim Montgomery discusses some ways that analysis of smart grid sensor data can lead to better methods for replacing equipment before catastrophic failures occur.

    Photo of Kim Montgomery

    Kim Montgomery

    GridCure

    Kim Montgomery is the head of analytics at GridCure, where she works on predictive modeling for the utility industry. Kim has a broad applied mathematics background with expertise in both predictive modeling and differential equations. Previously, as a postdoctoral scholar at the University of Utah and as a visiting professor at the Rose Hulman Institute of Technology, she did mathematical biology research and taught applied mathematics. Her research has included using feedback control to stabilize solutions to differential equations, modeling hair cells in the inner ear, and studying signaling between retinal cells during development. She has completed more than 30 predictive modeling projects through Kaggle.com on topics such as predicting which used cars would be bad buys, predicting the jobs that would most interest a job seeker, and predicting the composition of soil from its spectral properties. She has been ranked 15th on Kaggle. She holds a PhD in applied mathematics from Northwestern University.