Analytics and Learning from Non-Stationary Data to Overcome Severe Weather Impact on Power Grid Infrastructure

Technology #7248

Questions about this technology? Ask a Technology Manager

Download Printable PDF

Image Gallery
Categories
Researchers
Chuanyi Ji
Faculty Inventor Profile
External Link (www.ece.gatech.edu)
Yun Wei
Inventor Profile
External Link (www.linkedin.com)
Heqing Mei
Managed By
Rene' Meadors
Marketing Associate (404)385-0434

Background: Severe weather such as hurricanes, flooding, and thunderstorms often disrupt or damage utility supply mechanisms thus causing interference or loss of utility service to customers. As a result, it is important for utility companies to have a detailed picture of the resilience of their infrastructure to exogenous disruptions, such as severe weather, and develop mitigation strategies in advance.

Technology:Chuanyi Ji, Yun Wei, and Heqing Mei from the School of Electrial and Computer Engineering at Georgia Tech have applied model-guided data analysis to identify and increase resilience of these large-scale infrastructure installations. It uses data gleaned from the analysis of non-stationary, spatiotemporal random processes to detect potential failure points in large infrastructure installations such as power grids. The technique developed in this invention can identify pertinent vulnerabilities in the infrastructure and associated services and determine the impact level that will be felt during severe events. The invention models granular data through algorithms that learn from analysis of large-scale non-stationary data sets which vary both spatially and temporarily to identify vulnerabilities, such as large scale failure points, and reveal enhancement options that would be otherwise obscured. The model uses pertinent and simple quantities of data to derive underlying processes like disruption rates, time-varying probabilities of failures, and recoveries and expected costs that will arise to restore the impacted services and repair destroyed or damaged equipment. Network-wide analysis of the large-scale data sets are then used to help identify vulnerabilities. Results from applying this technique can also help in developing triage strategies around recovery, infrastructure repair, and occurrence of severe events.

Potential Commercial Applications: This invention would be especially useful to companies that supply services to a large number of customers and have equipment installation distributed over wide areas, such as electric power companies. Oil-and-gas, water supply, road repair, wireless communication service providers, and other market segments in which severe weather can have an impact on the delivery of services to customers/users could also benefit from this technology.

Benefits / Advantages:

  • Improve infrastructure resilience using model-guided data analysis of non-stationary, large data sets
  • Predict and resolve potential system-level vulnerabilities prior to them affecting consumers or users
  • Pertinent and simple data quantities help in the identification of network-wide vulnerabilities
  • Lower costs of repair and resolution of outagesĀ