This technology is a boosting-based machine learning algorithm for ranking electrical grid components according to their susceptibility to failure and their maintenance needs.
The cost of operations for maintenance and repairs of power grid components can be significantly reduced by more accurate planning and prediction of repair needs. Organizations with large networks of infrastructure components require an accurate ranking system to direct maintenance operations. Thus, there is a clear need for a system to accurately predict component failures and maintenance needs.
This technology is a Martingale boosting-based machine learning algorithm that ranks components of an electrical grid based on their susceptibility to failure. The system ranks the risks of component failures for accurate prediction of needed maintenance operations, providing an automated method that can assist power grid operators in optimizing maintenance plans and repair tasks. A feedback loop is used to evaluate the responses of the electrical distribution system to maintenance actions and a decision support application is also provided to facilitate the optimization of electrical network performance.
This technology has been successfully tested by Consolidated Edison Company of New York, Inc. through direct application to its electrical distribution system in New York City.
M06-086
Licensing Contact: Richard Nguyen