This technology is a machine learning platform that can be used for determining the likelihood of failures in power grid components based on historical and real-time data.
Unmet Need: Proactive power grid maintenance for enhanced reliability
The aging power grid infrastructure in major US cities is responsible for frequent failures and power outages. Without a robust method to identify key vulnerabilities in the power grid, maintenance is limited to dealing with failures, resulting in higher operating costs and reduced reliability. The implementation of smart power grids enabling proactive maintenance will require systems capable of prioritizing repair tasks through the analysis of tremendous amounts of past and real-time data. As a result, a need exists for proactive and predictive maintenance programs that can utilize existing data resources for electrical grid reliability.
The Technology: Machine learning system for predicting power grid failures
This technology is a statistical machine learning system that uses historical and real-time data collected by the utility company to predict the risk of failure in the power grid. The system generates vulnerability rankings and mean time between failure (MTBF) estimates for various components of the power grid such as cables, terminators, manholes, and transformers, which can be used to prioritize maintenance work proactively before breakdowns occur. Modules of this technology can be conveniently integrated into corporate management platforms to enhance cost-benefit analyses, maintenance planning, and effort allocation.
This technology has been validated by Consolidated Edison Company of New York, Inc. through direct application to the New York City power grid.
Applications:
- Statistical model for proactive decision-making systems
- Accurate failure prediction in power grid components and other utilities
- Ranking of key vulnerabilities within a system based on historical and real-time data
- Advanced decision-making tool for utility companies
Advantages:
- Can process historical and real-time data
- Uses state-of-the-art machine learning algorithms for predicting the risk of failures and prioritizing maintenance tasks
- Is applicable to several different power grid components
- Enables proactive maintenance
- Can be integrated into corporate business management and decision-making
- Can improve the reliability of the power grid
Lead Inventor:
Roger N. Anderson, Ph.D.
Patent Information:
Patent Status
Related Publications:
Simão HP, Jeong HB, Defourny B, Powell WB, Boulanger A, Gagneja A, Wu L, Anderson RN. “A Robust Solution to the Load Curtailment Problem” IEEE Transactions on Smart Grid 2013 Sep; 4(4): 2209-2219.
Rudin C, Waltz D, Anderson RN, Boulanger A, Salleb-Aouissi A, Chow M, Dutta H, Gross PN, Huang B, Ierome S, Isaac DF, Kressner A, Passonneau RJ, Radeva A, Wu L. “Machine learning for the New York City power grid” IEEE Trans Pattern Anal Mach Intell. 2012 Feb; 34(2): 328-345.
Wu L, Teräväinen T, Kaiser G, Anderson R, Boulanger A, Rudin C. “Estimation of System Reliability Using a Semiparametric Model” IEEE 2011 EnergyTech. 2011 Jul
Wu LL, Kaiser GE, Rudin C, Waltz DL, Anderson RN, Boulanger AG, Salleb-Aouissi A, Dutta H, Pooleery M. “Evaluating Machine Learning for Improving Power Grid Reliability” Columbia University Academic Commons 2011
Wu LL, Teräväinen TK, Kaiser GE, Anderson RN, Boulanger AG, Rudin C. “Estimation of System Reliability Using a Semiparametric Model” Columbia University Academic Commons 2011
Tech Ventures Reference: