Advanced Learning for Energy Risk Tracking (ALERT) – Mitigating Outage and Minimizing Catastrophes
Project Abstract
This NSF S&CC project aims to predict the State of Risk (SoR) of electricity outage occurrence and develop risk management and mitigation strategies to minimize the impact of outages. Currently, electric utilities are only able to reactively respond to outages. Consumer are left in a passive role of struggling to cope with the consequence without a preemptive option to manage the outage impacts. The project brings a transformative change that will allow utilities to predict outages, and then provide consumers with both individual and community mitigation measures. This will be achieved by increasing the S&CC awareness of how to deal with the outage impacts equitably and effectively. We will deploy advanced data analytics to train machine learning outage prediction algorithms using weather and historical outage data. The intellectual merits of the project include new risk prediction approaches, study of behavioral aspects of the outage prediction, and experiments that measure the effectiveness of predictive alert messages. Broader impacts include education and outreach efforts across PreK-20 students, their teachers and parents, through public services of museums and libraries by all-inclusive age-appropriate STEM programming. We will communicate with the broader community of citizens through the invited talks and videos at appropriate city offices in San Antonio, and at the headquarters of one of the major retail providers of electricity in Philadelphia. The emphasis on inclusive workforce development is broadly applicable and highly impactful to advance the S&CC human resource needs and resilience plans. To achieve the spatiotemporal prediction of the SoR, ALERT will perform integrative research by merging methodologies from several disciplines:
a) Advanced Data Analytics (ADA);
b) Social, Behavioral, and Economic Sciences (SBE); and
c) Smart Grid Fundamentals (SGF).
The project activity will integrate ADA and SGF data and physical power system models, respectively, and then design SBE interventions based on the survey and experimental data to define SoR models and make mitigation decisions to reduce outage risk. The innovation is in the physics-constrained and structured learning based prediction of the SoR using big data.