2026 Event Schedule

Extreme weather events are becoming more frequent and severe, posing substantial challenges to electric utilities in maintaining reliable services. The growing frequency of storms highlights the urgency for advanced technological solutions that enhance outage management and accelerate restoration efforts. Power disruptions not only inconvenience customers but also pose risks to public safety and economic stability. There the ability to predict storm-related outages with precision is critical to minimizing damage, strengthening grid resilience, and ensuring swift recovery.

The integration of in-field and AI-driven technologies is revolutionizing the way utilities respond to outages. By leveraging cutting-edge predictive analytics, utilities can anticipate storm impacts more accurately, strategically allocate resources, and optimize crew deployments for efficient restoration. Moreover, AI-powered systems enable better communication with customers and stakeholders, keeping them informed and prepared. However, challenges persist, including ensuring high-quality data, integration of new technologies with existing infrastructure, refining upstream processes, and managing organizational change. Addressing these complexities is essential for maximizing the benefits of modern outage management solutions.

As the demands on the power grid continue to evolve, utilities must embrace innovative approaches to strengthen their ability to withstand and recover from disruptions. By harnessing AI and other emerging technologies, they can build a more resilient, adaptive, and customer-centric outage management framework – ensuring communities remain powered even in the face of extreme weather events.

In this panel industry experts from Oncor, Eversource, and Exelon will explore how cutting-edge AI and machine learning technologies are being implemented in real-world outage management and restoration efforts across major US utilities. The discussion will focus on the challenges associated with AI models, data integration, and platform scalability, offering insights from diverse fields such as data science and engineering, electrical grid, and IT and cloud.

Oncor Electric Delivery Company has made significant strides in improving its storm response capabilities. Historically, visibility during large-scale storms was limited to what could be gathered from SCADA systems, meters, and operator reports. However, a concerted effort by various teams has led to the development of more sophisticated dashboards and reports. The data science and engineering department has played a crucial role in this transformation. By analyzing data from 4.1 million meters, these teams have gained valuable insights into storm patterns and their impact on the grid. This information has enabled Oncor to make proactive operational decisions, such as adjusting substation transformer capacities, feeder settings, and switch configurations to prevent overload during extreme weather events. Another significant development is the GIS department's overhaul. By leveraging data analysis techniques, these teams have improved the accuracy and completeness of the company's geographic information system (GIS). This has enabled operators to work with a more up-to-date and accurate map, which in turn has reduced mapping and connectivity issues. The integration of GIS and Engineering departments has also become more seamless. By utilizing large datasets, Oncor is now able to clean up and update its maps more efficiently. This collaboration has resulted in a more efficient storm response process, where operators can quickly identify areas of concern and take corrective action.

Exelon Corporation is leading a transformative effort to elevate storm-restoration operations with advanced artificial intelligence, centered on two core solutions: the Outage Prediction Model (OPM) and the Estimated Time of Restoration tool (POSEIDON). The Outage Prediction Model (OPM) is Exelon’s AI-powered platform for forecasting storm-related outages across its operating companies. It integrates weather data, historical outage patterns, and AI to predict outage counts and locations, enabling proactive resource allocation and operational planning. OPM includes storm-specific models—such as thunderstorm and rain-wind predictors—and is continuously refined through validation exercises, automated performance reports, and subject matter expert reviews. Its goal is to improve accuracy, reduce operational costs, and enhance grid resilience.

The POSEIDON is Exelon’s AI-powered platform for forecasting Estimated time of restoration at a more granular depth to meet the evolving needs of our customer needs especially during major storms. POSEIDON predicts ETRs at customer-specific level with high accuracy and speed compared to existing ETR tools and processes. The goal of POSEIDON is to meet customer communication requirements during storms within the operational constraints and challenges inherent to major restoration efforts.

Eversource Energy utilizes accurate and up-to-date Geographic Information System (GIS) records are vital for the safe and efficient operation of electric distribution networks. Traditional GIS verification methods are labor-intensive and prone to inconsistencies due to manual data collection. To address these challenges, the Eversource team have explored a pilot project of the use of computer vision techniques for automated asset inspection and GIS verification. Leveraging video footage captured via vehicle-mounted dashcams, frames were extracted and processed using AI-based object detection models trained to identify utility poles and associated assets such as transformers, crossarms, and insulators. The AI models demonstrated high detection accuracy, achieving over 90% precision across key asset categories. These detected features were geolocated using synchronized GPS data, allowing for direct cross-comparison with existing GIS records. The process enabled the identification of previously unmapped poles and discrepancies in asset records, which were subsequently used to update and enhance GIS accuracy. In addition to improving data fidelity, the computer vision approach reduced manual effort and inspection time, showcasing its potential for scalable, real-time infrastructure validation. This advancement positions AI-driven computer vision as a transformative tool in utility asset management and digital infrastructure modernization.

In summary, the panel will highlight the opportunities and obstacles in leveraging AI for outage management, providing valuable takeaways on how utilities could enhance their resilience and efficiency.

• The use of AI-generated ensemble weather predictions to improve storm forecasting and response.
• Application of outage management system (OMS) to streamline restoration operations.
• The development and refinement of Estimated Time to Restoration models to enhance recovery timelines.
• Optimization of Geographical Information System (GIS) data pipelines for better decision-making and resource allocation.