Name
Innovative Use of Analytics and Machine Learning for Anomaly Detection
Track
AI & Digitization Track
Date & Time
Wednesday, October 2, 2024, 11:15 AM - 12:00 PM
Ronald Chebra Jing Yang Junhui Zhao
Description

Eversource Energy pioneered a series of significant advancements in the detection and management of leaks within High Pressure Fluid Filled (HPFF) transmission systems, marking a substantial leap forward in operational efficiency and environmental protection. The presenters in this session will discuss how they developed innovative analytics methods and implemented advanced data visualization and machine learning (ML) techniques. Using these tools and methodologies, Eversource’s engineers achieved a remarkable reduction in leak detection times—from the traditional span of two to three days to within four hours. This acceleration in detection capability demonstrates not only Eversource’s leading role in developing data-driven infrastructure management but its commitment to environmental protection and sustainability. At the heart of the achievements is the novel application of pressure variation data to calculate pump starts. They also introduced a ML-based method to dynamically adjust alert thresholds and created a powerful cross-check tool—a PowerBI dashboard that integrates tank level data, temperature readings, and load change information, providing operation engineers with a 360-degree view of system operation. Presenters in this session will provide meaningful details about this data analytics based use case.