Research & Technical Content

Leveraging NLP and ML for Underground Transmission Asset Defect Categorization: A data driven approach using NLP and machine learning was applied to large, unstructured underground transmission work order records to classify maintenance and defect information more efficiently. By converting 17,605 historical records into structured categories, the models improved insights into asset health, reliability, and risk. The study achieved strong classification accuracy, with future efforts focused on expanding datasets, refining taxonomy detail, and integrating these tools into broader asset management workflows.

Underground Transmission System Risk Assessment: Due to the nature of underground transmission systems the collection of condition assessment data on underground assets is often limited. EPRI is exploring the use of readily available data sources (such as alarm data, maintenance records, and others) to in risk assessment of underground transmission systems within a utility. This site provides an overview of a utility use case where EPRI evaluated the use of alarm data to understand performance of underground transmission circuits across the system.