Predicting Distribution Transformer Failure with Smart Meter Data

This project was a client based project for an electric distribution utility. The goal was to predict equipment failure, particularly for transformers, using data provided from the new smart meters at their customer’s premises. These smart meters are relatively new to the industry and there was nothing published in any academic journals which aimed to do something similar. The approach taken was to estimate the impedance of the transformers from the smart meter data and grid connectivity. Then use the trends found in that impedance over the previous 90 days to predict whether the transformer would fail in the next six weeks.