While there is no substitute for experienced forestry professionals, mathematical models can be an effective complement, providing additional information to support tree maintenance decisions. Throughout 2021, FirstEnergy Corp. created a machine learning model to predict areas of distribution line most susceptible to tree-caused outages. The goal is to improve reliability by effectively directing maintenance resources to areas that will provide the greatest value. This session will cover the two key components of the tree outage predictive model solution:

1.  A back-end model that mathematically predicts the likelihood of a tree-caused outage for a particular location of distribution line. Model inputs and techniques will be described.
2.  A user-facing GIS visualization tool to display results in an easy to digest format. A demo/screenshot presentation will be shown. The tool has been in use for several months in a pilot program with a single Pennsylvania operating company within the FirstEnergy Corp. footprint. Starting in January 2022, the tool will be entering the A/B testing phase to measure model performance.

Attend this session and learn:

• How machine learning can be implemented to assist with forestry maintenance
• Benefits of a robust data pipeline build an effective predictive model - Advanced use of GIS software to support operations decisions