11-02, 10:10–10:50 (America/New_York), Radio City (Room 6604)
The great majority (more than 2/3) of power outages are caused by contact from vegetation to an active power line, with the added risk of fire and public safety. The goal for utility companies is to manage the vegetation near their above-ground infrastructure in order to reduce these types of contact during inclement weather. This session will discuss an AI-driven vegetation management solution, using LiDAR and satellite imagery to determine the highest risk areas of a utility's service area. Our approach is able to give insights across a service territory, by line down to the foot level as far as how close a branch may be to a power line. This work will enable them to shift from cycle-based to condition-based trimming. We will go over the data used, the various technical challenges, and our approach to scale the solution.
This session will introduce Vegetation Management for Utility companies. It will discuss how it is currently done for almost all companies, which is cycle-based. This means they will often revisit a particular line every x years. This approach does not take into account any changes to the vegetation, whether due to growth, previous storms, etc. Using accurate Geiger-mode LiDAR data, we can produce line by line results that present the actual physical distance of any vegetation to a given line. This allows the utility to prioritize areas that have imminent issues and focus on those. The result is reducing outages, risks of fire, as well as being more efficient with their own vegetation management costs. Of particular interest will be how we deal with various challenges, such as sparsity of wire data using automated gap filling, identification of type of wire (energized vs not), and automatic detection of spans.
Topics we will cover
- Satellite Data.
- Discuss the use of Satellite data, and differences between it and LiDAR.
- Why to use LiDAR for encroachment calculation vs Satellite.
- LiDAR Data.
- How it is collected, processed and classified.
- Different types (Geiger Mode vs Linear).
- How to visualize using open source tool QGIS.
- Python packages used to process and work with this data. These include laspy, pdal and others.
- Working with raster and vector data
- Packages to use to process this data. These include GeoPandas, rasterio, and others.
- Challenges with LiDAR data.
- Size and Scale of data.
- Sparse Wire data due to various reasons. And how we use techniques to regenerate the catenary curves in these cases.
- Programmatically determining which wires are electric vs other (comms, neutral, etc).
- Scaling
- How to process TeraBytes of LiDAR data.
- How to manage the pipeline of the various steps:
- Pre-process LiDAR Data
- Automated Gap Filling and line creation for wire data.
- Calculate distances of Vegetation to power lines.
- Generate consumable insights and KPIs for a utility.
- Tree Growth prediction models.
- Tree Species prediction models.
The target audience would be those interested in weather/climate related AI solutions, GIS solutions, or those interested in working with LiDAR, especially in Python. We expect the audience to have a working knowledge of Python, and a basic understanding of imagery.
No previous knowledge expected
Kewen Gu is a Machine Learning Engineer in the Sustainability Software division of IBM, focusing on building AI solutions for businesses using weather, remote-sensing Imagery and client related data sets. He has extensive experience working with very large data sets including scaling runs using PySpark, Kubernetes, Argo. Currently, he is focused on building and deploying two main AI solutions in the IBM Environment Intelligence Suite - a) Outage Prediction which predicts where and how many outages a Utility should expect during a storm. b) Vegetation Management, which uses 3D Imagery to determine where a Utility needs to trim trees that are closest to power lines.
Anjani Prasad Atluri is a Data Scientist in the Sustainability Software division of IBM, focusing on building AI solutions for businesses using remote-sensing Imagery and client related data sets. He graduated with Masters in Data Science from Columbia University in 2022. He has extensive experience working with satellite imagery (several resolutions) and lidar imagery. Currently, he is focused on building and deploying AI solutions in IBM Environment Intelligence Suite for Vegetation Management. Vegetation Management: uses 3D Imagery to determine where a Utility needs to trim trees that are closest to powerlines.