Rethinking vegetation management along powerlines with satellite imagery

Geospatial developer platform and marketplace UP42 helped AiDash, an AI-first vertical SaaS company based in the USA and India, transform vegetation management for a Fortune 500 power utility company. An AiDash customer, the utility company wanted to manage vegetation effectively along distribution lines spanning over 80,000 km.

Vegetation management remains one of the single largest line items in most power utilities’ annual operations and maintenance budgets. It exceeds $100 million annually in the US at many larger utilities.

Poor vegetation management is one of the primary reasons for frequent power outages and wildfires, with consequences costing an average of around $33 billion per year in the United States alone. Furthermore, increasing scrutiny from customers, the media, regulators and more has led utility companies to understand the risk of liabilities better.

For most power utilities, the vegetation management process is manual, routine and unoptimised. AiDash’s client faced challenges with this traditional approach. Scheduled trimming cycles spanning 4–5 years and manual data collection left the company with minimal predictive capabilities to prevent future outages, damages and costs.

Managing distribution lines and assets, spread over 80,000 km across multiple states, proved expensive, difficult and time-consuming. The lack of visibility concerning urgent situations and hazards and an inability to identify the exact point of failure resulted in reactive and ad hoc maintenance that is primarily expensive.

AiDash wanted a solution for its client that was reliable, accurate and easy to use. To solve this, the company developed an end-to-end AI proprietary model called the Intelligent Vegetation Management System (IVMS), using satellite imagery powered by UP42.

Through the platform, AiDash could easily access both ‘archive’ and ‘tasked’ very-high-resolution Airbus Pléiades satellite imagery of utility lines in more than one state in the Southern US. This 50 cm resolution multispectral imagery from the twin satellite constellation provided AiDash with detailed coverage for the customer’s remote asset monitoring needs that would have otherwise been very difficult to obtain.

By combining these images with a deep neural network model, AiDash’s system was able to learn and then predict the growth rate of the vegetation species in each feeder at a span level — ie, powerlines between two consecutive poles. Knowing these span-level predictions then led AiDash to cluster these predictions at section or sub-feeder and feeder levels.

The plug-and-play AI model for vegetation management enabled the utility company to predict the growth rate of different species along powerlines — at an individual tree level. The company was able to prepare a 3- to 5-year trim plan for the entire network with an accuracy of over 85%.

The IVMS solution ultimately enabled the utility company to improve system reliability by 15% and reduce its annual budget for vegetation management by 20%. The company can now identify risks and urgent situations in real time and plan vegetation management operations years in advance, saving costs and managing contractors through an app-based approach. Post-trim satellite image analysis can also be used to audit the compliance of vegetation clearance without the need for a supervisor on the field.

“Working with UP42 helped us with the most critical element of our AI-first solutions: data,” said Abhishek Singh, co-founder and CEO of AiDash. “Using archive satellite imagery alongside tasked imagery helps us provide precise and timely predictions for our customers, enabling them to make their infrastructure climate resilient and sustainable.”