Visual Intelligence Reduces Risk of Power Outages

Visual Intelligence Reduces Risk of Power Outages
(Jacob Ford/Odessa American via AP)
X
Story Stream
recent articles

A handful of recent power outages around the country have prompted serious evaluation within the utility industry. Most recently, the Texas deep freeze of February 2021 highlighted the risks associated with extreme weather events and, in combination with several other variables, a vulnerable, under-supplied power grid. Electricity price cap changes, weatherization of generating units, and other systemic market and policy changes are being debated to address and minimize the potential for these types of outages.

Or go further back in time. The Northeast Blackout of 2003 prompted remedial industry actions, including better interconnection rules, enhanced security, minimum standards for reliability, and greater use of sensors. These changes have demonstrably increased the reliability of the nation’s power grid.

But unless we bury all our transmission and distribution lines – which would have economic and technical challenges – we’ll always be somewhat vulnerable to the triggering event of that cascading outage: a power line in Ohio sagging and contacting overgrown trees. And although FERC (Federal Regulatory Energy Commission) implemented mandatory standards to minimize vegetation encroachment, utilities now need to balance the risk of significant non-compliance penalties against the not-inconsequential costs associated with vegetation management.

Enter artificial intelligence. Back in 2003, vegetation management consisted primarily of utility crews trimming trees based on time-based maintenance intervals, supplemented with specific reports of encroaching vegetation. Fast forward to today, and the emergence of artificial intelligence – visual artificial intelligence in particular – enables utilities to target vegetation management to high-risk areas while minimizing operations and maintenance costs.

Utilities collect huge amounts of data from visual inspections conducted by satellites, helicopters, drones, and land-based surveys. This data can be integrated with computational engines to:

  • Predict weather patterns and their impact on vegetation growth
  • Identify wind corridors and real-time fire risks in conjunction with multi-species vegetation growth rates
  • Evaluate the impact of trees with dynamic growth rates (e.g., bamboo) and growth patterns (e.g., crown shape) and incorporate into a predictive methodology
  • Assess the value of vegetation growth retardants
  • Expand assessment capabilities beyond trees and vegetation to include buildings and other fixed structures

But data is little more than numbers on a page without direction or actionable insight. We sought to take data collected by utilities and turn it into a software solution to bring cost savings to our utility customers and increased energy reliability to consumers.

GE Digital’s Visual Intelligence Platform (VIP) for grid T&D offers an integrated data-driven approach to vegetation management, asset inspection, and asset inventory using visual intelligence. Data analysis results can be used by utilities to analyze asset conditions and develop targeted action plans. Significantly, because the vegetation management and asset inspection tools reside on the same platform and share a common data set, users benefit from a 90% data processing productivity gain over other applications.

The Visual Intelligence Platform contains four key deliverables: 

  1. Machine learning automates vegetation growth calculations and 2D/3D dangerous tree maps and then produces risk-based trimming plans that specify trim prioritization, trim volume calculation, and ground clearance. The report is formatted ready for use by the trim crews.
  2. 3D georeferenced files for each phase of the power lines. The files may then be used for sway, swag, ground clearance analysis, and planning and maintenance activities.
  3. Time-series and multivariable tree growth modeling, which enables accurate vegetation and tree growth predictions. 
  4. A flexible reporting engine that supports data integration with other applications. 

The vegetation management portion of the VIP application has been benchmarked against current vegetation management processes to quantify its value. The results show that the solution enables users to reduce tree-caused outages by up to 30% and O&M trimming costs by up to 20%, while reducing liability exposure, operating risk, and data processing time and cost.

Visual intelligence can also support asset inspection and storm recovery. For example, the AI engine can detect defects in the T&D infrastructure, such as broken or missing equipment and rust; and embedded mapping of thermal images can identify defects at grid-scale solar generation sites. These capabilities help avoid efficiency losses and downtime. 

The asset inspection feature is also helpful for rapidly assessing storm damage. Change detection analytics identify anomalies between baseline data and data collected immediately after a storm by leveraging the user’s geo-data archives. This automated approach allows utilities to better position repair crews for faster and safer restoration.

As utilities grapple with the best ways to incorporate advanced data analytics into their business models, identifying use cases with demonstrable benefits to both the utility and the customer is essential. Vegetation management and asset inspection using visual intelligence can enhance reliability at lower cost and can also reduce the damaging reputational risk that accompanies power outages.

 

Brian Hoff is a senior executive for GE Digital’s Grid Innovation Solutions group. Driven by Artificial Intelligence, GE Digital’s Visual Intelligence Platform mitigates threats and strengthens the Grid with AI based vegetation and asset inspection programs. 



Comment
Show comments Hide Comments