Utilizing Artificial Intelligence To Detect & Estimate Wear Of C-hooks
One major public utility in the Southeast embarked on an advanced analytics journey to enhance system reliability, reduce costs, and improve safety.
In the wake of the 2018 Camp wildfire in the US, when investigators pegged partial responsibility on a broken C-hook component of transmission infrastructure, a utility wanted to explore the capabilities of AI to detect C-hooks and estimate C-hooks wear.
Grid Vision solution was used to upload, store, visualize, and manually inspect images. There were a total of 1,332 digital images captured for 3 types of connector hooks, including C-hook rings, C-hook metal plates, and pylon top shackles.
eSmart Systems worked with its partner Microsoft to augment this data set with an additional 17,000 synthetic images of C-hooks to help train and improve the models. The algorithms were properly trained to identify C-hook components and accurately estimate the percentage of wear for each C-hook component.
The use of Grid Vision resulted in the analysis of thousands of high-quality digital images provided by the utility. Grid Vision demonstrated success detecting wear on C-hooks on the transmission towers. This project created the opportunity for the utility company to:
• Increase reliability in major assets.
• Reduce their O&M spending.
• Mitigate the risk of catastrophic events.