Proof of Concept: Utilizing Artificial Intelligence with Ground Level Imagery to Identify and Inventory Rural Electric Utility Overhead Infrastructure

The evolution of Artificial intelligence (AI) continues to positively impact utility asset management, creating value-add solutions that were not feasible until now. The case study presented in this white paper was used to define and prove out an automated, cost-effective data collection and asset database population methodology. By building on this case study, utilities can streamline the identification of third-party attachments, pole features, and design discrepancies. This pilot project integrated West Kentucky Rural Electric Cooperative Corporation (West Kentucky) assets with AI technology provided by Mesa Associates, Inc. (Mesa), to remotely generate asset management level information for the utility. Through the Tennessee Valley Public Power Association, Inc. (TVPPA) R&D program, the pilot explored the feasibility of using AI tools to gather and analyze distribution structure data. The pilot successfully identified and geo-located Rural Utilities Service (RUS) 1728F-804 C1 and C2 pole top assemblies on distribution poles, achieving a classification accuracy rate greater than $98 \%$. The results demonstrate the viability of AI for improving asset management in a variety of utility settings.