Assets management on electrical grid using Faster-RCNN

• Published in: Annals of operations research 2022, Vol.308 (1-2), p.307-320
• Main Authors: Kala, Jules Raymond, Kre, Didier Michael, Gnassou, Armelle N’Guessan, Kala, Jean Robert Kamdjoug, Akpablin, Yves Melaine Akpablin, Coulibaly, Tiorna
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Asset management; Business and Management; Combinatorics; Electric power grids; Helicopters; Neural networks; Object recognition; Operations research; Operations Research/Decision Theory; S.I.: Artificial Intelligence in Operations Management; Theory of Computation
• ISSN: 0254-5330; 1572-9338
• DOI: 10.1007/s10479-020-03650-4

Electrical utility companies around the world are keeping track of all equipment on their distribution grid, because it will help them improve the management and the quality of the services they offer to their customers. Asset management of the electric grid is usually conducted manually, which is expensive, time consuming and the results obtained are often not accurate. In this article an automated asset management system for electricity, transport infrastructures is proposed, it is based on images taken by drones and analysed by Faster Region proposal Convolutional Neural Networks (Faster-RCNN) to generate the inventory. The designs of CNN are inspired from the human brain structures, they have been applied to many fields such as object recognition and crowed counting with promising results that are proven to be better than human observer. In order to evaluate the proposed asset management approach, a sample of images was randomly selected from a given dataset, the inventory results generated by the CNN based model are accurate, faster and cheaper than the previous approach based on human observers and helicopters.