Optimizing surveillance activities with AI solutions

Learning, and Computer Vision technologies are being leveraged to enhance the maintenance and inspection of electrical transmission networks, with a particular focus on the practices, the strategy and challenges faced by the French transmission system operator.

The main goal of this paper is to explore how AI-driven solutions can facilitate more effective asset identification and anomaly detection, thereby improving the reliability and efficiency of network inspections and to describe the strategy to tend to an AI-enhanced inspection process.

To achieve these goals, the paper outlines the use of convolutional neural networks (CNNs) and other machine learning models for processing and interpreting complex visual data acquired during inspections. These approaches enable the automatic detection of faults, weak signals, and anomalies across extensive networks. The integration of AI aims not only to enhance the accuracy and speed of inspections but also to contribute to more data-driven decision-making in maintenance planning.

Key results highlighted in the paper include validation of technical feasibility of automatised data acquisition and of inspection machine learning models to detect equipment and anomalies.

The strategy can be summarised as follows and is detailed in this paper:

- - A precise business need: the starting point

Preliminary proofs of concept and validation of technical feasibility

A company strategy to scale AI assisted inspection solutions:

o Partner selection among best actors on the market o Framework definition to favour collaboration o GDPR alignment

Development of a cloud-based IT infrastructure to host the global solution