Artificial Intelligence Powered Non-Destructive Testing for Inline Inspection

Abstract In this work, we present a novel AI-powered non-destructive testing (AI-NDT) technique for the examination and diagnostic of pipelines. This digital transformation development targets the detection and identification of degradation in metallic pipelines. It stems from various business needs to address an increasingly urgent challenge in the Oil and Gas sector and other related industries for the characterization of energy pipelines. This effort employs cutting-edge and bespoke AI models applied to hybrid data-physics modelling and acquisition techniques coupled with advanced sensing methods. The system interrogates the structural integrity of pipes via novel in-pipe advanced electromagnetic based sensing techniques. The AI inversion models learn on the hybrid data-physics modelling with high fidelity simulation techniques. The recorded data are fed into the AI model that directly maps them into images to represent the health status of inspected pipelines. Notably, not only this IR4.0 driven system allows for the examination of pipelines for their integrity, it also provides a direct mapping of the recorded spatiotemporal sensors’ responses to images that represent the cross-sectional profiles of the inspected pipelines along their entirety. The outcome is a system that equips end-users with effective and easy-to-use inspection methods to determine the health status of targeted assets; leading to a precise, accurate, and near real-time assessment, a first of its kind.