The Subsea AI Body of Knowledge (SAIBOK) project, supported by the Net Zero Technology Centre (NZTC), is a collaboration between RGU’s School of Computing and the NSC alongside industry partners TotalEnergies, BP, Chevron, Intel and Xodus.
The SAIBOK project has leveraged AI and machine learning technologies to train algorithms to detect and interpret anomalies within a collection of subsea inspection images. Automation is the direction of travel for offshore surveys in the drive to reduce emissions.
The project could lead to the deployment of automated unmanned vehicles with real-time anomaly detection and interpretation capabilities driven by the SAIBOK algorithms.
Aggregating and anonymising datasets from multiple sources has created a broader training dataset for AI than what would otherwise have been possible within individual company silos.
The outcomes of this collaboration underscore the significance of data sharing, as it has the potential to greatly propel AI advancements in the energy industry and facilitate the digital transformation of current practices into more data-oriented and intelligent solutions.
Professor Eyad Elyan from RGU’s School of Computing and SAIBOK project lead said: “The performance of AI models is heavily reliant on the quality and diversity of available data. Through the SAIBOK project, we have innovatively harnessed rich and diverse subsea data from different industry partners for asset inspections, resulting in intelligent anomaly detection methods.
“This endeavour has effectively showcased the immediate necessity for data-sharing within the energy industry, presenting a unique opportunity for companies to derive significant benefits from AI.”
Professor John McCall, NSC Director said: “The SAIBOK project exemplifies what the NSC is trying to achieve with its focus on the Transparent Ocean research programme.
“We’re aiming to bring together industry datasets in a way that will accelerate our capability to understand subsea and marine activities, including supporting infrastructures and surrounding environments, using the full range of state-of-the-art platforms and sources for data acquisition, visualisation, analysis, interpretation and prediction.”
Blair O’Connor, Digital & Data Architecture Project Manager at NZTC said: “Driven by industry from the start, the SAIBOK project is an exemplar in data sharing in the offshore energy sector. We have proved that it can be done. The machine learning advances made during this project resulting from access to a bigger, multi-sourced dataset have been inspiring.
“SAIBOK is realising value for the project contributors, and the hope is that the wider AI development community will also benefit upon public release of the anonymised repository curated for this project.”
Source: DIGIT