Session: 02-06-01 Data-driven Models for Marine Structures

Submission Number: 157080

Using Artificial Intelligence to Analyze Incidents Involving Falling Outfitting Structures 

This paper presents a feasibility study to investigate the value of modern search methods, using ontologies (knowledge models), artificial intelligence (AI) and text analysis, can provide to establish insight into trends and learning for falling construction elements. As well as the value of the methods for other events.

The objective is four-fold. Firstly, to address the risk awareness associated with other parts of the structure than main structure, based on reported incidents. Secondly, report on methods for these components. Thirdly identify possible state-of-the-art search methodologies with ontologies and AI to analyze data about a given type of events, based on Havtil's incidents database. The database contains all incidents reported by the operators in Norwegian petroleum sector (onshore and offshore). The reporting criterium is falling of heavy objects that could cause a third-party damage.

Given, to the authors` knowledge, the novelty of the work we will present the process by example of a visual representation of the ontology/knowledge development with some examples of the changes over time.  The intention is give a better mutual understanding og communication between the ontologists and designers.

In addition to the findings some examples of incidents, connection methods and root causes from in depth investigations are presented.

Presenting Author: Morten Langøy Norwegian Ocean Industry Authority

Presenting Author Biography: ..