On commission from the Norwegian Ocean Industry Authority (Havtil), DNV has prepared a report that explores how AI is used to analyse data on dropped objects. The report focuses on improving learning and insights from existing data using modern digital technologies.
Structural incidents
Follow-up of structural safety is a crucial feature of Havtil’s work. Over time, both we and the industry have paid close attention to structural incidents with
major accident potential. These are included as one of the indicators in the Trends in Risk Level (RNNP) reports. The reporting of dropped objects in which structural elements are involved is also included in RNNP, as such incidents can have fatal consequences at the individual level.
Extensive data collected
Havtil has collected extensive data on dropped-object incidents for several decades. This information is hard to access for the purposes of systematic learning. The report highlights the need to exploit new technology to make this information more accessible and useful.
DNV has used expertise in damage analysis, risk management, AI, language technology and knowledge models (ontologies) in its work to create a search platform for incident data.
Result
The study shows that the use of AI can improve the searchability and analysis of incident data. At the same time, the study emphasises that the tool still requires extensive human evaluation of the answers that the model provides. Subject experts must be involved early in the development of ontologies and knowledge graphs.
The report also contains other recommendations for how the tool can be improved.
The report is available in Norwegian only.