Engineering

Geotechnical engineering

Bayesian network

Marine engineering

Computer security

Probabilistic logic

Risk assessment

Data mining

Computer science

Collision

Risk analysis (engineering)

Business

Mechanical engineering

Automatic Identification System

Submarine pipeline

Environmental science

Artificial intelligence

Turbine

Wind power

Medicine

Offshore wind power

Electrical engineering

Reliability Engineering & System Safety

Offshore wind farms (OWFs) are relatively new installations at sea. Accident records related to vessel collisions with OWFs are insufficient to support a full quantitative risk analysis using traditional probabilistic approaches. This paper aims to develop a semi-qualitative risk model to assess the vessel-turbine collision risks by incorporating Bayesian networks (BN) with evidential reasoning (ER) approaches. First, a BN is trained based on Automatic Identification Systems (AIS) data to characterise real vessel traffic flows, including the detailed information and relationships between traffic flow parameters. Secondly, through synthesising expert judgements by ER, five risk factors influencing the probability and consequence of vessel-turbine collisions are identified (incl. the associated conditional probabilities) in the established BN. Finally, the updated BN with ER input is tested through ten real scenarios and validated by processing a validity framework. This paper pioneers the use of multi-data-driven BNs to characterise traffic flows and assess vessel-turbine collision risk for navigational safety assurance near OWFs. The research findings provide empirical evidence of using ER to supplement BN subjective data to advance its applications in risk analysis.

Realising advanced risk assessment of vessel traffic flows near offshore wind farms