A novel data-driven risk assessment framework for improved inspection efficiency of port state control

Accurate risk assessment of visiting vessels is of crucial importance for Port State Control (PSC) to ensure a highly-efficient inspection system. Although significant efforts are put forward, the inspection efficiency of PSC system still have large room for improvement, evident by deficiency records in major Memorandum of Understandings (MoUs). The ignorance of characteristics of deficiency types, as well as the lack of simultaneous consideration of probability and consequence, are among important factors making the assessment results unreliable. This research aims to develop a novel data-driven Bayesian network-based risk assessment framework to assist port authorities in assessing vessel risks accurately and selecting high-risk vessels for inspection efficiently. It makes new contributions by employing the new framework to investigate coastal ports located in the Greater Bay Area (GBA) for the first time. The findings reveal that the proposed risk assessment framework is a better risk classification tool and can deliver more precise results than the current ship risk profile. It is able to not only calculate the exact risk scores of visiting vessels, but more importantly distinguish their risks clearly even they are under similar conditions. Further, an improved vessel selection strategy is proposed for port authorities to ensure the accurate selection of high-risk vessels for inspection with limited resources and low costs dynamically, which is of great significance in better controlling substandard vessels with poor quality. This paper therefore provides insightful implications for practitioners to craft a highly-efficient inspection system, as well as develop a safer and more sustainable maritime transport.