Risk-Based Game Modelling for Port State Control Inspections
This thesis aims to develop a new way for port authorities to predict, analyse and make decisions in Port State Control (PSC) inspections. Under the New Inspection Regime (NIR), it is necessary to not only figure out the influence of new regime to the PSC system, but also provide some technical tool...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://researchonline.ljmu.ac.uk/id/eprint/9472/ https://researchonline.ljmu.ac.uk/id/eprint/9472/1/2018ZhisenYangphd.pdf |
| Summary: | This thesis aims to develop a new way for port authorities to predict, analyse and make decisions in Port State Control (PSC) inspections. Under the New Inspection Regime (NIR), it is necessary to not only figure out the influence of new regime to the PSC system, but also provide some technical tools capable of predicting the inspection results and supporting the decision-making of port authorities when regulating the inspection policy. The study consists of analysis from multiple perspectives, both qualitative and quantitative. The risk factors influencing the inspection results and the decision-making of port authorities under NIR are identified through the practical inspection records and related literature. The Paris Memorandum of Understanding (MoU) offers the historical inspection records within the region of Europe and the North Atlantic basin, reflecting different conditions in different periods. Given the different inspection system since 2011, port authorities require a brand new perception of the new inspection regime to estimate the inspection results, and further make decisions when making their own inspection policy. To achieve the objective, an incorporation of two types of models that have proved popular and superior is applied in this study. One is the risk assessment model of Bayesian network (BN), the other is the decision-making model of game theory. The BN models in this research utilize a data-driven approach called Tree Augmented Naïve (TAN) learning to derive the structure of the models. Based on the inspection reports collected from Paris MoU, two BNs that represent the situations of Paris MoU inspection system in different periods are constructed. Company performance, the new indicator, is viewed as one of the important factors influencing the inspection results for the first time and considered in the models. The BN model after the implementation of NIR can serve as the prediction tool for estimating inspection results under dynamic situations. Additionally, a comparative analysis ... |
|---|