Dynamic Risk Assessment of Oil Spill Accident on Offshore Platform Based on the Bayesian Network

Oil spill accidents on offshore platforms will cause serious environmental damage and huge economic losses. In this research, the offshore platform system is divided into six subsystems: “Human,” “Production and Storage,” “Special ship,” “Operation platform,” “Environment,” and “Management.” Combined with the historical data of oil spill accidents and the cause flow chart of accidents, the dynamic risk assessment method of offshore platform systems is constructed with the Bayesian network. First, the Dempster–Shafer evidence theory method is used to obtain the prior probability value of nodes, based on using the fuzzy set method to determine the prior probability value given by experts. Second, the Noisy-Max/Min algorithm is used to determine the conditional probability, and the risk node sensitivity is determined by combining the model. Finally, the model is verified by the receiver operating characteristic curve. The results show that the area under curve of the model is 0.870, which verifies the feasibility and effectiveness of the proposed method.