For taking counter measures in advance to prevent accidental risks, it is of
significance to explore the causes and evolutionary mechanism of ship
collisions. This article collects 70 ship collision accidents in Zhejiang
coastal waters, where 60 cases are used for modeling while 10 cases are used for
verification (testing). By analyzing influencing factors (IFs) and causal chains
of accidents, a Bayesian network (BN) model with 19 causal nodes and 1
consequential node is constructed. Parameters of the BN model, namely the
conditional probability tables (CPTs), are determined by mathematical statistics
methods and Bayesian formulas. Regarding each testing case, the BN model’s
prediction on probability of occurrence is above 80% (approaching 100% indicates
the certainty of occurrence), which verifies the availability of the model.
Causal analysis based on the backward reasoning process shows that H (Human
error) is the main IF resulting in ship collisions. The causal chain that
maximizes the likelihood of an accident occurring is: H1 (improper lookout) → H4
(underestimation of collision) → H7 (failure of taking effective
collision-avoidance measures) → H (human error) → C (ship collision). By
implementing sensitivity analysis process, key IFs of ship collisions are found
and are ranked as: H9 (improper emergency handling), H7 (failure of taking
effective collision-avoidance measures), H6 (without using safe speed), H4
(underestimation of collision), H1 (improper lookout), H3 (nonstandard duty), H8
(failure of fulfilling “giving way” responsibility), H5 (unaware of target
ships), and H2 (crew incompetence). Among them, H9 (improper emergency handling)
and H7 (failure to take effective collision-avoidance measures) have relatively
high sensitivity and greater impact on collision accidents. Results show that
the BN model can be used to analyze the causes of ship collisions in Zhejiang
coastal waters and to predict the probability of occurrence of accidents. The
research will provide theoretical and practical support for exploring the causes
and revealing the evolutionary mechanism of accidents, and for taking targeted
risk control measures to prevent future accidents.
