Abstract: To investigate the patterns of ship detention defects in Port State Control (PSC) inspections and reduce the probability of ship detention, this paper constructs an association rule analysis model for ship detention defects based on an improved Apriori algorithm, considering the characteristics of PSC inspections and ship attributes. Using the PSC inspection data of detained ships from the Tokyo Memorandum of Understanding (Tokyo MOU) between 2018 and 2023 as the research sample, the study first processes the ship attribute data by applying association rule mining technology to perform dimensionality reduction and discrete standardization, thereby forming effective samples for analysis. An improved Apriori algorithm, which incorporates constraints on both antecedent and consequent and introduces the evaluation metric of confidence boost, is employed to conduct an in-depth mining analysis of ship detention defects. Empirical results demonstrate that the improved Apriori algorithm can efficiently screen out significant strong association rules and accurately reveal the patterns of ship detention defects. These patterns can provide an important basis for ship safety risk management and help enhance navigation safety.

Key words: ship detention defects, association rules, Apriori algorithm, confidence boost

摘要： 为研究港口国监督（Port State Control,PSC）检查中船舶滞留缺陷的规律,降低船舶在PSC检查中的滞留概率,该文根据PSC检查和船舶属性特点,基于改进Apriori算法构建船舶滞留缺陷的关联规则分析模型。以东京备忘录中2018—2023年PSC检查船舶滞留数据为研究对象,首先从船舶属性数据入手,运用关联规则挖掘技术对船舶数据进行降维和离散标准化处理,形成有效分析样本;采用基于前后项约束并引入置信增强度评价指标的改进Apriori算法,从而实现船舶滞留缺陷的深度挖掘与分析。实证表明：改进的Apriori算法能够高效筛选出具有显著意义的强关联规则,准确揭示船舶滞留缺陷规律;该规律可为船舶的安全风险管理提供重要依据,有助于提升船舶航行安全。

关键词: 船舶滞留缺陷, 关联规则, Apriori算法, 置信增强度