Scenario evolutionary analysis for maritime emergencies using an ensemble belief rule base. (September 2022)

• Record Type: Journal Article
• Authors: Li, Baode; Lu, Jing; Li, Jing; Zhu, Xuebin; Huang, Chuan; Su, Wan
• Is Part Of: Reliability engineering & system safety. Volume 225(2022)
• Publication Date: 2022-09
• Subjects: Maritime emergencies -- Scenario evolutionary analysis -- Ensemble belief rule base -- Random subspace -- Multiple outputs; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2022.108627
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: The scenario evolutionary analysis of maritime emergencies from the perspective of the "scenario-response" decision-making mode. An Ensemble-BRB model from random subspace development combined with mutual information is proposed. The structural form of the belief rules is extended to enable the BRB models to solve maritime emergency scenarios that generate multiple consequences. Abstract: Maritime emergencies exhibit uncertainty and complex evolution in the process of development. Scenario evolutionary analysis can identify the development of maritime emergencies, which is essential for an effective response. This paper proposes a novel ensemble belief rule base model (Ensemble-BRB) for scenario evolutionary analysis of maritime emergencies. Specifically, multiple low-dimensional random subspaces are generated randomly by combining mutual information so as to avoid combinatorial explosion, and to reduce the interference of redundant information. Subsequently, each random subspace is developed into a BRB subsystem that can be used to solve multiple-output problems, and the parameters of each BRB subsystem are learned using a differential evolutionary algorithm. Then, evidential reasoning is employed to combine the reasoning results of each BRB subsystem rule. Furthermore, the reasoning results of each BRB subsystem are combined using a cautious conjunctive rule approach to obtain the final results. The scenario evolutionary analysis of the proposed model is demonstrated and validated using maritime accidents as a case study, and the experimental results show that the proposed model can be effectively implemented. Moreover, in comparison with other well-known methods, the proposed method demonstrates good interpretability, high accuracy, and an effective solution for combinatorial explosion.