Model Investigation for Determining Danger Level of Observable Sea Craft in Heavy Traffic Zone

Abstract

A human factor aspect accruing in situations of a person s interaction with sophisticated engineering systems is considered to be a reason for most sea incidents and accidents. Even the most skilled experts sometimes happen to make wrong decisions even in their simple work conditions, treating them as correct or the most appropriate ones. Automatic or automated systems that assist a navigator to determine the best algorithm for solving different problems, especially those that are to be solved emergently are potentially able to mitigate the risks caused by human factor aspect in making decisions. This paper presents the findings of studying two models designed to determine the danger level of the observable sea craft (target) that may appear in a heavy traffic zone. They are a fuzzy model and the neural-fuzzy network (NFN) model. The authors define the main stages of modeling, for example, linguistic variables, an algorithm for developing a fuzzy rule base for a fuzzy model, and the test results proved by 525 test samples. The flow chart is given for the neural-fuzzy network. The NFN rule base is formalized. The authors simulated 192 different NFN, the generation of which was performed by the lattice method without clustering, and 288 NFN, where the generation of neural-fuzzy networks was performed by the subclustering method. The test results of the neural-fuzzy model are presented both in pictures and charts).