決定木最適化法による避航操船に関する知識構造

抄録

There are many works on collision avoidance maneuver based on analyses of marine traffic survey, simulator studies and/or theoretical approaches. The greater part of them are mainly focused on the effect of evasive action after making decision on the safety of the maneuver. In this paper, authors made rules on decision making process of risk judgment by navigators at various encounters using optimizing theory of a decision tree, which was one of inductive approaches to make knowledge structure. Subjective assessment were expressed with objective data on encounter such as relative distance and relative change of bearing etc., in a form of production rules, so called if〜then rule. Results indicated that inductive approach based on many actual observed data was very effective to make rules on collision avoidance maneuver. These rules would be easily installed on supporting or training system of collision avoidance maneuver.