海员避碰行动各指标相关性的多元回归分析

在获取大量海员避碰行动样本的基础上，通过建立多元线性回归的数学模型，分析了转向时机、转向幅度、目标船舷角、会遇两船的船速比等4个指标与目标船最后通过距离之间的相关性大小，结论对海员避碰行动决策有一定的参考价值。由于样本是在对海员进行航海模拟器训练中获得的，从而说明了用航海模拟器来研究避碰行为的可行性以及STCW公约强化模拟器训练的必要性。

Multiple Regress Analysis of the Correlation of Seamen''''s Collision Avoidance Action Factors

Multivariate regression analysis is a mathematical method which calculates the correlation of multiple variable factors. The quantity samples of seamen's behavior during collision avoidance(CA) are obtained during seamen training with nautical simulator. Based on the samples, the mathematical model of multivariate linear regression is set up, and the correlation of the five factors in seamen's CA action, i.e. time and magnitude of course alternation, relative direction of target ship, velocity ratio of two ships and target ship's DCPA are analyzed in the paper. The conclusions presented are useful for seamen's CA action decision-making. As these samples are obtained in seamen training with nautical simulator, it proves the feasibility of studying CA with nautical simulator and the necessity of training with simulator which is strengthened by STCW convention.