To assess the current risk degree and predict the future risk degree of vessel traffic, a novel method is put forward in this study. Different from the existing literature, the available evidence of vessel traffic is directly transformed into the weighted basic probabilistic assignment (BPA) based on the optimal solution to the intersection of fuzzy membership functions in the framework of D-S evidence theory. The matrix deformation algorithm towards the combination rule makes the time complexity low in the process of the risk degree assessment. With respect to the risk degree prediction, the required Sigma points are effectively extracted. We derive the adaptive filtering gain that is suitable for the rapidly changing BPA. Finally, the experiments of vessel traffic in the Dalin Bay are made to indicate performance of the proposed method.

为了评估当前的风险程度并预测船舶交通的未来风险程度，本研究提出了一种新的方法。与现有文献不同，在DS证据理论的框架下，基于模糊隶属度函数的交集的最优解，将船舶交通的可用证据直接转化为加权基本概率分配（BPA）。针对组合规则的矩阵变形算法使得风险度评估过程中的时间复杂度较低。对于风险度预测，有效提取了所需的Sigma点。我们推导出适用于快速变化的 BPA 的自适应滤波增益。最后，大林湾船舶通行实验 被用来表明所提出的方法的性能。