航道内船舶交通事故预测模型选择及敏感性分析

Selection and sensitivity analysis of vessel traffic accidents forecasting models in waterway

为了探索航道内事故发生的关键诱因,依托长江江苏段9个水道的历史事故数据和水道交通数据,验证了航道内事故服从对数正态分布.以2009年至2014年间事故数为因变量,以船舶交通量、航道长度和设计一致性(包含航道深度、航道宽度、曲率变化率、船舶速度差及船舶速度变化率)为解释变量,建立了5个泊松对数正态事故预测模型,验证了以船舶速度变化率表征一致性的事故预测模型具有最好的拟合度.基于最优拟合度模型的敏感性分析表明:船舶速度变化率是最关键因素,其次是航道长度.基于安全考虑,船舶速度变化率应控制在20%以内,航道长度不宜大于30 km.

In order to explore the key factors causing accidents,historical accident data and waterway traffic data of 9 waterways of the Yangtze river in Jiangsu are employed,and it is verified that the accidents follow lognormal distribution. Five Poisson-lognormal accident forecasting models are established by taking the accidents from2009 to 2014 as dependent variables and vessel traffic,waterway length,and design consistency（ including waterway depth,waterway width,curvature changing rate,vessel speed difference and vessel speed changing rate）as explanatory variables. Among the five models,the one with vessel speed changing rate as design consistency measure has the best goodness-of-fit and is chosen as the optimal model. Sensitivity analysis based on the best model shows that vessel speed changing rate is the most significant factor and waterway length is the second. For safety purposes,the vessel speed changing rate should be within 20% and the waterway length should not exceed30 km.