| 实时交通流条件下船舶航道航行风险预警模型 |
| |
| 引用本文： | 张树奎,肖英杰.实时交通流条件下船舶航道航行风险预警模型[J].上海海事大学学报,2015,36(4):11-15. |
| |
| 作者姓名： | 张树奎 肖英杰 |
| |
| 作者单位： | 1. 江苏海事职业技术学院 航海技术学院 2. 上海海事大学 商船学院,上海海事大学 商船学院 |
| |
| 基金项目： | 中国交通教育研究会课题（20140233）；江苏省高校哲学社会科学研究项目(2014SJD286) |
| |
| 摘 要： | 为提高航道内实时交通事故预警精度,基于长江下游两个水道内发生的交通事故数据和船舶检测器数据,运用随机森林(Random Forest,RF)模型对事故发生前20~40 min内的航道状态初始数据进行重要变量筛选,根据筛选出的4个最重要的变量结合高斯混合模型和最大期望算法构建新的事故预警贝叶斯网络(Bayesian Network,BN)模型.实例检验表明:新的BN模型优于直接利用初始数据建立的模型,事故预警正确率达到81.29%;可转移性测试中新的BN模型的事故预警正确率虽有所降低,但整体预警正确率和事故预警正确率仍高于利用初始数据建立的模型;新的BN模型用于实时船舶交通流航行风险预警是有效的.
|
| 关 键 词： | 航道 航行风险 预警 随机森林模型 贝叶斯网络 |
| 收稿时间： | 2015/4/10 0:00:00 |
| 修稿时间： | 5/3/2015 12:00:00 AM |
Navigation risk early warning model of ships based on real time traffic flow in waterway |
| |
| Institution: | Navigational Department Jiangsu Maritime Institute |
| |
| Abstract: | In order to improve early warning accuracy of real time ship traffic accidents in waterway, based on the traffic accident data and the ship detector data collected on two downstream waterways of Yangtze River, important variables are picked out by the Random Forest (RF) model from the initial data of waterways within 20 40 min before traffic accidents, and then a new Bayesian Network (BN) model on accident early warning is established with 4 most important variables combined with the Gaussian mixture model and the maximum expectation algorithm. A case shows that: a new BN model is better than the model using initial data directly, and the accident early warning correct rate is 81.29%; the transferability test shows that the overall early warning correct rate and the accident early warning correct rate of the new BN model are still higher than those of the model using initial data directly, though the accident early warning correct rate of the new BN model decreases. The new BN model is effective for navigation risk early warning based on real time traffic flow. |
| |
| Keywords: | waterway navigation risk early warning random forest model Bayesian network |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《上海海事大学学报》浏览原始摘要信息 |
| 点击此处可从《上海海事大学学报》下载免费的PDF全文 |
