﻿ 一种实时船舶交通事故识别方法
 舰船科学技术  2022, Vol. 44 Issue (23): 66-69    DOI: 10.3404/j.issn.1672-7649.2022.23.013 PDF

A identification method on real-time ship traffic accident
DING Zhen-guo, ZHANG Shu-kui
Jiangsu Maritime Institute Navigation College, Nanjing 211170, China
Abstract: In order to improve the ability of real-time prediction of ship traffic accidents, support vector machine (SVM) prediction model is built based on SVM technology. The historical accident data and non accident data are divided into training set samples and test set samples according to the ratio of 7∶ 3 to train the model and test the prediction accuracy respectively. The results show that the accuracy of accident classification is 82.13%, and the overall accuracy of classification is 80.34%. Compared with the classification results of other prediction models, it is found that although the accuracy of non accident classification of the SVM model is slightly lower, the accuracy of accident classification is significantly higher than that of other models. The example shows that the SVM model is effective in accident prediction.
Key words: waterway     traffic accident     SVM     classification
0 引　言

1 支持向量机模型

 $W \bullet X + {{b}}$

 $\min \frac{1}{2}{W^{\text{T}}}W + C\sum\limits_{i = 1}^N {{\mu _i}} x$

 ${y_k}({W^{\text{T}}}\omega ({x_i}) + {{b}}) \geqslant 1 - {\mu _i}，$ (1)

 $f(x)={\displaystyle \sum _{i=1}^{N}({\theta }_{i}}-{\theta }_{i}^{\ast })h({x}_{i}·{x}_{j})+{b} ，$ (2)
 $h({x}_{i}·{x}_{j})=\omega ({x}_{i})·\omega ({x}_{j})。$ (3)

 ${h}_{\text{line}}({x}_{i}·{x}_{j})={x}_{i}^{\text{T}}{x}_{j} ，$ (4)
 ${h}_{\text{nonline}}({x}_{i}·{x}_{j})={\left[\eta ({x}_{i}·{x}_{j})+1\right]}^{p}，$ (5)
 ${h}_{\text{high}}({x}_{i}·{x}_{j})=\mathrm{exp}(-\eta {\Vert {x}_{i}-{x}_{j}\Vert }^{2}) 。$ (6)

2 模型检验 2.1 数据选择

2.2 模型检验

2.3 分类结果及分析

2.4 与其他模型分类结果比较

3 结　语

1）利用支持向量机技术构建了支持向量机模型，运用长江下游某水道历史数据验证了该模型的有效性，其中事故分类正确率可达82.13%，非事故分类正确率与整体分类正确率也达到较高水平。

2）通过与BP神经网络模型、贝叶斯方法分类结果比较，支持向量机模型在事故分类正确率和整体分类正确率方面更优，但非事故分类正确率却低于两者。

 [1] PEREZ M A. SUDWEEKS J D. SEARS E. Performance of basic kinematic thresholds in the identification of crash and near-crash events within naturalistic driving data[J]. Accident Analysis & Prevention, 2017, 103: 10. [2] 王雪松, 徐晓妍. 基于自然驾驶数据的危险事件识别方法[J]. 同济大学学报(自然科学版), 2020, 48(1): 51-59. WANG Xuesong, XU Xiaoyan. Detection of Safety-critical Events Based on Naturalistic Driving Data[J]. Journal of Tongji University (Natural Science), 2020, 48(1): 51-59. DOI:10.11908/j.issn.0253-374x.19147 [3] DOZZA M, GONZALEZ N P. Recognizing safety critical events: can automatic video processing improve naturalistic data analyses?[J]. Accident Analysis & Prevention, 2013, 60: 298. [4] SHEW C, PANDE A, NUWORSOO C. Transferability and robustness of real-time freeway crash risk assessment[J]. Journal of Safety Research, 2013, 46(9): 83. [5] SUN J, SUN J. Real-time crash prediction on urban expressways Identification of key variables and a hybrid support vector machine model[J]. IET Intelligent Transport System, 2016, 10(5): 331. DOI:10.1049/iet-its.2014.0288 [6] 游锦明, 方守恩, 张兰芳, 等. 高速公路实时事故风险研判模型及可移植性[J]. 同济大学学报(自然科学版), 2019, 47(3): 347-352. YOU Jinming, FANG Shouen, ZHANG Lanfang, et al. Real-Time Crash Risk Prediction Modelsand Transferability Analysison Freeway[J]. Journal of Tongji University (Natural Science), 2019, 47(3): 347-352. DOI:10.11908/j.issn.0253-374x.2019.03.007 [7] BOYLES S. A comparison of interpolation methods for missing traffic volume data[C]// Proceedings of the 90th Annual Meeting of the Transportation Research Board. Washington DC: Transportation Research Board, 2011: 23–27. [8] 苗旭, 王忠宇, 邹亚杰, 等. 改进的固定交通检测器缺失数据综合修复方法[J]. 同济大学学报(自然科学版), 2019, 47(10): 1477-1484. MIAO Xu, WANG Zhouyu, ZOU Yajie, et al. Improved Modification Method of Missing Datafor Location-specific Detector[J]. Journal of Tongji University (Natural Science), 2019, 47(10): 1477-1484. DOI:10.11908/j.issn.0253-374x.2019.10.013 [9] ZOU Y, HUA X, ZHANG Y. Hybrid short-term freeway speed prediction methods based on periodic analysis[J]. Canadian Journal of Civil Engineering, 2015, 42(8): 570. DOI:10.1139/cjce-2014-0447