基于GA-ELM的水上交通事故严重程度影响因素识别研究  

Research on Identification Influence Factors of Waterway Traffic Accident Severity Based on Genetic Algorithm-Extreme Learning Machine

作　　者：张丽丽[1] 吕靖[1] 艾云飞[2] ZHANG Li-li;LU Jing;AI Yun-fei(School of Transportation Management,Dalian Maritime University,Dalian 116026,China;China Trans-port Telecommunications ＆ Information Center,Beifing 100000,China)

机构地区：[1]大连海事大学交通运输管理学院,辽宁大连116026 [2]中国交通通信信息中心,北京100000

出　　处：《运筹与管理》2018年第9期105-111,共7页Operations Research and Management Science

基　　金：国家自然科学基金资助项目(71473023);中央高校基本科研业务费专项资金资助项目(3132015068)

摘　　要：水上交通事故严重程度影响因素的识别对从根本上减少严重事故件数、降低事故危害和损失具有重要意义。在历史事故报告的基础上,构建并量化事故影响因素集,提出以极限学习机(ELM)为一般事故、严重事故的二分类器,以遗传算法(GA)为因素搜索算法的GA-ELM因素识别模型。对发生在我国水域的737件水上交通事故进行实证研究,并与以支持向量机(SVM)为分类器的GA-SVM模型进行对比分析。结果表明,GA-ELM模型识别出时段、人为致因、环境致因等9个事故严重程度影响因素,较GA-SVM模型结果更为精简,且分类精度较不做因素识别时分别提高8. 2%、7. 1%。此外,GA-ELM大大缩短运算时间。由此可见,GA-ELM可为水上交通事故严重程度影响因素识别提供一个较好的方法。Identifying influence factors of waterway traffic accident severity is of great significance in reducing number of serious accidents and reducing the hazards and loss fundamentally. In order to identify the influence factors, it first constructs and quantifies the accidents influence factors system on the basis of historical waterway traffic accidents reports. Then it presents a GA-ELM model using extreme learning machine as classifier of generous accidents and serious accidents, and using genetic algorithm as search tool. Finally, it carries out an empirical study using 737 waterway traffic accidents dates happened in Chinese waters with both GA-ELM and GA-SVM. The results show that GA-ELM identifies 9 influence factors of waterway traffic accidents severity which are leaner than the results of GA-SVM. In the meantime, GA - ELM arid GA-SVM respectively improves the classification accuracy by 8.2% 、7.1% compared with the results using all influence factors. Besides, the former operation time is much shorter. Thus it can be seen, GA-ELM model can be well used in identifying influ- ence factors of waterway traffic accident severity.

关 键 词：水路运输 因素识别 极限学习机 事故严重程度 遗传算法 

分 类 号：U698.6[交通运输工程—港口、海岸及近海工程]