基于文本挖掘的内河船舶碰撞事故致因因素分析与风险预测
|
摘要
作为一种典型的高风险海事事故,内河船舶碰撞事故致因因素众多。为了明确内河船舶碰撞事故致因因素,分析内河船舶碰撞风险,选取2013年至2017年长江内河航道419起船舶碰撞事故报告作为文本挖掘语料,将语料库中的人为因素、船舶因素、自然环境因素以及管理因素作为目标数据,利用R语言和文本挖掘方法,获得包含特征值和特征值权重属性的高维稀疏的原始特征向量空间集合,对tf-idf公式进行平滑改进,解决了文本识别过程中无法识别统计较为生僻的专业名词的问题,提升了文本挖掘方法在交通运输领域的适应性。使用统计对其进行降维,获得最终33个维度的文本特征项,确定了船舶碰撞风险致因因素。通过实验验证了文本挖掘得出的船舶碰撞致因因素置信率达到了81%。以碰撞过程中的4个步骤为主线,构建基于"人-船-环境-管理"系统的船舶碰撞风险贝叶斯网络结构,计算贝叶斯网络结构中各节点的条件概率表,进行船舶碰撞风险建模。结合贝叶斯网络实现内河船舶碰撞事故的风险预测,并通过反向推理确定了人为因素是船舶碰撞事故致因中的首要因素。
As a typical high-risk maritime accident,inland ship collision is caused by multiply factors.In order to clarify cause factors and analyze risks of inland ship collision,419 accidents in inland waterway of the Yangtze River from2013 to 2017 are selected as a text mining corpus.Human factor,ship factor,natural environment factor,and management factor in the corpus are regarded as target data.Through R language and a text-mining method,a vector space set with high dimensional and sparse primitive character which contains eigenvalue and weight property of the eigenvalue is obtained.The tf-idfformula is improved by a smoothing method,solving the problem that relatively uncommon technical terms cannot be recognized,and their statistics cannot be made in the process of text recognition.The adaptability in transportation of the text mining method is enhanced.The statistic is used to reduce the dimension of the set,finally the text characteristics of 33 dimensions are obtained,and cause factors of ship collision are determined.Through case studies,confidence ratio of the cause factors reaches 81% by text mining.Taking four steps in a collision process as the main line,a Bayesian Network structure of ship collision risks based on a"human-ship-environment-management"system is developed.The conditional probability table of each node in the Bayesian Network structure is calculated to simulate ship collision risks.Combined with the Bayesian Network,risk forecasting of inland ship collision is realized.Human factor is determined as the primary factor of ship collision through backward reasoning.
As a typical high-risk maritime accident,inland ship collision is caused by multiply factors.In order to clarify cause factors and analyze risks of inland ship collision,419 accidents in inland waterway of the Yangtze River from2013 to 2017 are selected as a text mining corpus.Human factor,ship factor,natural environment factor,and management factor in the corpus are regarded as target data.Through R language and a text-mining method,a vector space set with high dimensional and sparse primitive character which contains eigenvalue and weight property of the eigenvalue is obtained.The tf-idfformula is improved by a smoothing method,solving the problem that relatively uncommon technical terms cannot be recognized,and their statistics cannot be made in the process of text recognition.The adaptability in transportation of the text mining method is enhanced.The statistic is used to reduce the dimension of the set,finally the text characteristics of 33 dimensions are obtained,and cause factors of ship collision are determined.Through case studies,confidence ratio of the cause factors reaches 81% by text mining.Taking four steps in a collision process as the main line,a Bayesian Network structure of ship collision risks based on a"human-ship-environment-management"system is developed.The conditional probability table of each node in the Bayesian Network structure is calculated to simulate ship collision risks.Combined with the Bayesian Network,risk forecasting of inland ship collision is realized.Human factor is determined as the primary factor of ship collision through backward reasoning.
引文
[1]交通运输部.2015交通运输行业发展统计公报[R].北京:交通运输部,2016.Ministry of Transport.2015statistical bulletin of the development of transportation industry[R].Beijing:Ministry of Transport,2016.(in Chinese)
[2]KUJALA P,HANNINEN M,AROLA T,et al.Analysis of the marine traffic safety in the Gulf of Finland[J].Reliability Engineering&System Safety,2009,94(8):1349-1357.
[3]ZAMAN M B.Study on Safety of Navigation using automatic identification system for marine traffic area case study:Malacca straits[J].Int.J of Marine Engineering Innovation and Research,2016,1(1):26-30.
[4]张磊.基于贝叶斯网络的内河船舶碰撞致因分析[D].太原:中北大学,2016.ZHANG Lei.The Causation analysis of inland river ship collision based on bayesian network[D].Taiyuan:North University of China,2016.(in Chinese)
[5]WU Bing,ZONG Liking,YAN Xinping,et al,modelling the collision risk in the Yangtze River using bayesian networks[C].2017 4th International Conference on Transportation Information and Safety(ICTIS),Alberta,Canada:Banff,2017.
[6]孙立成.人的个性与避碰行为关系的统计分析[J].大连海事大学学报,1997,23(1):41-45.SUN Licheng.Statistical analysis of the relationship between personality and collision avoidance[J].Journal of Dalian Maritime University,1997,23(1):41-45.(in Chinese)
[7]CURTIS R G.Determination of mariners'reaction times[J].Journal of Navigation,1980,31:409-417.
[8]魏晓阳,吴兵,黄金辉,等.船舶碰撞事故取证中的操纵意图分析[J].中国科技论文,2016,11(19):2173-2178.WEI Xiaoyang,WU Bing,HUANG Jinhui,et al.Ship maneuvering intention analysis in the forensic study of ship collision accidents[J].China Sciencepaper,2016,11(19):2173-2178.(in Chinese)
[9]陈昌源,戴冉,牛佳伟,等.改进灰色理论在海上交通事故预测中的应用[J].船海工程,2016,45(6):67-71.CHEN Changyuan,DAI Ran,NIU Jiawei,et al.Study on maritime traffic accidents prediction based on improved grey model[J].Ship&Ocean Engineering,2016,45(6):67-71.(in Chinese)
[10]尚前明,曹玉佩,梁峥.基于事故树分析法的“东方之星”翻沉事故研究[J].中国修船,2017,30(1):23-27.SHANG Qianming,CAO Yupei,LIANG Zheng.Study on the transfiguration of oriental star based on fault tree analysis[J].China Shiprepair,2017,30(1):23-27.(in Chinese)
[11]牛佳伟,李连博,陈昌源,等.加权灰色关联理论模型在辽宁海上交通事故分析与预测中的应用[J].上海海事大学学报,2016(1):65-69.NIU Jiawei,LI Lianbo,CHEN Changyuan,et al.Application of weighted grey correlation theory model in analysis and prediction of maritime traffic accidents in Liaoning[J].Journal of Shanghai Maritime University.2016(1):65-69.(in Chinese)
[12]梅馨,邢桂芬.文本挖掘技术综述[J].江苏大学学报(自然科学版),2003,24(5):72-76.MEI Xin,XING Guifen.On technology of document mining[J].Journal of Jiangsu University(Natural Science Edition).2003,24(5):72-76.(in Chinese)
[13]TAN A H,RIDGE K,LABS D,et al.Text Mining:The state of the art and the challenges[C].3rd Pacific Asia Knowledge Discovery and Data Mining(PAKDD),Beijing,China:DBLP,1999.
[14]ELMASRI R,NAVATHE S B.Fundamentals of database systems[M].San Francisco:Benjamin Cummings,1989.
[15]GUPTA V,LEHAL G S.A survey of text mining techniques and applications[J].Journal of Emerging Technologies in Web Intelligence,2009(1):60-76.
[16]陈阳,邵曦,赵海博.基于UR-LDA的微博主题挖掘[J].计算机技术与发展,2017,27(6):173-182.CHEN Yang,SHAO Xi,ZHAO Haibo.Microblog topic mining based on UR-LDA[J].Computer Technology and Development,2017,27(6):173-182.(in Chinese)
[17]袁闻,王晓晔,邓高登,等.基于贝叶斯分类器的中文垃圾短信辨识[J].科技资讯,2017,15(5):10-13.YUAN Wen,WANG Xiaoye,DENG Gaoden,et al.Recognition of Chinese spam messages based on Bayesian classifier[J]Science&Technology Information,2017,15(5):10-13.(in Chinese)
[18]黄发良,冯时,王大玲,等.基于多特征融合的微博主题情感挖掘[J].计算机学报,2017,40(4):872-888.HUANG Faliang,FENG Shi,WANG Daling,et al.Mining topic sentiment in microblogging based on multrfeature fusion[J].Chinese Journal of Computers,2017,40(4):872-888.(in Chinese)
[19]USAMA F,GREGORY P S,PADHRASIC S.The KDD processfor extracting useful knowledge from volume of data[J].Communications of the ACM,1996,39(11):27-34.
[20]严新平,张笛,张明阳,等.安全与可靠性领域研究动态与展望[J].交通信息与安全,2015,33(5):1-8.YAN Xinping,ZHANG Di,ZHANG Mingyang,et al.Research trends of transportation safety:I-dentified from ESREL 2015conference[J].Journal of Transport Information&Safety,2015,37(5):1-8.(in Chinese)
[2]KUJALA P,HANNINEN M,AROLA T,et al.Analysis of the marine traffic safety in the Gulf of Finland[J].Reliability Engineering&System Safety,2009,94(8):1349-1357.
[3]ZAMAN M B.Study on Safety of Navigation using automatic identification system for marine traffic area case study:Malacca straits[J].Int.J of Marine Engineering Innovation and Research,2016,1(1):26-30.
[4]张磊.基于贝叶斯网络的内河船舶碰撞致因分析[D].太原:中北大学,2016.ZHANG Lei.The Causation analysis of inland river ship collision based on bayesian network[D].Taiyuan:North University of China,2016.(in Chinese)
[5]WU Bing,ZONG Liking,YAN Xinping,et al,modelling the collision risk in the Yangtze River using bayesian networks[C].2017 4th International Conference on Transportation Information and Safety(ICTIS),Alberta,Canada:Banff,2017.
[6]孙立成.人的个性与避碰行为关系的统计分析[J].大连海事大学学报,1997,23(1):41-45.SUN Licheng.Statistical analysis of the relationship between personality and collision avoidance[J].Journal of Dalian Maritime University,1997,23(1):41-45.(in Chinese)
[7]CURTIS R G.Determination of mariners'reaction times[J].Journal of Navigation,1980,31:409-417.
[8]魏晓阳,吴兵,黄金辉,等.船舶碰撞事故取证中的操纵意图分析[J].中国科技论文,2016,11(19):2173-2178.WEI Xiaoyang,WU Bing,HUANG Jinhui,et al.Ship maneuvering intention analysis in the forensic study of ship collision accidents[J].China Sciencepaper,2016,11(19):2173-2178.(in Chinese)
[9]陈昌源,戴冉,牛佳伟,等.改进灰色理论在海上交通事故预测中的应用[J].船海工程,2016,45(6):67-71.CHEN Changyuan,DAI Ran,NIU Jiawei,et al.Study on maritime traffic accidents prediction based on improved grey model[J].Ship&Ocean Engineering,2016,45(6):67-71.(in Chinese)
[10]尚前明,曹玉佩,梁峥.基于事故树分析法的“东方之星”翻沉事故研究[J].中国修船,2017,30(1):23-27.SHANG Qianming,CAO Yupei,LIANG Zheng.Study on the transfiguration of oriental star based on fault tree analysis[J].China Shiprepair,2017,30(1):23-27.(in Chinese)
[11]牛佳伟,李连博,陈昌源,等.加权灰色关联理论模型在辽宁海上交通事故分析与预测中的应用[J].上海海事大学学报,2016(1):65-69.NIU Jiawei,LI Lianbo,CHEN Changyuan,et al.Application of weighted grey correlation theory model in analysis and prediction of maritime traffic accidents in Liaoning[J].Journal of Shanghai Maritime University.2016(1):65-69.(in Chinese)
[12]梅馨,邢桂芬.文本挖掘技术综述[J].江苏大学学报(自然科学版),2003,24(5):72-76.MEI Xin,XING Guifen.On technology of document mining[J].Journal of Jiangsu University(Natural Science Edition).2003,24(5):72-76.(in Chinese)
[13]TAN A H,RIDGE K,LABS D,et al.Text Mining:The state of the art and the challenges[C].3rd Pacific Asia Knowledge Discovery and Data Mining(PAKDD),Beijing,China:DBLP,1999.
[14]ELMASRI R,NAVATHE S B.Fundamentals of database systems[M].San Francisco:Benjamin Cummings,1989.
[15]GUPTA V,LEHAL G S.A survey of text mining techniques and applications[J].Journal of Emerging Technologies in Web Intelligence,2009(1):60-76.
[16]陈阳,邵曦,赵海博.基于UR-LDA的微博主题挖掘[J].计算机技术与发展,2017,27(6):173-182.CHEN Yang,SHAO Xi,ZHAO Haibo.Microblog topic mining based on UR-LDA[J].Computer Technology and Development,2017,27(6):173-182.(in Chinese)
[17]袁闻,王晓晔,邓高登,等.基于贝叶斯分类器的中文垃圾短信辨识[J].科技资讯,2017,15(5):10-13.YUAN Wen,WANG Xiaoye,DENG Gaoden,et al.Recognition of Chinese spam messages based on Bayesian classifier[J]Science&Technology Information,2017,15(5):10-13.(in Chinese)
[18]黄发良,冯时,王大玲,等.基于多特征融合的微博主题情感挖掘[J].计算机学报,2017,40(4):872-888.HUANG Faliang,FENG Shi,WANG Daling,et al.Mining topic sentiment in microblogging based on multrfeature fusion[J].Chinese Journal of Computers,2017,40(4):872-888.(in Chinese)
[19]USAMA F,GREGORY P S,PADHRASIC S.The KDD processfor extracting useful knowledge from volume of data[J].Communications of the ACM,1996,39(11):27-34.
[20]严新平,张笛,张明阳,等.安全与可靠性领域研究动态与展望[J].交通信息与安全,2015,33(5):1-8.YAN Xinping,ZHANG Di,ZHANG Mingyang,et al.Research trends of transportation safety:I-dentified from ESREL 2015conference[J].Journal of Transport Information&Safety,2015,37(5):1-8.(in Chinese)
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |