Article(id=1149738771702264027, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1149738762382524507, articleNumber=1003-3033(2024)07-0153-10, orderNo=null, doi=10.16265/j.cnki.issn1003-3033.2024.07.0089, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1704988800000, receivedDateStr=2024-01-12, revisedDate=1713369600000, revisedDateStr=2024-04-18, acceptedDate=null, acceptedDateStr=null, onlineDate=1752048684286, onlineDateStr=2025-07-09, pubDate=1722096000000, pubDateStr=2024-07-28, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1752048684286, onlineIssueDateStr=2025-07-09, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1752048684286, creator=13701087609, updateTime=1752048684286, updator=13701087609, issue=Issue{id=1149738762382524507, tenantId=1146029695717560320, journalId=1146031787341344770, year='2024', volume='34', issue='7', pageStart='1', pageEnd='252', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=0, createTime=1752048682065, creator=13701087609, updateTime=1757316437713, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1171833331021824745, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1149738762382524507, language=EN, specialIssueTitle=, coverIllustrator=, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1171833331021824746, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1149738762382524507, language=CN, specialIssueTitle=, coverIllustrator=, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=153, endPage=160, ext={EN=ArticleExt(id=1149738772062974173, articleId=1149738771702264027, tenantId=1146029695717560320, journalId=1146031787341344770, language=EN, title=Urban taxi traffic flow prediction based on attentive ConvLSTM-ResNet model, columnId=1149733270084042840, journalTitle=China Safety Science Journal, columnName=Public safety, runingTitle=null, highlight=null, articleAbstract=

In order to address the challenges of urban traffic congestion and safety,an ACLR model was proposed. By integrating ConvLSTM,attention mechanisms,and residual structures,the ACLR model effectively enhanced the extraction of spatio-temporal traffic features.The time,space and other characteristics of taxi traffic were processed respectively,and the influence of regional point of interest(POI) data on taxi traffic was mined. Additionally,a specialized learning component was incorporated to capture the impact of external factors and point-of-interest density on traffic flow. Using taxi trajectory data from Beijing,the ACLR model demonstrates superior prediction accuracy compared to other models such as the autoregressive integrated moving average (ARIMA) model,long short-term memory (LSTM),deep spatio-temporal residual networks (ST-ResNet),convolutional neural network(CNN)-ResNet-LSTM (CRL),and attentive crowd flow machines (ACFM) in urban traffic flow forecasting,which is helpful to improve the prediction performance of the model without POI density or considering POI density. The predicted value of the ACLA model is basically consistent with the real value,and it can also be in good agreement with the real value during peak hours,which effectively improves the ability to extract traffic temporal and spatial characteristics,reduces the prediction error,and optimizes the traffic flow prediction performance.

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Xinmin ZHOU, Jiangtao JIN, Nana BAO, Tao YUAN, Ye CUI), CN=ArticleExt(id=1149738788273959754, articleId=1149738771702264027, tenantId=1146029695717560320, journalId=1146031787341344770, language=CN, title=基于注意力卷积长短时记忆模型的城市出租车流量预测, columnId=1149733271510106222, journalTitle=中国安全科学学报, columnName=公共安全, runingTitle=null, highlight=null, articleAbstract=

为解决城市交通拥堵和安全问题,提出一种注意力卷积长短时记忆(ConvLSTM)残差(ACLR)模型,该模型通过结合ConvLSTM、注意力机制和残差结构,分别处理出租车流量的时间、空间、和其他特征,挖掘区域兴趣点(POI)数据对出租车流量的影响,有效提升交通时空特征的提取能力。同时,引入专门的学习元件考虑外部因素和POI密度对交通流量的影响,并利用北京市出租车轨迹数据验证。结果表明:ACLR模型在城市交通流预测中的精度高于差分自回归滑动平均(ARIMA)模型、长短时记忆(LSTM)网络、深度时空残差网络(ST-ResNet)、卷积神经网络(CNN)-残差神经单元-LSTM(CRL)循环神经网络、ACFM等模型,在无POI密度和考虑POI密度的情况下,均有助于提升模型的预测性能,ACLA模型的预测值与真实值基本一致,高峰时段也能与真实值较好地吻合,有效提升交通时空特征的提取能力,降低预测误差,使得交通流量预测性能得到优化。

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=zfd6hDq8ADx51U+eiuNUJw==, magXml=ApPYYyS75yRZbvJLubLKgQ==, pdfUrl=null, pdf=NLRCctmO06Wdu4VMyfmhxg==, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=vOBdKjxJLmlx9pbCVpT+Bw==, mapNumber=null, authorCompany=null, fund=null, authors=

周新民 (1977—),男,湖南新邵人,博士,教授,主要从事新型智慧城市、商务智能与大数据、互联网安全与服务等方面的研究。E-mail:

, authorsList=周新民, 金江涛, 鲍娜娜, 袁涛, 崔烨)}, authors=[Author(id=1168186651227467861, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=zhouxinmin2699@163.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1168186651298771032, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186651227467861, language=EN, stringName=Xinmin ZHOU, firstName=Xinmin, middleName=null, lastName=ZHOU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, address=1 School of Artificial Intelligence and Advanced Computing,Hunan University of Technology and Business,Changsha Hunan 410205,China
2 Xiangjiang Lab,Changsha Hunan 410205,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1168186651357491289, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186651227467861, language=CN, stringName=周新民, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, address=1 湖南工商大学 人工智能与先进计算学院,湖南 长沙 410205
2 湘江实验室,湖南 长沙 410205, bio={"img":"kSFJ1WWJt5ApJ6+2x4xu8w==","content":"

周新民 (1977—),男,湖南新邵人,博士,教授,主要从事新型智慧城市、商务智能与大数据、互联网安全与服务等方面的研究。E-mail:

"}, bioImg=kSFJ1WWJt5ApJ6+2x4xu8w==, bioContent=

周新民 (1977—),男,湖南新邵人,博士,教授,主要从事新型智慧城市、商务智能与大数据、互联网安全与服务等方面的研究。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1168186650938060872, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=1, ext=[AuthorCompanyExt(id=1168186650942255177, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186650938060872, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Artificial Intelligence and Advanced Computing,Hunan University of Technology and Business,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186650950643786, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186650938060872, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 湖南工商大学 人工智能与先进计算学院,湖南 长沙 410205)]), AuthorCompany(id=1168186650996781131, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=2, ext=[AuthorCompanyExt(id=1168186651005169740, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186650996781131, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Xiangjiang Lab,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186651009364045, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186650996781131, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 湘江实验室,湖南 长沙 410205)])]), Author(id=1168186651420405851, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1168186651504291933, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186651420405851, language=EN, stringName=Jiangtao JIN, firstName=Jiangtao, middleName=null, lastName=JIN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=3, address=3 School of Frontier Crossover Studies,Hunan University of Technology and Business,Changsha Hunan 410205,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1168186651563012190, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186651420405851, language=CN, stringName=金江涛, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=3, address=3 湖南工商大学 前沿交叉学院,湖南 长沙 410205, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1168186651072278606, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=3, ext=[AuthorCompanyExt(id=1168186651076472911, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651072278606, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 School of Frontier Crossover Studies,Hunan University of Technology and Business,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186651084861520, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651072278606, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 湖南工商大学 前沿交叉学院,湖南 长沙 410205)])]), Author(id=1168186651613343840, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1168186651701424226, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186651613343840, language=EN, stringName=Nana BAO, firstName=Nana, middleName=null, lastName=BAO, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=4, address=4 School of Computer Science,Hunan University of Technology and Business,Changsha Hunan 410205,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1168186651755950179, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186651613343840, language=CN, stringName=鲍娜娜, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=4, address=4 湖南工商大学 计算机学院,湖南 长沙 410205, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1168186651164553297, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=4, ext=[AuthorCompanyExt(id=1168186651168747602, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651164553297, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4 School of Computer Science,Hunan University of Technology and Business,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186651177136211, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651164553297, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4 湖南工商大学 计算机学院,湖南 长沙 410205)])]), Author(id=1168186651810476133, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1168186651869196391, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186651810476133, language=EN, stringName=Tao YUAN, firstName=Tao, middleName=null, lastName=YUAN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=3, address=3 School of Frontier Crossover Studies,Hunan University of Technology and Business,Changsha Hunan 410205,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1168186651923722344, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186651810476133, language=CN, stringName=袁涛, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=3, address=3 湖南工商大学 前沿交叉学院,湖南 长沙 410205, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1168186651072278606, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=3, ext=[AuthorCompanyExt(id=1168186651076472911, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651072278606, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 School of Frontier Crossover Studies,Hunan University of Technology and Business,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186651084861520, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651072278606, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 湖南工商大学 前沿交叉学院,湖南 长沙 410205)])]), Author(id=1168186652049551466, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1168186652112466028, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186652049551466, language=EN, stringName=Ye CUI, firstName=Ye, middleName=null, lastName=CUI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=4, address=4 School of Computer Science,Hunan University of Technology and Business,Changsha Hunan 410205,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1168186652208935021, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, authorId=1168186652049551466, language=CN, stringName=崔烨, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=4, address=4 湖南工商大学 计算机学院,湖南 长沙 410205, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1168186651164553297, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=4, ext=[AuthorCompanyExt(id=1168186651168747602, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651164553297, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4 School of Computer Science,Hunan University of Technology and Business,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186651177136211, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651164553297, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4 湖南工商大学 计算机学院,湖南 长沙 410205)])])], keywords=[Keyword(id=1168186652309598318, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, orderNo=1, keyword=attentive convolutional long short-term memory(ConvLSTM) residual network(ResNet)(ACLR)), Keyword(id=1168186652380901487, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, orderNo=2, keyword=urban taxi), Keyword(id=1168186652477370480, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, orderNo=3, keyword=traffic flow prediction), Keyword(id=1168186652536090737, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, orderNo=4, keyword=space-time characteristics), Keyword(id=1168186652582228082, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, orderNo=5, keyword=residual structure), Keyword(id=1168186652636754035, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, orderNo=1, keyword=注意力卷积长短时记忆残差网络(ACLR)模型), Keyword(id=1168186652682891380, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, orderNo=2, keyword=交通流量预测), Keyword(id=1168186652737417333, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, orderNo=3, keyword=城市出租车), Keyword(id=1168186652791943286, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, orderNo=4, keyword=时空特征), Keyword(id=1168186652842274935, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, orderNo=5, keyword=残差结构)], refs=[Reference(id=1168186655161725079, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2021, volume=31, issue=3, pageStart=112, pageEnd=120, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=姬浩, 王永东, 李佩, journalName=中国安全科学学报, refType=null, unstructuredReference=姬浩, 王永东, 李佩, 等. 事故车辆影响下的城市三车道道路交通流仿真[J]. 中国安全科学学报, 2021, 31(3):112-120., articleTitle=事故车辆影响下的城市三车道道路交通流仿真, refAbstract=null), Reference(id=1168186655216251032, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2021, volume=31, issue=3, pageStart=112, pageEnd=120, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=JI Hao, WANG Yongdong, LI Pei, journalName=China Safety Science Journal, refType=null, unstructuredReference=JI Hao, WANG Yongdong, LI Pei, et al. Traffic flow simulation of urban three-lane road considering influence of accident vehicle[J]. China Safety Science Journal, 2021, 31(3):112-120., articleTitle=Traffic flow simulation of urban three-lane road considering influence of accident vehicle, refAbstract=null), Reference(id=1168186655274971289, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2022, volume=32, issue=5, pageStart=170, pageEnd=176, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=熊晓夏, 刘擎超, 沈钰杰, journalName=中国安全科学学报, refType=null, unstructuredReference=熊晓夏, 刘擎超, 沈钰杰, 等. 基于LSTM-BF的高速公路交通事故风险模型[J]. 中国安全科学学报, 2022, 32(5):170-176., articleTitle=基于LSTM-BF的高速公路交通事故风险模型, refAbstract=null), Reference(id=1168186655337885850, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2022, volume=32, issue=5, pageStart=170, pageEnd=176, url=null, language=null, rfNumber=[2], rfOrder=3, authorNames=XIONG Xiaoxia, LIU Qingchao, SHEN Yujie, journalName=China Safety Science Journal, refType=null, unstructuredReference=XIONG Xiaoxia, LIU Qingchao, SHEN Yujie, et al. Study on risk model of highway traffic accidents based on LSTM-BF[J]. China Safety Science Journal, 2022, 32(5):170-176., articleTitle=Study on risk model of highway traffic accidents based on LSTM-BF, refAbstract=null), Reference(id=1168186655392411803, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2002, volume=10, issue=4, pageStart=303, pageEnd=321, url=null, language=null, rfNumber=[3], rfOrder=4, authorNames=SMITH B L, WILLIAMS B M, OSWALD R K, journalName=Transportation Research Part C, refType=null, unstructuredReference=SMITH B L, WILLIAMS B M, OSWALD R K. Comparison of parametric and nonparametric models for traffic flow forecasting[J]. Transportation Research Part C, 2002, 10(4):303-321., articleTitle=Comparison of parametric and nonparametric models for traffic flow forecasting, refAbstract=null), Reference(id=1168186655446937756, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2014, volume=43, issue=2, pageStart=50, pageEnd=64, url=null, language=null, rfNumber=[4], rfOrder=5, authorNames=GUO Jianhua, HUANG Wei, WILLIAMS B M, journalName=Transportation Research Part C, refType=null, unstructuredReference=GUO Jianhua, HUANG Wei, WILLIAMS B M. Adaptive Kalman filter approach for stochastic short-term traffic flow rate prediction and uncertainty quantification[J]. Transportation Research Part C, 2014, 43(2):50-64., articleTitle=Adaptive Kalman filter approach for stochastic short-term traffic flow rate prediction and uncertainty quantification, refAbstract=null), Reference(id=1168186655543406749, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2019, volume=21, issue=9, pageStart=3848, pageEnd=3858, url=null, language=null, rfNumber=[5], rfOrder=6, authorNames=ZHAO Ling, SONG Yujiao, ZHANG Chao, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=ZHAO Ling, SONG Yujiao, ZHANG Chao, et al. T-gcn: a temporal graph convolutional network for traffic prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 21(9): 3848-3858., articleTitle=T-gcn: a temporal graph convolutional network for traffic prediction, refAbstract=null), Reference(id=1168186655614709918, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2017, volume=17, issue=4, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[6], rfOrder=7, authorNames=MA Xiaolei, DAI Zhuang, HE Zhengbing, journalName=Sensors, refType=null, unstructuredReference=MA Xiaolei, DAI Zhuang, HE Zhengbing, et al. Learning traffic as images: a deep convolutional neural network for large-scale transportation network speed prediction[J]. Sensors, 2017, 17(4):DOI:10.3390/s17040818., articleTitle=Learning traffic as images: a deep convolutional neural network for large-scale transportation network speed prediction, refAbstract=null), Reference(id=1168186655690207391, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2015, volume=54, issue=null, pageStart=187, pageEnd=197, url=null, language=null, rfNumber=[7], rfOrder=8, authorNames=MA Xiaolei, TAO Zhimin, WANG Yinhai, journalName=Transportation Research Part C: Emerging Technologies, refType=null, unstructuredReference=MA Xiaolei, TAO Zhimin, WANG Yinhai, et al. Long short-term memory neural network for traffic speed prediction using remote microwave sensor data[J]. Transportation Research Part C: Emerging Technologies, 2015, 54: 187-197., articleTitle=Long short-term memory neural network for traffic speed prediction using remote microwave sensor data, refAbstract=null), Reference(id=1168186655740539040, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2018, volume=18, issue=4, pageStart=215, pageEnd=223, url=null, language=null, rfNumber=[8], rfOrder=9, authorNames=段宗涛, 张凯, 杨云, journalName=交通运输系统工程与信息, refType=null, unstructuredReference=段宗涛, 张凯, 杨云, 等. 基于深度CNN-LSTM-ResNet组合模型的出租车需求预测[J]. 交通运输系统工程与信息, 2018, 18(4):215-223., articleTitle=基于深度CNN-LSTM-ResNet组合模型的出租车需求预测, refAbstract=null), Reference(id=1168186655795064993, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2018, volume=18, issue=4, pageStart=215, pageEnd=223, url=null, language=null, rfNumber=[8], rfOrder=10, authorNames=DUAN Zongtao, ZHANG Kai, YANG Yun, etal, journalName=Journal of Transportation Systems Engineering and Information Technology, refType=null, unstructuredReference=DUAN Zongtao, ZHANG Kai, YANG Yun, etal. Taxi Demand prediction based on CNN-LSTM-ResNet hybrid depth learning model[J]. Journal of Transportation Systems Engineering and Information Technology, 2018, 18(4):215-223., articleTitle=Taxi Demand prediction based on CNN-LSTM-ResNet hybrid depth learning model, refAbstract=null), Reference(id=1168186655899922594, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2017, volume=6, issue=null, pageStart=1662, pageEnd=1669, url=null, language=null, rfNumber=[9], rfOrder=11, authorNames=KARIM F, MAJUMDAR S, DARABI H, journalName=IEEE Access, refType=null, unstructuredReference=KARIM F, MAJUMDAR S, DARABI H, et al. LSTM fully convolutional networks fortime series classification[J]. IEEE Access, 2017, 6: 1662-1669., articleTitle=LSTM fully convolutional networks fortime series classification, refAbstract=null), Reference(id=1168186655967031459, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2021, volume=578, issue=null, pageStart=401, pageEnd=416, url=null, language=null, rfNumber=[10], rfOrder=12, authorNames=PENG Hao, DU Bowen, LIU Mingsheng, journalName=Information Sciences, refType=null, unstructuredReference=PENG Hao, DU Bowen, LIU Mingsheng, et al. Dynamic graph convolutional network for long-term traffic flow prediction with reinforcement learning[J]. Information Sciences, 2021, 578:401-416., articleTitle=Dynamic graph convolutional network for long-term traffic flow prediction with reinforcement learning, refAbstract=null), Reference(id=1168186656034140324, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2021, volume=23, issue=9, pageStart=810, pageEnd=16, url=null, language=null, rfNumber=[11], rfOrder=13, authorNames=XING Lumin, LIU Wenjian, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=XING Lumin, LIU Wenjian. A data fusion powered Bi-Directional long short term memory model for predicting multi-lane short term traffic flow[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(9):16810-16 819., articleTitle=A data fusion powered Bi-Directional long short term memory model for predicting multi-lane short term traffic flow, refAbstract=null), Reference(id=1168186656084471973, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2021, volume=23, issue=9, pageStart=654, pageEnd=16, url=null, language=null, rfNumber=[12], rfOrder=14, authorNames=SHU Wanneng, CAI Ken, XIONG Naixue, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=SHU Wanneng, CAI Ken, XIONG Naixue. A short-term traffic flow prediction model based on an improved gate recurrent unit neural network[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(9):16654-16 665., articleTitle=A short-term traffic flow prediction model based on an improved gate recurrent unit neural network, refAbstract=null), Reference(id=1168186656218689702, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=802, pageEnd=810, url=null, language=null, rfNumber=[13], rfOrder=15, authorNames=SHI Xingjian, CHEN Zhourong, WANG Hao, journalName=International Conference on Neural Information Processing Systems, refType=null, unstructuredReference=SHI Xingjian, CHEN Zhourong, WANG Hao, et al. Convolutional LSTM network: a machine learning approach for precipitation nowcasting[C]. International Conference on Neural Information Processing Systems, 2015: 802-810., articleTitle=Convolutional LSTM network: a machine learning approach for precipitation nowcasting, refAbstract=null), Reference(id=1168186656285798567, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2023, volume=24, issue=8, pageStart=8650, pageEnd=8666, url=null, language=null, rfNumber=[14], rfOrder=16, authorNames=ZHAO Yiji, LIN Youfang, WEN Haomin, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=ZHAO Yiji, LIN Youfang, WEN Haomin, et al. Spatial-temporal position-aware graph convolution networks for traffic flow forecasting[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(8):8650-8666., articleTitle=Spatial-temporal position-aware graph convolution networks for traffic flow forecasting, refAbstract=null), Reference(id=1168186656336130216, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, doi=null, pmid=null, pmcid=null, year=2019, volume=20, issue=10, pageStart=3913, pageEnd=3926, url=null, language=null, rfNumber=[15], rfOrder=17, authorNames=GUO Shengnan, LIN Youfang, LI Shijie, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=GUO Shengnan, LIN Youfang, LI Shijie, et al. Deep spatial-temporal 3D convolutional neural networks for traffic data forecasting[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 20(10):3913-3926., articleTitle=Deep spatial-temporal 3D convolutional neural networks for traffic data forecasting, refAbstract=null)], funds=[Fund(id=1168186655035895958, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, awardId=21BGL231, language=CN, fundingSource=国家社会科学基金资助(21BGL231), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1168186650938060872, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=1, ext=[AuthorCompanyExt(id=1168186650942255177, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186650938060872, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Artificial Intelligence and Advanced Computing,Hunan University of Technology and Business,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186650950643786, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186650938060872, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 湖南工商大学 人工智能与先进计算学院,湖南 长沙 410205)]), AuthorCompany(id=1168186650996781131, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=2, ext=[AuthorCompanyExt(id=1168186651005169740, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186650996781131, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Xiangjiang Lab,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186651009364045, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186650996781131, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 湘江实验室,湖南 长沙 410205)]), AuthorCompany(id=1168186651072278606, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=3, ext=[AuthorCompanyExt(id=1168186651076472911, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651072278606, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 School of Frontier Crossover Studies,Hunan University of Technology and Business,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186651084861520, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651072278606, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 湖南工商大学 前沿交叉学院,湖南 长沙 410205)]), AuthorCompany(id=1168186651164553297, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, xref=4, ext=[AuthorCompanyExt(id=1168186651168747602, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651164553297, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4 School of Computer Science,Hunan University of Technology and Business,Changsha Hunan 410205,China), AuthorCompanyExt(id=1168186651177136211, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, companyId=1168186651164553297, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4 湖南工商大学 计算机学院,湖南 长沙 410205)])], figs=[ArticleFig(id=1168186652989075576, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.1, caption=Framework of ACLR prediction model, figureFileSmall=twN8gtebQysPJGFOgv/H5A==, figureFileBig=Ykwb5k2jVk3oXEivNQZTMg==, tableContent=null), ArticleFig(id=1168186653043601529, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图1, caption=ACLR预测模型框架, figureFileSmall=twN8gtebQysPJGFOgv/H5A==, figureFileBig=Ykwb5k2jVk3oXEivNQZTMg==, tableContent=null), ArticleFig(id=1168186653102321786, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.2, caption=Data of traffic trajectory is converted into "space-time image", figureFileSmall=bpWeapyjl7L5Q4CPAFmemw==, figureFileBig=qou62FkpbdIF1UCKDftkkw==, tableContent=null), ArticleFig(id=1168186653156847739, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图2, caption=交通轨迹数据转换为“时空图像”, figureFileSmall=bpWeapyjl7L5Q4CPAFmemw==, figureFileBig=qou62FkpbdIF1UCKDftkkw==, tableContent=null), ArticleFig(id=1168186653215567996, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.3, caption=Construction method of "closeness,period and trend" data sets, figureFileSmall=7/tb++cI8GeEYc8XJcInVg==, figureFileBig=Wxvb6lWaLL+FRWf607hAmw==, tableContent=null), ArticleFig(id=1168186653278482557, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图3, caption=“邻近性、周期性、趋势性”数据集构建方法, figureFileSmall=7/tb++cI8GeEYc8XJcInVg==, figureFileBig=Wxvb6lWaLL+FRWf607hAmw==, tableContent=null), ArticleFig(id=1168186653345591422, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.4, caption=structure of ACLR model, figureFileSmall=e38OCBOSo2u2X9yI3gHYLw==, figureFileBig=WbPs7Fn1tWz2mUb8vNEa2A==, tableContent=null), ArticleFig(id=1168186653400117375, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图4, caption=ACLR模型结构, figureFileSmall=e38OCBOSo2u2X9yI3gHYLw==, figureFileBig=WbPs7Fn1tWz2mUb8vNEa2A==, tableContent=null), ArticleFig(id=1168186653463031936, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.5, caption=Process of Squeeze-and-Excitation(SE), figureFileSmall=4C5M2+g5EOLl86pBfFJvbw==, figureFileBig=yAxBLPM0jcZdMNMjQPcSkw==, tableContent=null), ArticleFig(id=1168186653513363585, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图5, caption=Squeeze-and-Excitation(SE)过程, figureFileSmall=4C5M2+g5EOLl86pBfFJvbw==, figureFileBig=yAxBLPM0jcZdMNMjQPcSkw==, tableContent=null), ArticleFig(id=1168186653572083842, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.6, caption=Learning component structure, figureFileSmall=gI/25FTlLyiEI/aJ1OsAIg==, figureFileBig=1G04+GN5UJ7I+NnnMxu+BA==, tableContent=null), ArticleFig(id=1168186653630804099, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图6, caption=学习元件结构, figureFileSmall=gI/25FTlLyiEI/aJ1OsAIg==, figureFileBig=1G04+GN5UJ7I+NnnMxu+BA==, tableContent=null), ArticleFig(id=1168186653693718660, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.7, caption=Influence of convolution kernel size on RMSE, figureFileSmall=n8sABSFhpHUeSksZsLL1ug==, figureFileBig=knRWsgcsfLA5xgDoNVNnqA==, tableContent=null), ArticleFig(id=1168186653748244613, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图7, caption=卷积核尺寸对RMSE的影响, figureFileSmall=n8sABSFhpHUeSksZsLL1ug==, figureFileBig=knRWsgcsfLA5xgDoNVNnqA==, tableContent=null), ArticleFig(id=1168186653811159174, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.8, caption=Influence of filter number on RMSE, figureFileSmall=0M53MZOQDmsuqSOrWPbhGQ==, figureFileBig=Cj2x2euZgjEuAFjB23LHyA==, tableContent=null), ArticleFig(id=1168186653903433863, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图8, caption=滤波器数量对RMSE的影响, figureFileSmall=0M53MZOQDmsuqSOrWPbhGQ==, figureFileBig=Cj2x2euZgjEuAFjB23LHyA==, tableContent=null), ArticleFig(id=1168186653970542728, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.9, caption=Influence of iteration times on RMSE, figureFileSmall=wDnLzWTpwgAgcbZ207pYsw==, figureFileBig=W3Arqo9TnjvsxRlX+f6u7Q==, tableContent=null), ArticleFig(id=1168186654020874377, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图9, caption=迭代次数对RMSE的影响, figureFileSmall=wDnLzWTpwgAgcbZ207pYsw==, figureFileBig=W3Arqo9TnjvsxRlX+f6u7Q==, tableContent=null), ArticleFig(id=1168186654075400330, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.10, caption=Evaluation of each module of model, figureFileSmall=xY5n6mvgydmNMoXS+UccHg==, figureFileBig=+QxdCruyZ6ng2lksiml6WQ==, tableContent=null), ArticleFig(id=1168186654213812363, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图10, caption=模型各模块评估, figureFileSmall=xY5n6mvgydmNMoXS+UccHg==, figureFileBig=+QxdCruyZ6ng2lksiml6WQ==, tableContent=null), ArticleFig(id=1168186654310281356, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Fig.11, caption=Comparison between predicted value and real value, figureFileSmall=Mz3EqGiC2Ln3WCJW3WBqZg==, figureFileBig=UBvigVHRtagud+7riDKzFA==, tableContent=null), ArticleFig(id=1168186654377390221, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=图11, caption=预测值与真实值对比, figureFileSmall=Mz3EqGiC2Ln3WCJW3WBqZg==, figureFileBig=UBvigVHRtagud+7riDKzFA==, tableContent=null), ArticleFig(id=1168186654436110478, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=, caption=

, figureFileSmall=null, figureFileBig=null, tableContent=
算法1:基于ACLR的交通流量预测算法
Input:历史数据: { X 0 X 1 X n - 1 };
外部特征: { E 0 E 1 E n - 1 };
POI密度: ρ;
“邻近性、周期性、趋势性”数据集的步长: l c l P l t r;
周期跨度:p;趋势跨度:q;
Output:训练完成的ACLR模型M
//构造训练实例
1 D ←Ø
2 for ( 1 t n - 1 ) do
//t为所有可用时间间隔
3 S c = [ X t - l c X t - ( l c - 1 ) X t - 1 ]
4 S p = [ X t - l p · p X t - ( l p - 1 ) · p X t - p ]
5 S t r = [ X t - l t r · q X t - ( l t r - 1 ) · q X t - q ]
// X t为目标时刻t的预测流量
6将一个训练实例 ( { S c S p S t r E t ρ p o i } X t )输入D
//训练模型
7初始化参数 θ
8 repeat
9 从D中随机选择一批训练实例 D b
10 输入 D b,调整参数 θ,使得式(6)最小化
11 until达到最大迭代次数
12 return M
), ArticleFig(id=1168186654507413647, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=, caption=

, figureFileSmall=null, figureFileBig=null, tableContent=
算法1:基于ACLR的交通流量预测算法
Input:历史数据: { X 0 X 1 X n - 1 };
外部特征: { E 0 E 1 E n - 1 };
POI密度: ρ;
“邻近性、周期性、趋势性”数据集的步长: l c l P l t r;
周期跨度:p;趋势跨度:q;
Output:训练完成的ACLR模型M
//构造训练实例
1 D ←Ø
2 for ( 1 t n - 1 ) do
//t为所有可用时间间隔
3 S c = [ X t - l c X t - ( l c - 1 ) X t - 1 ]
4 S p = [ X t - l p · p X t - ( l p - 1 ) · p X t - p ]
5 S t r = [ X t - l t r · q X t - ( l t r - 1 ) · q X t - q ]
// X t为目标时刻t的预测流量
6将一个训练实例 ( { S c S p S t r E t ρ p o i } X t )输入D
//训练模型
7初始化参数 θ
8 repeat
9 从D中随机选择一批训练实例 D b
10 输入 D b,调整参数 θ,使得式(6)最小化
11 until达到最大迭代次数
12 return M
), ArticleFig(id=1168186654595494032, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Table 1, caption=

Beijing taxi track data set

, figureFileSmall=null, figureFileBig=null, tableContent=
参数 数值
数据类型 北京市出租车GPS数据
时间跨度 2013-07-01—2013-10-30
2014-03-01—2014-06-30
2015-03-01—2015-06-30
2015-11-01—2016-04-10
时间间隙/min 30
网格尺寸 32×32
网格尺寸 32×32
), ArticleFig(id=1168186654650019985, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=表1, caption=

北京出租车轨迹数据集合

, figureFileSmall=null, figureFileBig=null, tableContent=
参数 数值
数据类型 北京市出租车GPS数据
时间跨度 2013-07-01—2013-10-30
2014-03-01—2014-06-30
2015-03-01—2015-06-30
2015-11-01—2016-04-10
时间间隙/min 30
网格尺寸 32×32
网格尺寸 32×32
), ArticleFig(id=1168186654708740242, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Table 2, caption=

External data set

, figureFileSmall=null, figureFileBig=null, tableContent=
外部数据 描述
天气情况 16种天气类型
温度/℃ [-24.6,41.0]
风速/(m·s-1) [0,48.6]
节假日 周末、法定节假日
), ArticleFig(id=1168186654763266195, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=表2, caption=

外部数据集

, figureFileSmall=null, figureFileBig=null, tableContent=
外部数据 描述
天气情况 16种天气类型
温度/℃ [-24.6,41.0]
风速/(m·s-1) [0,48.6]
节假日 周末、法定节假日
), ArticleFig(id=1168186654830375060, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=EN, label=Table 3, caption=

Model comparison

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 条件 RMSE MAPE/%
ARIMA 无POI密度 22.78 43.16
LSTM 无POI密度 24.81 35.64
ST-ResNet 无POI密度 16.69 23.92
CRL 无POI密度 16.10 21.35
ACFM 无POI密度 15.40 19.31
ACLR 无POI密度 15.31 17.83
考虑POI密度 14.52 16.57
), ArticleFig(id=1168186654897483925, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1149738771702264027, language=CN, label=表3, caption=

模型对比

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 条件 RMSE MAPE/%
ARIMA 无POI密度 22.78 43.16
LSTM 无POI密度 24.81 35.64
ST-ResNet 无POI密度 16.69 23.92
CRL 无POI密度 16.10 21.35
ACFM 无POI密度 15.40 19.31
ACLR 无POI密度 15.31 17.83
考虑POI密度 14.52 16.57
)], attaches=null, journal=Journal(id=1123942128916217864, delFlag=0, nameCn=中国安全科学学报, nameEn=China Safety Science Journal, nameHistory1=null, nameHistory2=null, issn=1003-3033, eissn=, cn=11-2865/X, coden=null, periodic=0, language=CN, oaType=0, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=fkqsFM6VKlHC4gCtS5XqTw==, journalPrice=null, startedYear=null, abbrevIsoEn=Chin Saf Sci J, journalRemark=null, publicationField=null, createdTime=null, updatedTime=1754269350027, createdBy=null, updatedBy=13701087609, firstLetterCn=C, firstLetterEn=C, subjectCode=Engineering, subjectName=工程, subjectCodeEn=Engineering, subjectNameEn=null, picCn=fkqsFM6VKlHC4gCtS5XqTw==, picEn=SHn9HgqSxtJrOcAxqD++4Q==, jcr=null, cjcr=null, exts=[JournalExt(id=1159052918994595848, language=CN, name=中国安全科学学报, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.cssjj.com.cn/, createdTime=1754269350050, updatedTime=1754269350050, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=http://www.cssjj.com.cn/CN/column/item15.shtml, submissionAuthorUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionReviewUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionCeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionAeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, option={"copyright":""}), JournalExt(id=1159052919040733193, language=EN, name=China Safety Science Journal, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.cssjj.com.cn/EN/1003-3033/home.shtml, createdTime=1754269350061, updatedTime=1754269350061, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=https://synbioj.cip.com.cn/EN/column/column3.shtml, submissionAuthorUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionReviewUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionCeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionAeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, option={"copyright":""})], databaseList=null, tenantJournalId=1146031787341344770, websiteList=[Website(id=1148243202345263519, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146031787341344770, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN, language=CN, createTime=1751692112766, createBy=18614031015, updateTime=1753502583634, updateBy=18614031015, name=《中国安全科学学报》中文站点, tplId=1146099689490845704, title=中国安全科学学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1148618794941046792, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=articleTextType, value=kx, createTime=1751781661020, updateTime=1751781661020, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794911686661, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=banner, value=null, createTime=1751781661012, updateTime=1751781661012, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794894909444, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=tui0IVO9FMwB61HHtX5scg==, createTime=1751781661008, updateTime=1751781661008, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794932658183, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1751781661017, updateTime=1751781661017, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794924269574, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1751781661015, updateTime=1751781661015, creator=18614031015, updator=18614031015)]), Website(id=1155836763751993353, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146031787341344770, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/zgaqkxxb/EN, language=EN, createTime=1753502558893, createBy=18614031015, updateTime=1753524450387, updateBy=18614031015, name=《中国安全科学学报》英文站点, tplId=1146101810881728533, title=China Safety Science Journal, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1155895925743669425, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=articleTextType, value=kx, createTime=1753516664205, updateTime=1753516664205, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925722697902, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=banner, value=null, createTime=1753516664200, updateTime=1753516664200, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925714309293, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=tui0IVO9FMwB61HHtX5scg==, createTime=1753516664198, updateTime=1753516664198, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925735280816, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1753516664203, updateTime=1753516664203, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925731086511, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1753516664202, updateTime=1753516664202, creator=18614031015, updator=18614031015)])], journalTitle=中国安全科学学报, weixinUrl=null, journalUrl=null, iacademicId=null, status=0, seqNo=null, journalTitleEn=China Safety Science Journal, journalPhotoCn=fkqsFM6VKlHC4gCtS5XqTw==, journalPhotoEn=SHn9HgqSxtJrOcAxqD++4Q==, journalFirstLetter=C, journalRecommend=null, journalNew=null, journalCollection=1, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=null, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/10.16265/j.cnki.issn1003-3033.2024.07.0089, detailUrlEn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/EN/10.16265/j.cnki.issn1003-3033.2024.07.0089, pdfUrlCn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/PDF/10.16265/j.cnki.issn1003-3033.2024.07.0089, pdfUrlEn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/EN/PDF/10.16265/j.cnki.issn1003-3033.2024.07.0089, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于注意力卷积长短时记忆模型的城市出租车流量预测
收藏切换
PDF下载
周新民 1, 2 , 金江涛 3 , 鲍娜娜 4 , 袁涛 3 , 崔烨 4
中国安全科学学报 | 公共安全 2024,34(7): 153-160
收起
收藏切换
中国安全科学学报 | 公共安全 2024, 34(7): 153-160
基于注意力卷积长短时记忆模型的城市出租车流量预测
全屏
周新民1, 2 , 金江涛3, 鲍娜娜4, 袁涛3, 崔烨4
作者信息
  • 1 湖南工商大学 人工智能与先进计算学院,湖南 长沙 410205
  • 2 湘江实验室,湖南 长沙 410205
  • 3 湖南工商大学 前沿交叉学院,湖南 长沙 410205
  • 4 湖南工商大学 计算机学院,湖南 长沙 410205
  • 周新民 (1977—),男,湖南新邵人,博士,教授,主要从事新型智慧城市、商务智能与大数据、互联网安全与服务等方面的研究。E-mail:

Urban taxi traffic flow prediction based on attentive ConvLSTM-ResNet model
Xinmin ZHOU1, 2 , Jiangtao JIN3, Nana BAO4, Tao YUAN3, Ye CUI4
Affiliations
  • 1 School of Artificial Intelligence and Advanced Computing,Hunan University of Technology and Business,Changsha Hunan 410205,China
  • 2 Xiangjiang Lab,Changsha Hunan 410205,China
  • 3 School of Frontier Crossover Studies,Hunan University of Technology and Business,Changsha Hunan 410205,China
  • 4 School of Computer Science,Hunan University of Technology and Business,Changsha Hunan 410205,China
出版时间: 2024-07-28 doi: 10.16265/j.cnki.issn1003-3033.2024.07.0089
文章导航
收藏切换

为解决城市交通拥堵和安全问题,提出一种注意力卷积长短时记忆(ConvLSTM)残差(ACLR)模型,该模型通过结合ConvLSTM、注意力机制和残差结构,分别处理出租车流量的时间、空间、和其他特征,挖掘区域兴趣点(POI)数据对出租车流量的影响,有效提升交通时空特征的提取能力。同时,引入专门的学习元件考虑外部因素和POI密度对交通流量的影响,并利用北京市出租车轨迹数据验证。结果表明:ACLR模型在城市交通流预测中的精度高于差分自回归滑动平均(ARIMA)模型、长短时记忆(LSTM)网络、深度时空残差网络(ST-ResNet)、卷积神经网络(CNN)-残差神经单元-LSTM(CRL)循环神经网络、ACFM等模型,在无POI密度和考虑POI密度的情况下,均有助于提升模型的预测性能,ACLA模型的预测值与真实值基本一致,高峰时段也能与真实值较好地吻合,有效提升交通时空特征的提取能力,降低预测误差,使得交通流量预测性能得到优化。

注意力卷积长短时记忆残差网络(ACLR)模型  /  交通流量预测  /  城市出租车  /  时空特征  /  残差结构

In order to address the challenges of urban traffic congestion and safety,an ACLR model was proposed. By integrating ConvLSTM,attention mechanisms,and residual structures,the ACLR model effectively enhanced the extraction of spatio-temporal traffic features.The time,space and other characteristics of taxi traffic were processed respectively,and the influence of regional point of interest(POI) data on taxi traffic was mined. Additionally,a specialized learning component was incorporated to capture the impact of external factors and point-of-interest density on traffic flow. Using taxi trajectory data from Beijing,the ACLR model demonstrates superior prediction accuracy compared to other models such as the autoregressive integrated moving average (ARIMA) model,long short-term memory (LSTM),deep spatio-temporal residual networks (ST-ResNet),convolutional neural network(CNN)-ResNet-LSTM (CRL),and attentive crowd flow machines (ACFM) in urban traffic flow forecasting,which is helpful to improve the prediction performance of the model without POI density or considering POI density. The predicted value of the ACLA model is basically consistent with the real value,and it can also be in good agreement with the real value during peak hours,which effectively improves the ability to extract traffic temporal and spatial characteristics,reduces the prediction error,and optimizes the traffic flow prediction performance.

attentive convolutional long short-term memory(ConvLSTM) residual network(ResNet)(ACLR)  /  urban taxi  /  traffic flow prediction  /  space-time characteristics  /  residual structure
周新民, 金江涛, 鲍娜娜, 袁涛, 崔烨. 基于注意力卷积长短时记忆模型的城市出租车流量预测. 中国安全科学学报, 2024 , 34 (7) : 153 -160 . DOI: 10.16265/j.cnki.issn1003-3033.2024.07.0089
Xinmin ZHOU, Jiangtao JIN, Nana BAO, Tao YUAN, Ye CUI. Urban taxi traffic flow prediction based on attentive ConvLSTM-ResNet model[J]. China Safety Science Journal, 2024 , 34 (7) : 153 -160 . DOI: 10.16265/j.cnki.issn1003-3033.2024.07.0089
城市化进程加快和城市人口快速增长给城市交通管理带来了巨大压力,容易引发重大交通事故和大范围交通拥堵[1]。出租车作为城市中重要的交通工具,数据量大且范围广,能较好地反映城市交通流量问题。准确预测出租车流量有助于合理规划出行路线,缓解城市交通拥堵压力,进而提高城市交通安全水平[2]。由于出租车流量在空间和时间维度上都具有动态相关性,且流量变化受多因素影响。因此,如何利用海量出行轨迹数据挖掘非线性、复杂的时空特征,进而作出准确预测出租车流量是一个现实而关键的问题。
目前,交通流量预测模型主要包括统计理论模型、深度学习模型等。自回归积分滑动平均模型[3](Autoregressive Integrated Moving Average Model,ARIMA)、卡尔曼滤波模型[4]等统计理论模型能较好地捕获中长期交通流量变化趋势,但对交通流量的随机性变化不够敏感,难以应对复杂的交通流量预测问题。深度学习模型[5]由于具有较强的非线性建模能力被广泛应用于交通预测,如卷积神经网络(Convolutional Neural Network,CNN)[6]、长短时记忆网络(Long Short Term Memory,LSTM)[7]等。然而单一CNN或LSTM都无法充分挖掘交通流量变化的时空特征,基于CNN的方法通常需要大量堆叠才能挖掘数据的全局特征,基于LSTM的方法无法高效处理长序列数据,且易造成信息丢失。段宗涛等[8]将残差卷积网络与LSTM网络结合,构建出租车需求预测模型;KARIM等[9]提出LSTM全卷积网络(LSTM Fully Convolutional Networks,LSTM-FCN),通过LSTM、CNN分别提取交通数据的时间、空间特征;PENG Hao等[10]提出一种基于动态图卷积网络的长期交通流量预测方法,克服了交通流量预测的数据缺陷;XING Lumin等[11]构建了单车道和聚合交通流数据融合驱动的双向LSTM(Data Fusion Powered Bi-Directional LSTM,DFBD-LSTM)模型,预测多车道短期交通流;SHU Wanneng等[12]利用改进的门控循环(Gated Recurrent Unit,GRU)神经网络模型,并采用自适应整流优化器(Rectified Adam,RAdam),加快了模型收敛的速度和稳定性。此外,SHI Xingjian等[13]提出的端到端卷积LSTM(Convolutional LSTM,ConvLSTM)网络结构,与CNN与LSTM的简单堆叠组合相比,具有更高效的时空序列处理能力,然而,简单堆叠ConvLSTM网络层的方法无法有效学习交通数据的全局特征。ZHAO Yiji等[14]通过引入时空位置感知关系推理模块,改进时空图卷积网络,可在位置嵌入的指导下捕获节点特定的模式特征。交通流量影响因素方面,大多研究基于网格分割法[15],通过多分支结构挖掘交通时空相关性,并融合天气、节假日等外部因素构建预测模型。然而,该类方法往往未考虑土地功能因素,如兴趣点(Point of Interest,POI)对交通流量的影响。
鉴于此,笔者拟将POI数据进行栅格化处理,同时构建“学习元件”对POI密度和外部因素进行特征提取与融合;针对堆叠ConvLSTM模型无法充分挖掘交通流量时空变化特征的问题,设计注意力卷积LSTM残差网络(Attentive ConvLSTM-Residual Network(ResNet)ResNet,该模型以ConvLSTM为基础网络,引入残差结构与注意力机制构建多分支结构,分别处理出租车流量的不同信息。并基于出租车轨迹数据验证进行模型验证,以期有效降低预测误差。
充分考虑出租车流量时空特性,是提高预测效果的关键。ACLR模型主要通过以下3个步骤加强时空特征提取:①考虑交通数据在空间以及时间上分布的邻近性、周期性、趋势性,构建基于历史数据的时空图像,构造以不同时间戳为索引的输入数据集,实现交通流量变化在时间层次的细化。②基于嵌入式一体化的ConvLSTM网络,引入残差结构和注意力机制构建ConvLSTM-ResNet模块,同时构建基于残差改进的LSTM-ResNet模块,组合以上2个模块以提高对交通流量时空特征的提取能力。③设计“学习元件”,同步学习POI密度特征和外部影响因素对交通流量变化的影响,连接主体结构构建一个综合考虑空间、时间和外部关系的高效城市交通流量预测模型。ACLR预测模型框架如图1所示。
出租车流量预测实质上是解决一个包含空间位置和时间序列结构的问题,即依据长期历史轨迹数据 { X t t = 0,1 n - 1 }来预测未来轨迹数据Xn。从该角度看,出租车流量预测问题可视为时空序列预测问题。
假设需要预测由MN列组成的网格空间区域上的出租车流量,M×N网格中的每一单元格都有P个不同意义的测量值,且随时间不断变化。因此,用 X R p × M × N表示任意时刻收集到的观测值,R表示观测到的特征域。历史观测值组成的集合可表示为 X ^ 1 X ^ 2 X ^ t。出租车流量时空序列预测问题可理解为预测未来长度k的序列,如下式:
X ^ t + 1 X ^ t + k = a r g   m a x X ^ t + 1 X ^ t + k   ρ ( X t + 1 X t + k | X ^ t - j + 1 X ^ t - j + 2 X ^ t )
交通流量预测需综合考虑时间和空间的动态关联性,为准确预测城市出租车流量,采用序列快照方式处理出行轨迹数据。首先,根据经纬度将城市划分为 I × J个网格,每个网格代表一个交通区域,同时也作为时空图像的像素点。然后,以 t为时间间隙,统计该时段所有网格的累计流量。其中,每个时段都包含流入图和流出图2个图像,流入代表进入一个区域的出租车流量,流出代表离开一个区域的出租车流量。
位于第i行和第j列的单元格(ij)在t时间间隔内的流入出租车交通量 X t y i j和流出出租车交通量 X t z i j分别定义为:
X t y i j = T 1 T { S > 1 | P n ( i j ) P n + 1 ( i j ) } | X t z i j = T 1 T { S > 1 | P n ( i j ) P n + 1 ( i j ) } |
式中: T为轨迹集合; T 1 T中的一条轨迹,即一系列地理空间坐标; P n T轨迹集合中的轨迹点, P n ( i j )表示点 P n位于单元格 ( i j )内; S为集合的基数。式(2)中,某一时刻 P n不属于单元格(ij),且下一时刻属于单元格 ( i j ),就将其统计为 ( i j )区域的流入出租车流量;某一时刻 P n属于单元格 ( i j ),且下一时刻不属于单元格 ( i j ),就将其统计为 ( i j )区域的流出出租车流量。在时间间隔内,所有 I × J区域的流入量和流出量都可表示为 X t R 2 × I × J
以30min为时间间隔建立时空矩阵 P t。如下式,矩阵中的每一项 X i j表示 t时段内,网格区域 ( i j )的出租车流量。
P t = P 11 P 12 P 1 j   P 21 P 22 P 2 j       P i 1 P i 2 P i j
出行轨迹数据转化为时空图像具体操作如图2所示。先统计单位时间间隔内所有网格区域的流入流出量,依据时间间隙转化为流入流出矩阵,再转化为双通道时空图像。图像中颜色越深表示该网格区域流入和流出出租车数量越多。
选用北京市出租车轨迹数据集,该数据集采用32×32的网格对北京市进行栅格化处理,每个交通小区实际大小为1km×1km。将1天分为48个时间段,统计每30min内所有网格区域流入和流出出租车数量,其中,一张“时空图像”代表一个时间段所有网格区域的出租车流量。
交通流量普遍受出行规律、天气、节假日等因素影响,会表现出不同状态。分析数据特征,设计网络模型结构,是交通流量预测的核心工作,出租车流量预测同样如此。
1) 空间特征。出租车流量在空间上具有局部相似特征,区域间相互影响,且间隔距离越小,影响程度越大。
2) 时间特征。邻近性:对同一个区域,某一时间间隙的流量受前几个时间间隙流量影响,并影响后几个时间间隙流量,且流量在邻近隙内保持相对稳定,不会有大范围波动。周期性:每日交通峰值规律性出现,在工作日(周一到周五),人们的作息时间相对固定,总在一天中相同的时间段内出行,这导致早晚高峰出现时间大致相同。趋势性:在相对较长的一段时间内,由于城市人口增长、经济发展等要素,人们的出行需求将发生变化,这导致出租车流量的整体趋势将随之变化。
3) 其他特征。节假日、天气等外部因素对市民出行选择造成影响,如阴雨天气会减缓通行速度,影响区域交通流量。POI密度一定程度上反映该区域建设发展情况,考虑到出租车流量与城市发展的紧密联系,在模型设计中考虑POI密度对出租车流量的影响。
考虑出租车流量的邻近性、周期性、趋势性特征、外部因素以及POI密度,因此,需构建5类输入数据集来模拟出租车流量的不同特征。各输入数据集的定义及构建过程如下:
邻近性(closeness):
S c = ( X t - l c X t - ( l c - 1 ) X t - 1 ) R 2 × I × J
周期性(period):
S p = ( X t - l p · p X t - ( l p - 1 ) · p X t - p ) R 2 × I × J
趋势性(trend):
S t r = ( X t - l t r · q X t - ( l t r - 1 ) · q X t - q ) R 2 × I × J
式中: l c l p l t r分别为邻近性、周期性、趋势性的序列长度; c为邻近性时间跨度为30min; p为周期性时间跨度置为1天; q为趋势性时间跨度为1周。
“closeness、period、trend”3类数据集选取过程如图3所示。
外部因素: { E 0 E 1 E n - 1 },包含天气、温度、风速、节假日情况。以日期为索引构建外部因素数据集,其中,天气、节假日采用One-hot编码处理,温度、风速采用Min-Max归一化处理。
POI密度ρ,基于207 892条北京市POI数据,使用ArcGIS软件进行栅格化处理,统计各网格单元POI数量,Min-Max归一化处理后转化为32×32的POI密度图像。
交通数据具有较强的时空相关性,使用单一的CNN或LSTM无法充分提取交通流量时空特征,同时为避免CNN、LSTM堆叠组合造成的特征信息丢失,选取ConvLSTM作为模型的基础结构。ConvLSTM网络的本质是在LSTM中嵌入CNN,与普通CNN和LSTM堆叠的组合模型相比,具有更强的时空特征提取与融合能力。
由于简单堆叠ConvLSTM的方式难以对交通数据的邻近性、周期性和趋势性特征进行建模,因此,参考残差结构同时引入注意力机制,改进ConvLSTM模型,构建ACLR模型。该模型为多支输入结构,通过分别学习出租车流量数据的邻近性、周期性和趋势性特征,设计“学习元件”学习外部因素与POI密度对交通流量的影响。模型具体结构如图4所示。
图4可知:特征数据集输入后先通过ConvLSTM进行出租车流量时空相关性学习。
ConvLSTM模型对时空特征的提取由2个关键部分协同完成,分别是时间轴上的卷积结构和记忆组件中的细胞综合状态 C T t - j ( j = 1,2 q )。出租车流量空间特征提取由卷积结构完成,综合细胞综合状态 C T t - j不断更新和积累信息并向下传递,能有效挖掘长时间跨度出租车流量时间特征。
出租车流量时空信息通过遗忘门、输入门、输出门进行筛选和更新。其中,依据上一时段的状态信息 h T t - j - 1 f、细胞综合状态 C T t - j - 1 f和当前输入 ( X T t - j - 1 f ) ',遗忘门与输入门将分别产生 θ T t - j f ( C T t - j - 1 f被保留的比例)及 S T t - j f(候选状态 C ^ T t - j - 1 f被选择的比例),从而将无用信息删除,有用信息进行更新。输出门依据 h T t - j - 1 f ( X T t - j - 1 f ) '来判断当前流量状态特征,表示为 O T t - j f,并依据状态特征产生当下状态信息 h T t - j f。根据数据数量,重复该过程,最终提取出出租车流量时空特征 h T f
ConvLSTM完成初步特征学习后,连接ConvLSTM-ResNet模块进一步提取时空特征。
1) ConvLSTM-ResNet模块。图4中,ConvLSTM-ResNet模块由“Part-a”和“Attention”2部分组成。前者负责学习流量数据的时空依赖性,后者负责学习注意力权重,提取流入和流出交通量之间的关系。
“Part-a”部分参考基础残差结构,构建3层分支。ResNet通过在网络中增加跳跃连接,有效缓解随着神经网络深度增加所带来的信息丢失、梯度丢失等问题。
第1层分支由3个ConvLSTM组成,第1个ConvLSTM的卷积核大小为(1,1),目的是为了限制通道数量,提高模型速度;由于出租车流量数据非常稀疏,第2个和第3个ConvLSTM采用(1,3)(3,1)的卷积核,目的是为了提取特定区域与邻近区域之间的相关性,2个ConvLSTM相互补充。第2层分支包含1个ConvLSTM,为第1层分支提供信息补充。1、2层分支将特征信息进行“concat”相连后与第3层分支携带的残差信息进行相加,组成残差结构。为提高计算效率,“Part-a”部分学习到的特征随后连接批量归一化(Batch Normalization,BN)层进行批量规范化处理来加速训练。
“Attention”部分通过挤压-激励(Squeeze and Excitation,SE)机制计算通道间的相关性,筛选出更有价值的通道信息,具体工作过程如图5所示。该操作包含2层分支,第1层分支使用全局池化层(Global Average Pooling,GAP)压缩参数,再连接一个全连接层来提取各通道的特征表示,该全连接层以修正线性单元(Rectified Linear Unit,Relu)为激活函数来实现信息的非线性映射,再连接下一个以Sigmoid为激活函数的全连接层,实现对通道的注意力权重分配,进而学习通道间重要的关联信息。
第2层分支是包含初始输入的残差层,图4中,将初始输入与第1层分支进行逐元素相乘融合后连接ConvLSTM。模型通过ConvLSTM网络和LSTM-ResNet模块来加深网络深度,从而更好地捕获交通数据的全局相关性。
2) LSTM-ResNet模块。LSTM是循环神经网络的一种变体。LSTM通过精妙的门控制将短期与长期记忆结合起来,并一定程度上解决了梯度消失的问题,能学习长期依赖信息。LSTM在捕获当前时刻特征的同时,也考虑前一时刻特征。通过遗忘门来决定上一时刻需要遗忘的信息,通过输入门来决定当前时刻需要保留的信息,最后通过输出门得到预测输出。
考虑出租车流量数据的强时间依赖性,使用LSTM-ResNet模块来提高模型对时间特征的学习能力。该模块由3分支组成,第1分支由2个LSTM堆叠而成,提取深层时间特征;第2分支包含一个LSTM;第3分支携带残差信息;3分支通过元素相加进行特征融合,减少信息丢失,进而更好地学习出租车流量的长时间依赖性。
随后,将“closeness、period、trend”3分支的特征信息 X m X n X k融合得到外部数据 X m n k,过程如下式:
X m n k = ( X m X n X k )
式中 为按元素相加操作,即对应位置的元素相加。
3) 学习元件。为更好地学习外部因素和POI密度对出租车流量的影响,设计学习元件来提取特征信息并进行融合,如图6所示。学习元件包含特征提取与融合过程。外部因素由2层Dense进行特征提取,第1层Dense提取浅层关联性,第2层Dense挖掘深层的特征信息,提取的特征信息被重塑为I×J后与POI密度进行融合,表示为轨迹数据 O m
外部因素和POI密度从特征提取到融合,最后映射到各网格单元的过程如下式:
O m = f ( ρ p o i E t )
式中 E t为外部因素。
外部数据与轨迹数据通过元素相加,得到时间间隙的流量预测矩阵,过程如下式:
X ^ t = t a n h ( O m X c p t )
式中: t为时间间隙; X ^ t为流量预测矩阵。
ACLR模型以下式作为损失函数,通过不断调整参数 θ,使预测流量矩阵和真实流量矩阵之间的均方误差最小化,完成模型训练。
L ( θ ) = X t - X ^ t 2 2
ACLR模型流程如下所示。其中,第1—6步为从原始序列数据中构造训练实例,第7—11步为模型训练与优化。
如算法所示,每一次迭代都将一批样本输入到模型中,学习模型参数,使得损失最小化,当达到最大迭代次数后,训练中止。训练好后的模型就可以用于城市出租车流量预测。
出租车流量是城市交通流量的重要组成部分,出租车流量预测研究可为智能交通系统构建提供切实可行的解决方案。选用出租车轨迹数据集来研究城市交通流量,主要考虑如下:①出租车在运营期间,不会占用城市停车资源,流动性大,能较好地反映城市路网中交通流动情况。②出租车车型丰富,数量大且轨迹覆盖范围广,能较为准确地反映出城市流量变化特征。③出租车数据充分且完整性高,通过运营平台获取较为方便,且处理后数据质量高,便于试验仿真。
北京出租车轨迹数据集合见表1
北京出租车轨迹数据集包括34 000多辆出租车的轨迹数据和北京市外部气象数据(风速、天气状况、温度、节假日)。记录时间分别为:2013年7月1日—10月30日,2014年3月1日—6月30日,2015年3月1日—6月30日,2015年11月1日—2016年4月10日。预处理数据集中错误和异常交通流量数据,确保研究范围内所有时间戳数据有效性。外部数据集见表2
选用数据集的80%作为训练集,其余20%作为测试集,即最后4周数据(48×7×4个时段数据)作为测试集。
通过均方根误差(Root Mean Squard Error,RMSE)和平均绝对百分比误差(Mean Absolute Percentage Error,MAPE)来评估城市出租车流量预测模型性能。
R M S E = 1 N i = 1 N ( X i - X ^ i ) 2
M A P E = 100 % N   i = 1 N X i - X ^ i X i
式中: X i为第i个样本的真实值; X ^ i为第i个样本的预测值。
ACLR模型使用Python语言,基于tensorflow和keras框架搭建,选用Adam优化器进行训练。经多次优化,模型超参数设置如下:学习率设置为0.005;batch-size设置为32;epoch设置为200。
卷积结构的空间感知能力由卷积核的大小决定。考虑到ConvLSTM模型计算过程中包含卷积操作,因此,改变模型中ConvLSTM网络卷积核大小,验证不同尺寸的卷积核对预测效果的影响。试验结果如图7所示,在2×2、3×3、4×4、5×5这4种卷积核尺寸中,卷积核尺寸为5×5时,RMSE值最低。因此,将ACLR模型滤波器尺寸设置为5×5。
滤波器的数量与特征提取密切联系。分别将滤波器数量调整为8、16、32进行试验,并记录RMSE值,试验结果如图8所示。当滤波器数目为32时,RMSE值最低,此时模型预测效果最佳。因此,ACLR模型中滤波器数量设置为32。
选定参数后,观察RMSE值随迭代次数的变化曲线,如图9所示。随着迭代次数不断增加,RMSE值不断降低,100次epoch之后,RMSE值变化趋于平缓,模型实现收敛。
1) 模型变体。为验证ACLR模型各改进模块的有效性,评估4种ACLR的退化模型。4种退化模型分别如下:
变体1:同时缺少ConvLSTM-ResNet模块和LSTM-ResNet模块的ACLR退化模型,即只使用ConvLSTM模块的模型。
变体2:缺少ConvLSTM-ResNet模块的ACLR退化模型。
变体3:不考虑POI密度的ACLR退化模型。
变体4:缺少 LSTM-ResNet模块的ACLR退化模型。
模型各模块评估如图10所示。通过残差结构和注意力机制改进的ConvLSTM-ResNet模块对于模型预测性能具有较大提升,表明该模块能有效提取出租车流量数据的时空特性;对比变体3和ACLR的RMSE值,可知考虑城市区域的POI密度能降低预测误差;对比变体4和ACLR的RMSE值,发现组合LSTM-ResNet模块后能降低预测误差,表明该模块能提高模型的时间特征获取能力。
2) 模型对比。为验证模型的有效性,选取5个模型:差分整合移动平均自回归模型(Autoregressive Integrated Moving Average model,ARIMA)、LSTM网络、深度时空ResNet(Deep Spatio-Temporal ResNet,ST-ResNet)、CNN-ResNet-LSTM(CRL)循环神经网络、ACFM (Attentive Crowd Flow Machines)与提出的ACLR模型进行性能对比,见表3
表3可知:ARIMA和LSTM模型的预测误差较大;ST-ResNet采用残差的CNN结构加深网络,具有较强的空间特征获取能力,但无法学习时间特征,因此,预测效果较为一般;CNN-ResNet-LSTM模型通过残差CNN和LSTM堆叠组合实现交通流量时空特征提取,对比ST-ResNet模型,表现出更优的预测效果;ACFM模型引入注意力机制改进LSTM,提高了预测效果。表3中,ACLR模型在2种不同的条件下进行了比较,分别是:①在无POI密度的条件下,ACLR模型与其他5个模型的对比中表现出更优的预测性能,说明细化了交通流量特征,充分提取及融合时空特征能有效降低交通流量预测误差。②考虑POI密度的ACLR模型与未考虑POI密度的ACLR模型相比,RMSE值降低0.79%,MAPE值降低1.26%,说明在考虑POI密度的条件下有助于提高模型的预测性能。以上分析说明,ACLR模型具有更优的预测性能,能够较好完成城市交通流量预测任务。
3) 模型预测。选择某一交通流量较大的网格区域进行流量预测,工作日和周末的交通流量预测值和真实值的对比如图11所示。通过预测误差分析,工作日预测RMSE值为14.34,MAPE值为16.33%;周末预测RMSE值为15.21,MAPE值为17.46%。由于工作日的交通流量表现出更明显的规律性,模型能够较好地学习规律性变化,因此,工作日预测效果较好。对于规律性较弱的周末,ACLR模型也具有不错预测效果。
图11可知:ACLR模型的预测值与真实值基本一致,且高峰时段也能与真实值较好地吻合。
1) 针对城市出租车流量预测问题,考虑区域POI密度特征,设计一种高效的出租车流量预测ACLR模型。通过将POI数据进行栅格化处理,挖掘区域POI密度对出租车流量的影响,同时为城市交通流量预测提供一种新思路。
2) 通过设计“学习元件”提取外部因素和POI密度对城市出租车流量的影响,同时进行深度特征融合,有效提高了预测模型的可扩展性。
3) 基于ConvLSTM网络,引入残差结构和SE机制,构建一个综合考虑空间、时间、外部因素和POI密度的高效城市出租车流量预测模型,有效降低了预测误差。
4) 在未来的研究中,将尝试使用图神经网络等模型,考虑复杂路网信息,提高交通流量预测效果;选用不同交通数据分析,增强模型泛化能力。
  • 国家社会科学基金资助(21BGL231)
参考文献 引证文献
排序方式:
[1]
姬浩, 王永东, 李佩, 等. 事故车辆影响下的城市三车道道路交通流仿真[J]. 中国安全科学学报, 2021, 31(3):112-120.
JI Hao, WANG Yongdong, LI Pei, et al. Traffic flow simulation of urban three-lane road considering influence of accident vehicle[J]. China Safety Science Journal, 2021, 31(3):112-120.
[2]
熊晓夏, 刘擎超, 沈钰杰, 等. 基于LSTM-BF的高速公路交通事故风险模型[J]. 中国安全科学学报, 2022, 32(5):170-176.
XIONG Xiaoxia, LIU Qingchao, SHEN Yujie, et al. Study on risk model of highway traffic accidents based on LSTM-BF[J]. China Safety Science Journal, 2022, 32(5):170-176.
[3]
SMITH B L, WILLIAMS B M, OSWALD R K. Comparison of parametric and nonparametric models for traffic flow forecasting[J]. Transportation Research Part C, 2002, 10(4):303-321.
[4]
GUO Jianhua, HUANG Wei, WILLIAMS B M. Adaptive Kalman filter approach for stochastic short-term traffic flow rate prediction and uncertainty quantification[J]. Transportation Research Part C, 2014, 43(2):50-64.
[5]
ZHAO Ling, SONG Yujiao, ZHANG Chao, et al. T-gcn: a temporal graph convolutional network for traffic prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 21(9): 3848-3858.
[6]
MA Xiaolei, DAI Zhuang, HE Zhengbing, et al. Learning traffic as images: a deep convolutional neural network for large-scale transportation network speed prediction[J]. Sensors, 2017, 17(4):DOI:10.3390/s17040818.
[7]
MA Xiaolei, TAO Zhimin, WANG Yinhai, et al. Long short-term memory neural network for traffic speed prediction using remote microwave sensor data[J]. Transportation Research Part C: Emerging Technologies, 2015, 54: 187-197.
[8]
段宗涛, 张凯, 杨云, 等. 基于深度CNN-LSTM-ResNet组合模型的出租车需求预测[J]. 交通运输系统工程与信息, 2018, 18(4):215-223.
DUAN Zongtao, ZHANG Kai, YANG Yun, etal. Taxi Demand prediction based on CNN-LSTM-ResNet hybrid depth learning model[J]. Journal of Transportation Systems Engineering and Information Technology, 2018, 18(4):215-223.
[9]
KARIM F, MAJUMDAR S, DARABI H, et al. LSTM fully convolutional networks fortime series classification[J]. IEEE Access, 2017, 6: 1662-1669.
[10]
PENG Hao, DU Bowen, LIU Mingsheng, et al. Dynamic graph convolutional network for long-term traffic flow prediction with reinforcement learning[J]. Information Sciences, 2021, 578:401-416.
[11]
XING Lumin, LIU Wenjian. A data fusion powered Bi-Directional long short term memory model for predicting multi-lane short term traffic flow[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(9):16810-16 819.
[12]
SHU Wanneng, CAI Ken, XIONG Naixue. A short-term traffic flow prediction model based on an improved gate recurrent unit neural network[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(9):16654-16 665.
[13]
SHI Xingjian, CHEN Zhourong, WANG Hao, et al. Convolutional LSTM network: a machine learning approach for precipitation nowcasting[C]. International Conference on Neural Information Processing Systems, 2015: 802-810.
[14]
ZHAO Yiji, LIN Youfang, WEN Haomin, et al. Spatial-temporal position-aware graph convolution networks for traffic flow forecasting[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(8):8650-8666.
[15]
GUO Shengnan, LIN Youfang, LI Shijie, et al. Deep spatial-temporal 3D convolutional neural networks for traffic data forecasting[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 20(10):3913-3926.
2024年第34卷第7期
PDF下载
337
142
引用本文
BibTeX
文章信息
doi: 10.16265/j.cnki.issn1003-3033.2024.07.0089
  • 接收时间:2024-01-12
  • 首发时间:2025-07-09
  • 出版时间:2024-07-28
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2024-01-12
  • 修回日期:2024-04-18
基金
国家社会科学基金资助(21BGL231)
作者信息
    1 湖南工商大学 人工智能与先进计算学院,湖南 长沙 410205
    2 湘江实验室,湖南 长沙 410205
    3 湖南工商大学 前沿交叉学院,湖南 长沙 410205
    4 湖南工商大学 计算机学院,湖南 长沙 410205
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/10.16265/j.cnki.issn1003-3033.2024.07.0089
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏