城市快速路交通拥挤识别方法姜桂艳.pdf

第6卷　第3期2006年9月交 通 运 输 工 程 学 报Journal of Traffic and Transportation EngineeringVol.6　No.3Sept. 2006là°ù:2005-11-20Á[":SE1– SÐÁ×Ä["(50338030);SE1– SÐÁ["(50178072)T€eº:{I(1964-),o,É=¦,vНq,VYŒYýñùîÓcI|:1671-1637(2006)03-0087-05ìg yÎ^ŒYš*MYZE姜桂艳,冮龙晖,王江锋(vÐŒYÐý,É=　130025)K　1:为了从海量动态交通数据中快速识别路网中存在的交通拥挤,通过分析拥挤的特征模式和各种数据挖掘技术的特点后,设计了一种适用于城市快速路的交通拥挤自动识别方法该方法将占有率、速度和流量三个基础交通流参数进行组合得到新的特征变量,运用优化的多层前馈神经网络模型对特征变量进行处理来判断是否有拥挤发生,通过分析模型输出结果的变化趋势区分常发性拥挤和偶发性拥挤。

模拟数据和实测数据对比结果表明,该方法可以识别城市快速路上发生的交通拥挤,具有良好的实用性1oM:交通信息工程;交通拥挤识别;数据挖掘;人工神经网络;交通状态Ïms Ë|:U491　　　ÓDSM’:ATrafficcongestionidentificationmethodofurbanexpresswayJiang Gui-yan, Gang Long-hui, Wang Jiang-feng(School of Transportation, Jilin University, Changchun 130025, Jilin, China)Abstract:In order to quickly identify traffic congestion from mass dynamic traffic information, trafficcongestion pattern and the characteristics of various data mining technologies were analyzed, an auto-identifying method of urban expressway trafficcongestion was designed. The flow , speed and occupancyof expressway were combined into several new eigenvectors, optimized multi-layer feedforwardperceptron model was adopted to classify the eigenvectors during congestion and non-congestion,recurrent congestion and non-recurrent congestion could be distinguished by analyzing the variances ofthe model outputs, the method was tested with simulated data and actual data from an urbanexpressway. The result shows that the method has great practicability and can identify congestion stateson urban expressway correctly. 2 tabs, 5 figs, 11 refs.Key words:traffic information engineering ;traffic congestion identification;data mining;artificial neural network;traffic stateAuthorresume:Jiang Gui-yan(1964-), female, professor, 86-431-5095505, jianggy @public. .0　„　ýÛ"ÏSöÜ6¥?Z,Cµ¥¡^ŒY!@XÜ‚?¡@ŒY9É¥³1,ŒYš*XÜî¹òvÏìgë"¥]Ù5[ 1] 。

¹hŒYš*{ Ÿ¥å>,y ë¡^ŒYš*çÞ„Þ"d^dÈA1¥ÍM Ÿ,Û"Ս˜ŒY_©/Œ¥C„‹¨,ŒY5 ؆Êë"¥ö1Ù5‚^Â…¤ŒY” ,7^Â…HW=VZ¥L=” Ï4|µ¨¥,ƒ9^y ëŒYš*çÞ„Þ"d¥1o/Œ­B” ‹ !(Data Mining,DM)^20W@–¥” s/Œ,ñ V[Vv˜”  o”  oÏ1î4|?Cµ¨¥©MDMT¹Bñ–¥Ð SŒ­‹¨/Œ,ø ‚]5×¥©M,­^”  o、¦ýÆ?、” Ød9、 VjÄ、i›9Ø©Z륩M,ž¸5×¥ %¦¦ûîÏò"×1¥˜ä[2] 'Ó|” ‹ !/Œ‹¨žìg yÎ^ŒYš*ÿ¥1îMY,ŒYš*1îMYZEÉ›!9。

1　” ‹ !” ‹ !^Vv¥、‚›†¥、µ.2¥、´¥、Û¥” Ï,4|…cÏ¥,¦ÌY5‚©¡¥,Œ»^.µ¨¥„©M¥Vñ” ‹ !ðS”  V[^²Ä¥(Â1"˜”  oÏ¥” ),9 V[^ö²Ä¥(ÂÓ'、m™、m^” ©),.À^sƒ©Ž¥s˜” ‹ !¥©M V[¨¿5 Ø、°ØªÄ、%¦¦û、Vñ eÅ、” 1&¥»¹©” ‹ !¥ýTVñ VvÁs¹4ñ„ç:Ù5çl(Task Definition)、” l"„ç) Ø(DataPreparation and Preprocessing)、” ‹ !、ØE³›„²T¥³d#ß(Interpretation and Evalua-tion)[3] ” ‹ !^Bñ‚Q Œ¥Vñ,ÂT‚?¤žù¬¥²T,5³1ò„ç¥ýT=¸É›€ „Øœ,°ž¡@¨¼1p¹¸。

” ‹ !/Œ^[¦ýÆ?ÐÐZE¹$,[”  o¹ùî`™î¥B"” s/Œ¥9ë[ 4] ,Bî Vs¹‚/Œ、"†‚/Œ、¦ý*Ü©Ž/Œ、d9s/Œ、´”Ð/Œ„ VjÄ/Œ©6 Ë2　ìg yÎ^ŒYš*1îMYØE2.1　ŒYš*T¥+˜sŒYš* Vs¹È?Ÿš*„}?Ÿš*È?Ÿš*^·®¿ŒY³pÑV¡^Y›? ï7Á3¥š*,V‚š*žš*³1Bñ É¥HWVñ,š*?3-ªŒY@¥3ñ'•”(@、΄]µq)¥MÄ^vÉ¥¹?]B3 gÏAUƒ3ñ•”¥MÄVñ,Üa’¥1èMЪ,²Tnm1}?Ÿš*^®¿l?ŒYYq/î¡^Y›? ïH/†i®¿’H¥ŒY³p7„¥š*,y7Yq?3-ªŒY@3ñ'•”¥MÄ^lM¥。

ÜVЍM]¥MÐ,}?Ÿš*Hq/ŒY@3ñ'•”¥MÄVñnm2Ït1^ŒYYq?3¥H Y, t2^YqŒY@Á3¥•Y³M#_©_8C Ÿ¥H Y[5-6] m1　È?ŸŒYš*TFig. 1　Traffic pattern of recurrent congestionm2　}?ŸŒYš*TFig. 2　Traffic pattern of non-recurrent congestionô [s, V[“¸^‘YŒYš*¥Ë˜,'ÂT_©ž¥ŒYš*-ªŒY@3ñ'•”MÄ´ñ+çS¶=,5 V[‘羌Yš*¹È?ŸŒYš*;’MÄ´ÑV!ç¥S¶,ªüNHŒYÿïClMŸMÄ,5 Vª¹?3}?Ÿš*V@wL9 V[ AÈ?Ÿš*„}?Ÿš*¥uY:}?Ÿš*?3H,Yq¹Ä¸Í¥ŒYövv®¿N^L=Y›?ï;7È?Ÿš*YÈ?3¿Ú‚ù¥Þà^,NHŒYö¤Í¿¡^¥Y›? ï。

2.2　” ‹ !/Œ¥Ê4ÐØE!9yÎ^^v˜ìgŒY"d¥Fî†s­B,^ìgŒY5 Ø¥×Ä æ¨òÕ_©/Œ V[¤ yÎ^ŒY@¥v” ,ùî?VZ”Ï yÎ4|ŒYÿ¥/Œ^“µA1¥” ‹ !/ŒÏ,®¿ANN?E¦7¥:k„) Øÿ?,üÉVZ¥” Ï4|µ¨¥©M,¸5פžµr¥‹¨Ïª- Œ*Ü©Ž(Multi-Layer Feedforwar。