抄録
本論文では, 東京首都圏の電車を利用する通勤通学客を対象として, 利用者均衡に基づいて, 出発駅から到着駅への交通需要をほぼ時刻表の通りに運行される電車に配分するモデルを提案する.乗客は朝の定期券利用者のうち約700万人を対象とし, 電車はJR線私鉄線合計128路線の始発電車から午前10時台に終着駅に到着する通勤電車のほとんど(約7500本)を対象とする.電車のダイヤおよび交通需要が時間依存である場合を扱うために, 電車の時刻表を基に, 各駅における各電車の発着をそれぞれ頂点で表し, 駅間の電車の運行を頂点間の枝で表してネットワークを作成する.これによって, 確定した離散的な時間の経過を頂点間の移動で表すことができるので, 電車ネットワーク上の移動を計算する問題を静的な利用者均衡配分問題として定式化することが可能となる.そして, 2000年に行われた大都市交通センサスデータからOD交通需要を求めて利用者均衡配分を計算し, 同じセンサスに記録された実際の乗客の移動と比較して計算結果の妥当性を確かめる.その上で, 応用問題として時差出勤による混雑緩和の可能性を調べること, 南北線全線開通にともなう乗客数変化の予測を行って実際の利用数と比較することを行う.ここで提案した手法によって, 首都圏のような広い範囲の公共交通機関を利用する移動に対して, 詳細な検討を行うことができるひとつのツールが得られたと考えている.
In this paper we propose a traffic assignment model based on user equilibrium principle for commuters traveling through the public railway network in Tokyo Metropolitan area. Our model covers about seven million commuters in seventy five hundred trains traveling along 128 railway routes of JR and other private railway companies. Since such census data of commuter traffic that describes for each commuter his/her origin station, departure time and destination station are available, we formulate this problem as being time dependent. In order to treat the time dependent problem where OD traffic demand depends on time and commuters are assigned to each train running according to a timetable, we adopt such approach to construct a time-space network which honestly expresses the timetable of trains in the network. We extend a usual network to a time-space network by representing each train arrival/departure at a station by a node, each train travel between successive stations by an edge. Using this kind of network, the time dependent problem could be formulated as to find static equilibrium flow pattern being realized through "user-optimal" route choice of commuters, i.e. Wardrop's first principle. In order to obtain OD traffic demand as an input to the traffic assignment problem, we edit the census data for commuter traffic in Tokyo area in 2000. The calculated results show good coincidence with the commuters' flow pattern reproduced by aggregating commuters on every train according to records in the census data. After this preliminary calculation, we considered two applications to demonstrate usefulness of our model. One is to forecast how many commuters would change their routes to use newly opened railway line "Nanboku-sen". The other is to estimate the effect of staggered departure of commuters to their offices in order to reduce congestion during the morning rush hours. We believe that we could propose a useful tool to analyze precisely commuter traffic flow in a widespread public transportation network.
