ANALYSIS OF CAR BEHAVIOR IN MATSUYAMA CITY

Transcription

1 ANALYSIS OF CAR BEHAVIOR IN MATSUYAMA CITY (USING PROBE PERSON DATA) TEAM G 16-Sep 2018 T h e U n i v e r s i t y o f To k yo, Jap a n & T h e U n i v e r s i t y o f C e n t r a l P u n j a b, L a h o r e, P a k i s t a n

2 GROUP INTRODUCTION Ahsan Umer Ali Habib Khan Shehroze Khalid Abbasi Babar Yamamoto Shotaro University of Central Punjab, Lahore Pakistan The University of Tokyo, Japan 1

3 TARGET AREA MATSUYAMA CITY Located in Ehime Prefecture on Shikoku Island (Western part of Japan) Capital and Largest City of Ehime Prefecture with Population = 516, 643 (as of January 1, 2014), Area = m 2 and No. of Households = 229,916. According to the PT survey conducted in 2007, car usage is more than half showing the Expanding Car Usage in Matsuyama City 2

4 DATA CHARACTERISTICS & PREPARATION Probe Person (PP Data) Data (Given for Exercise) Network Data FOCUS Central Area of Matsuyama City Location Data (Sequential GPS Log i.e. Latitude & Longitude) Combination Path Choice Can be Assumed Trip Data (OD, Duration, Mode, Purpose) Extraction of Data Extracted the OD Data and Network Data for Central Area of Matsuyama City Problem Approach Reason of Extraction How to assume the Actual Path? Map Matching Algorithm Data Preparation for reducing computational load on RL Model 3

5 INTRODUCED POLICIES Focus on Central Area Attempts to make some roads Pedestrian Friendly Car Flow Restraint (In central Area) 4

6 Percentage shared by each Mode Percentage shared by each Mode PRELIMINARY ANALYSIS Whole City Central Area Legend Purpose of Trip Purpose of Trip 5

7 FORMULATION OF MODEL Existing Route Choice Models IIA, Path Enumeration Sequential Route Choice Models Spatial Cognition about downstream, Degree of Spatial Cognition β-scaled RECURSIVE LOGIT MODEL; OYAMA AND HATO 2016 Consider a directed connected graph; G = A, N, where A set of links, N set of nodes The instantaneous random utility of a link a j condition on being in state a j 1 is given by, u n a j a j 1 = v n a j a j 1 + με n a j The total utility of link a j given the state a j 1 is formulated by sum of the instantaneous utility u n a j a j 1 and maximum expected downstream utility up to the destination link d, denoted as value function V n d a j equation (Bellman, 1957); V n d a j = E max a j+1 A a j v n a j+1 a j + βv n d a j+1 + με n a j+1 a j εa and defined by the Bellman β is time discount rate represents the spatial cognition of driver for downstream links LINK CHOICE PROBABILITY (MULTINOMIAL LOGIT MODEL) P n d a j+1 a j = 1 eμ v n a j+1 a j +βv d n a j+1 1 σ aj+1 eμ v n a j+1 a j +βv d n a j+1 A a j 6

8 PRELIMINARY ESTIMATION RESULTS Variables Parameters t-value Travel Time ** Right-Turn Dummy ** β ** L (0) LL Rho-Square Adjusted Rho-Square

9 INTRODUCED POLICIES PROPOSED POLICY Focus on Central Area of Matsuyama City Making Transit Mall (A Pedestrian Friendly area) Car Flow Restraint (In central Area) PRECEDING POLICY Hanazonomachi Avenue Reduced the No. of Car Lanes from 4 to 2 8

10 TRAFFIC ASSIGNMENT β-scaled RECURSIVE LOGIT MODEL; OYAMA AND HATO 2016 u n a j a j 1 = Ѳ tt a j a j 1 TT + Ѳ RT a j a j 1 RT + με n a j P n d a j+1 a j = σ aj+1 1 eμ v n a j+1 a j +βv d n a j+1 1 eμ v n a j+1 a j +βv d n a j+1 A a j 1 e V n,t d a j = ቐ 0 μ σ a j+ 2 A a J + 1 e v n,d a j+ 2 a j+1 +βv n d a j+ 2 a j+1 d a j+1 = d z = Mz + b z = (I M) 1 b Link Flows Equation: I P T F = G 9

11 POLICY SIMULATION (CASE-0) Central Area (Without any change) 10

12 POLICY (CASE-1) Prohibit Cars in two (2) links (The road in front of Central Station) 11

13 POLICY SIMULATION (CASE-1) 12

14 POLICY (CASE-2) Prohibit Cars in ten (10) links (Case-1 + Hanazonomachi Avenue) 13

15 POLICY SIMULATION (CASE-2) 14

16 POLICY (CASE-3) Prohibit Cars in sixteen (16) links (Case-1,2 + Making a Small Traffic Cell) 15

17 POLICY SIMULATION (CASE-3) 16

18 POLICY (CASE-4) Prohibit Cars in sixty (60) links (Making a Large Traffic Cell) 17

19 POLICY SIMULATION (CASE-4) 18

20 VISUALIZATION OF FLOW CHANGE CASE-0 CASE-4 CASE-1 CASE-3 CASE-2 19

21 THANK YOU