서지주요정보
Bayesian mixture model for estimating travel time distribution based on low frequency probe vehicle data = 표본 프로브 차량 기반 베이즈 혼합 통행시간 분포 추정
서명 / 저자 Bayesian mixture model for estimating travel time distribution based on low frequency probe vehicle data = 표본 프로브 차량 기반 베이즈 혼합 통행시간 분포 추정 / Hyungjoo Kim.
발행사항 [대전 : 한국과학기술원, 2019].
Online Access 원문보기 원문인쇄

소장정보

등록번호

8033298

소장위치/청구기호

학술문화관(문화관)B1층 보존서고

DGT 19001

휴대폰 전송

도서상태

이용가능(대출불가)

사유안내

반납예정일

리뷰정보

초록정보

This dissertation aims to estimate travel time based on low and unknown sampling rates of probe vehicles and has two main objectives: i) enhance understanding of travel time properties and ii) develop a novel estimation model of travel time distribution. First, we explored the characteristics of travel time according to traffic state changes and congested traffic. The systematic relation between spot and section measurements is theoretically and empirically examined; as well, the characteristics of travel time variability are empirically observed. Then, we developed a novel estimation model of travel time distribution based on the Bayesian mixture model. To reduce the estimation error due to low and unknown sampling rate of probe vehicles, we used prior information in the Bayesian approach based on the characteristics of travel time regularity. To evaluate the accuracy of the methodology, we used empirical data from expressways and arterial roads. The Bayesian mixture model outperforms different estimation approaches.

본 연구는 표본 프로브 차량 기반 통행시간 분포 추정에 관한 연구로 i) 교통류 상태에 따른 통행시간 특성에 대한 이해, ii) 표본 데이터 기반 통행시간 분포 추정 방법론 개발 등으로 나누어 연구를 수행한다. 먼저 교통류 상태 변화 및 혼잡상태에서 통행시간 추정 시 발생하는 특성을 분석한다. 지점 및 구간검지 기반의 교통정보가 체계적인 관계를 가지고 있음을 이론적 및 실증적으로 분석하고, 혼잡 시 발생하는 통행시간 변동성의 원인을 실증적으로 분석한다. 이이서 표본 프로브 차량 기반 베이즈 혼합 분포에 기반한 통행시간 분포를 추정한다. 표본 데이터 기반 통행시간 추정 시 발생하는 오차를 베이즈 추론의 사전정보를 활용해서 오차를 최소화 시키며, 베이즈 혼합 모형를 활용해서 혼잡 시 통행시간 분포를 추정한다. 본 연구방법론의 정확도 검증을 위해서 연속류 및 단속류 구간에서 분석을 실시하며, 기존방법론의 결과와 비교함으로써 본 연구방법론의 유용성을 검증한다.

서지기타정보

서지기타정보
청구기호 {DGT 19001
형태사항 vi, 121 p. : 삽화 ; 30 cm
언어 영어
일반주기 저자명의 한글표기 : 김형주
지도교수의 영문표기 : Kitae Jang
지도교수의 한글표기 : 장기태
Including appendix
학위논문 학위논문(박사) - 한국과학기술원 : 조천식녹색교통대학원,
서지주기 References : p. 99-107
QR CODE

책소개

전체보기

목차

전체보기

이 주제의 인기대출도서

Dissertation layout

Domain of travel time estimation

Spatially stationary detection

Vehicle re-identification detection

Illustration of congestion pattern, and estimated and actual travel speeds

Categorization of different traffic states

Input values for numerical analysis

Changes in the relation between estimated and actual travel speeds: (a) origina relation; (b) free-flow speed, Vf, increases from 100 to 110 km/h; (c) congestion speed, Vc increases from 30 to 40 km/h; (d) queue growth wave speed, W1, diminishes from 20to 1 km/h; (e) queue dissipation wave speed, W2, diminishes from 15 to 5 km/h; (f) Duration 0 C.

Study site I: Gyeongbu Expressway

Spatiotemporal speed contours of(a) April 중요결정보 and (b) April 20"4201 (Seoul-bound Gyeongbu Expressway, South Korea)

Estimated intervals of traffic parameters on Korean Expressways

Theoretical and observed relations between estimated and actual travel speeds (a) April 12th,2011; and (b) April 20th,2011 (Seoul-bound Gyeongbu Expressway, Soutl Korea)

Study site II: Northbound Interstate 880, Hayward, Califomia

Spatiotemporal speed contourforFebruary 80,2008 (Northbound Interstate 880 Hayward, Califomia)

Estimated intervals of traffic parameters on Califomia Freeway

Theoretical and observed relation between estimated and actual travel speeds for February 8th, 2008 (Northbound Interstate 880, Hayward, California)

Descriptions ofindividual travel data and site

Vehicle trajectories in lane 1 and impactof'raffic oscillation: (a) vehicle trajectories in lane 1; (b) individual travel time in lane 1

Time-series of travel times across lanes

Cross correlation analysis across lanes: (a) lane 1 and2;(b)lane 2 and3;(c)lane 3 and 4; (d) lane4 and 5

Scatter plotanalysis across lanes: (a) lane 1 and 2; (b) lane2 and 3;(c) lane 3 and 4;(d)lane 4 and 5

Propagation of traffic oscillations in congested traffic

Summary statistics foreach time period

Travel time distribution under congestion: (a)17:50-7:55:(b)7:55-8:00;(c)8:00- 8:05; (d) 8:05-8:10; (e) 8:10-8:15; (f) 8:15-8:20; (g) 8:20-8:25;(h) 8:25-8:30;(i) 8:30-8:35

Comparison oftravel time estimation models undercongestion: (a) 7:50-7:55;(b) 7:55-8:00;(c) 8:00-8:05;(d) 8:05-8:10;(e) 8:10-8:15;(f) 8:15-8:20;(g) 8:20-8:25;(h)8:25- 8:30;(i) 8:30-8:35

Bayesian approach to travel time estimation

Framework of Bayesian mixture model fortravel time estimation

Similar trends with close travel time sequences

Nonlinear prediction based on k-NN algorithm

Parameter estimation in Bayesian mixture model

Performance evaluation

Case study: Gyeongbu Expressway

Study site: Gyeongbu Expressway, Anseong IC-Dongtan JC

Comparison oftraffic volume between VDS and DSRC, April 27th, 2017

Ratio between VDS and DSRC

Estimation error by penetration rate of probes

Scatter plotbetween ground truth and comparison groups: (a) 3% penetration rate ofprobes, total time;(b) 50% penetration rate ofprobes, total time

Speed contour based on loop detectors

Estimation errorin transition period (6:30-7:30, 12:00-13:00)

Estimation errorin congestion period (7:30-12:00)

Boxplot ofMAPE: (a) in transition period; (b) in congestion period

95% credible interval for different time intervals

Estimation error by penetration rate of probes

Scatter plotbetween groundtruth and comparison groups: (a) 3% penetration rate ofprobes, total time;(b) 50% penetration rate ofprobes, total time

Speed contour based on loop detectors

Estimation errorin transition period (6:30-7:30, 12:00-13:00)

Estimation errorin congestion period (7:30-12:00)

Boxplot ofMAPE: (a) in transition period; (b) in congestion period

95% credible interval for different time intervals

Estimation errorbypenetration rate ofprobes

Scatter plotbetween ground truth and comparison groups: (a) 3% penetration rate ofprobes, total time;(b) 50% penetration rate ofprobes, total time

Speed contour based on loop detectors

Estimation errorin transition period (6:30-7:30, 12:00-13:00)

Estimation errorin congestion period (7:30-12:00)

Boxplot ofMAPE: (a) in transition period; (b) in congestion period

95% credible interval for different time intervals

Estimation errorbypenetration rate ofprobes

Scatter plot between ground truth and comparison groups: (a) 3%penetration rate ofprobes, total time;(b) 50% penetration rate ofprobes, total time

Speed contour based on loop detectors

Estimation errorin transition period (6:30-7:30, 12:00-13:00)

Estimation errorin congestion period (7:30-12:00)

Boxplot ofMAPE: (a) in transition period; (b) in congestion period

95% credible interval for different time intervals

Study site: Dalgubeol arterial road

Comparison oftraffic volume between video camera and DSRC

Traffic oscillation-like congestion evolution patterns on urban road

Summary statistics for each timeperiod (Nov 6th,2017)

Travel time distribution on urban roads:(a) 18:30-18:35;(b) 18:35-18:40;(c) 18:40-18:45;(d) 18:45-18:50;(e) 18:50-18:55;(f) 18:55-19:00;(g) 19:00-19:05; (h) 19:05- 19:10;(i) 19:10-19:15; (j) 19:15-19:20; (k) 19:20-19:25;(1) 19:25-19:30

Estimation error by penetration rate ofprobes

Scatter plotbetween ground truth and comparison groups: (a) 3% penetration rate ofprobes, total time;(b) 50% penetration rate of probes, total time

Estimation errorin transition period (17:00-17:30, 19:30-20:00)

Estimation errorin congestionperiod (8:00-9:00, 17:30-19:30)

Boxplot ofMAPE: (a) in transition period; (b) in congestion period

95% credible interval for different time intervals

Estimation errorbypenetration rate ofprobes

Scatter plotbetween ground truth and comparison groups: (a) 3% penetration rate ofprobes, total time;(b) 50% penetration rate of probes, total time

Estimation errorin transition period (16:30-17:00, 19:00-19:30)

Estimation errorin congestion period (17:00-19:00)

Boxplot ofMAPE: (a) in transition period; (b) in congestion period

95% credible interval for different time intervals

Estimation error bypenetration rate ofprobes

Scatter plotbetween groundtruth and comparison groups: (a) 3% penetration rate ofprobes, total time;(b) 50% penetration rate ofprobes, total time

Estimation errorin transition period (16:30-17:00, 19:00-19:30)

Estimation errorin congestion period (17:00-19:00)

Boxplot ofMAPE: (a) in transition period; (b) in congestion period

95% credible interval for different time intervals

List ofvariables used for derivations

Illustration ofVE(t) and VA(t) in traffic state I(t<to orts<t)

Illustration of VE(t) and VA(t) in traffic state II(to <1<tI)

Illustration ofVE(t) and VA(t) in traffic state III (t1 <t<tz)

Illustration ofVE(t) and VA(t) in traffic state IV (t2 <t<t3)

Illustration ofVE(t) and VA(t) in traffic state V (t3 <t<t4)

Illustration of VE(t) and VA(t) in traffic state VI(t4 <t<ts)

Low frequency probe vehicles in Daegu, Dawn (0:00-7:00)

Low frequency probes in Daegu, Morningpeak(7:00-9:00)

Low frequency probes in Daegu, Morning (9:00-12:00)

Low frequency probes in Daegu, Aftemoon (12:00-18:00)

Low frequency probesin Daegu, Aftemoonpeak(18:00-20:00)

Low frequency probes in Daegu, Night(20:00-24:00)