Since knowledge of motorists' route choice behavior as well as origin-destination flows takes an important role in traffic management and control over an expressway network, this paper presents two models of estimating origin-destination patterns and route flows simultaneously, both formulated in the framework of the Kalman filtering methodology. The one model employs as the measurement equation for link traffic counts the causal relationship between input flows from on-ramps and flows measured at off-ramps and/or mainline links, ignoring the travel time it takes for a vehicle to get to a specific link from its origin. The other model takes such travel time into account in establishing the measurement equations with the help of informations collected from a limited number of equipped vehicles incorporated in the automatic vehicle navigation system to be planned in the near future. For on-line application of the models a recursive algorithm is utilized which does not require any matrix inversion operation and is numerically accurate and stable. Finally, these models are tested by making use of synthetic data obtaind from digital traffic simulation executed for a small network. The results demonstrated that the presented models are capable of producing satisfactory estimates.
[1]
G. Davis,et al.
Recursive estimation of origin-destination matrices from input/output counts
,
1987
.
[2]
H Keller,et al.
A SYSTEMS DYNAMICS APPROACH TO THE ESTIMATION OF ENTRY AND EXIT O-D FLOWS
,
1984
.
[3]
V F Hurdle,et al.
DYNAMIC IDENTIFICATION OF FLOWS FROM TRAFFIC COUNTS AT COMPLEX INTERSECTIONS
,
1983
.
[4]
M. Cremer,et al.
A new class of dynamic methods for the identification of origin-destination flows
,
1987
.
[5]
Johannes Ledolter,et al.
A Recursive Kalman Filter Forecasting Approach
,
1983
.
[6]
L G Willumsen,et al.
ESTIMATING TIME-DEPENDENT TRIP MATRICES FROM TRAFFIC COUNTS
,
1984
.
[7]
M. Cremer.
Determining the Time-Dependent Trip Distribution in a Complex Intersection for Traffic Responsive Control
,
1983
.