Title
Throughput-Improving Control of Highways Facing Stochastic Perturbations.
Abstract
In this article, we study the problem of controlling a highway segment facing stochastic perturbations, such as recurrent incidents and moving bottlenecks. To model traffic flow under perturbations, we use the cell-transmission model with Markovian capacities. The control inputs are: (i) the inflows that are sent to various on-ramps to the highway (for managing traffic demand), and (ii) the priority levels assigned to the on-ramp traffic relative to the mainline traffic (for allocating highway capacity). The objective is to maximize the throughput while ensuring that on-ramp queues remain bounded in the long-run. We develop a computational approach to solving this stability-constrained, throughput-maximization problem. Firstly, we use the classical drift condition in stability analysis of Markov processes to derive a sufficient condition for boundedness of on-ramp queues. Secondly, we show that our control design problem can be formulated as a mixed integer program with linear or bilinear constraints, depending on the complexity of Lyapunov function involved in the stability condition. Finally, for specific types of capacity perturbations, we derive intuitive criteria for managing demand and/or selecting priority levels. These criteria suggest that inflows and priority levels should be determined simultaneously such that traffic queues are placed at locations that discharge queues fast. We illustrate the performance benefits of these criteria through a computational study of a segment on Interstate 210 in California, USA.
Year
Venue
Field
2018
arXiv: Systems and Control
Mathematical optimization,Markov process,Traffic flow,Control theory,Queue,Flow (psychology),Linear programming,Throughput,Mathematics,Perturbation (astronomy),Bounded function
DocType
Volume
Citations 
Journal
abs/1809.07610
0
PageRank 
References 
Authors
0.34
1
3
Name
Order
Citations
PageRank
Li Jin12510.58
Alex A. Kurzhanskiy2719.23
Saurabh Amin3112287.45