A multi-stage looking-ahead holding strategy to stabilize a high-frequency bus line

TL;DR

This paper introduces an adaptive multi-stage look-ahead holding strategy for stabilizing high-frequency bus lines, utilizing global headway information and simulation to optimize bus operation and passenger waiting times.

Contribution

It proposes a novel multi-stage look-ahead mechanism combined with dynamic target headway to improve bus line stability over existing simpler strategies.

Findings

Optimal number of look-ahead stages is crucial.

Larger action set improves strategy performance.

Proper interval between holding times enhances stability.

Abstract

If a bus line becomes unstable, passengers waiting time will be lengthened and buses capacities will be mismatched. To stabilize a high-frequency bus line, many holding strategies have been proposed. Among these strategies, some need to take oversimplified assumptions to simplify the formulation of a real bus line; some may choose a myopic holding time because they take into account only the local and currently available headway information. To overcome the above shortcomings, we proposed an adaptive holding strategy which continuously reduces the deviations of the instantaneous headways from the Dynamic Target Headway -DTH. By using DTH, the new strategy can make use of the global headway information to determine a proper holding time. To fully estimate the influence of a holding time on the operation of a bus line in a relative long time period, we introduced the multi-stage looking-ahead mechanism into our new strategy. A detailed bus line simulation system was used to realize the above mechanism and to describe all kinds of complex components of a bus line. The numerical experiment demonstrated the effectiveness of our new strategy. Some meaningful insights were uncovered as follow: a. The number of stages to be looked ahead should not be too small or too big; b. The bigger the range of the action set, the better the performance of our strategy; c. To obtain the optimal performance, the interval between two neighbor holding times should not be too small or too big; d. the bigger the number of the control points, the better the performance of our strategy.