Increased Convergence Rates in Multiagent Transport Simulations with Pseudosimulation

Abstract

A multimodeling approach to large-scale, activity-based, multiagent simulation of travel demand is introduced. MATSIM is a full activity-based transport simulation. Its greatest current performance limitation is the network loading simulation, currently a queue simulation (QSim). QSim is iteratively executed for the entire agent population for evaluating the effects of random mutations on the activity plans of a fraction of the population. After each QSim, poorly performing plans are discarded, good plans are kept, and the agents slowly learn what works best for their individual activity needs. In the application presented, the system periodically replaces QSim for a number of iterations with a simplified pseudosimulation that runs approximately two orders of magnitude faster. The pseudosimulation uses travel time information from the preceding QSim iteration to estimate how well an agent day plan might perform. Repeated iterations of the pseudosimulation produce better-performing plans in a short time. These plans are passed to the QSim for updating of network travel time information, and the process repeats. The technique is tested in a scenario for Zurich, Switzerland, and incorporates mode choice, road pricing, secondary activity location choice, activity timing adjustment, and dynamic routing. The technique dramatically improves convergence rates for such complex, large-scale simulations and fully exploits modern multicore computer architectures.

The Multi-Agent Transport Simulation MATSim

“…First, the signi cantly larger number of network links and nodes increases time and resources for routing and dynamic queue simulation and could erase the advantages of a small-scale network. Extended simulation times can be tackled with the new pseudo-simulation methodology, currently developed by Fourie et al (2013). Second, total network capacity increase leads to reduction or even disappearance of congestion during peak hours, although including freight and through tra c in the scenario can make it more realistic and address congested conditions during peak-hours.…”

“…Relaxation Process The number of iterations needed by MATSim's co-evolutionary algorithm to reach a stable state was a critical variable; e orts were made to reduce it (Meister et al, 2006;Fourie et al, 2013).…”

“…An alternative path closer to the idea presented in was successfully attempted, using a JDEQSim based model (see Section 4.3.2) with within-day support and travel time approximation, as seen in PSim (see Chapter 39, Fourie et al (2013)). This removed overhead, Source: present in the previous approach, enabling exibility to implement many of the parking strategies presented in Axhausen and Polak (1989) and beyond.…”

“…For a detailed discussion of the meta-modeling approach and the results of applying this method to the Zürich scenario, refer to Fourie et al (2013).…”

An improvement in MATSim computing time for large-scale travel behaviour microsimulation

“…In response to these two issues, the improvement work is focused on the reduction of the number of the iterations and the decrease of the computing time in network loading, which are achieved by proposing a new framework for the MATSim replanning module, and new execution modules, respectively. The proposed improvement methods are tested in a Chinese medium-sized using MATSim calibrated with Cadyts (Calibration of dynamic traffic simulations) (Flötteröd 2009;Flötteröd et al 2012;Fourie et al 2013).…”