Geoscience Reference
In-Depth Information
This chapter is structured as follows: in the next section, MATSim is described
together with several mobility simulation implementations available for MATSim
and open issues in this regard. Thereafter, the implementation of the various
performance improvements with regards to the micro-simulation and the event-
handling model are described. This is followed by experiments which assess the
performance gains due to the newly implemented models. Before concluding, open
issues are discussed together with possible future work.
9.2
Related Work
In the following section, a description of MATSim is given followed by a presen-
tation of the different micro-simulation models and the event-handling module in
MATSim.
9.2.1
MATSim
In MATSim, individuals are modeled as agents who want to perform activities
throughout the day, such as being at home or work and going shopping. But due to
the spatial separation of the corresponding activity locations, agents need to travel.
This leads to many additional choices for the agent, such as the mode of travel,
the activity duration, the location, and the route choice. The goal of MATSim is
to find a plan for each agent, which maximizes the overall utility of the agent,
including items as travel time, ticket fares, or street toll prices. This optimization
needs to be performed while keeping constraints of the agent's environment in mind,
such as street network capacities or opening times at shops and working hours.
This optimization of agents' plans in MATSim is achieved by applying an iterative
process, which is depicted in Fig.
9.1
.
In the beginning, the simulation starts with an initial plan for each agent depicted
as
initial demand
. A simple plan for an agent, who wants to leave home at 7:25 a.m.
in the morning, work for 8 h and 20 min, and then drive back home, might look as
follows:
simulation
(execution)
initial demand
scoring
analysis
replanning
Fig. 9.1
Coevolutionary simulation process of MATSim
Search WWH ::
Custom Search