Next: Software
Architecture Up: Evaluation Framework Previous: Simulation Output and MOEs
A prototypical predictive control and route guidance generation system
is implemented in SIMLAB to demonstrate the use of the methodology and
simulation tools developed in this research. In this particular
application, MesoTS, the mesoscopic traffic simulator presented in Chapter 5, is used for
traffic prediction.
The addition of the real-time traffic prediction model significantly
complicates the synchronization of modules in the simulation
laboratory. The coordination among the simulators for a typical
traffic prediction cycle and testing of a candidate strategy is
depicted in Figure 2-4 and
summarized below:
Figure 2-4: Iterative guidance generation and
traffic prediction
- (1)
-
When TMS needs a new prediction of future traffic conditions
(e.g. next 45 minutes), it sends a message to MITSIM to request the
current network state as captured by the surveillance system. To
simplify the discussion, in the following sections, we assume that the
surveillance system provides complete information on the network
state. Hence, a model for estimating current network state is not
used.
- (2)
-
As MITSIM receives this request, it scans and writes the current
network state into a file. Network state is represented by the
positions, speeds, and destinations of vehicles currently in the
network and those waiting to enter the network. When MITSIM finishes
reporting the current network state, it sends a message to TMS and
continues the simulation.
- (3)
-
TMS also sends the current signal controls and route guidance to
MesoTS in order to start the simulation for the given prediction
horizon.
- (4)
-
When MesoTS receives the message from TMS, it utilizes the network
state from MITSIM in step (2) and predicts future traffic conditions
for a given prediction horizon (e.g., 45 minutes). When the
mesoscopic simulation for the given prediction horizon is complete,
MesoTS writes the predicted time dependent link flows and travel times
to a file. Then it sends a message to TMS to signal the completion of
the prediction task.
Having received this message, TMS may proceed in two ways: (i) if the
candidate solution is not satisfactory it restarts a mesoscopic
simulation with the same initial network state but updated route
guidance and signal controls; otherwise, (ii) it terminates the
iterations and accepts the route guidance and signal controls. In the
first case, steps (2)-(4) are repeated. In the second case, step (5)
is executed.
- (5)
-
TMS sends a message to MITSIM to announce that new guidance and signal
control strategies are available for implementation. After a
pre-specified computation delay - the difference between the
time the network state is reported and the expected completion time
for the generation of control strategies and route guidance - MITSIM
switches to the updated controls and route guidance and continues the
simulation.
Next: Software Architecture
Up: Evaluation Framework
Previous: Simulation Output and Measures
Qi Yang
Wed Feb 26 19:17:06 EST 1997