Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Large-Scale Model Calibration
Hello,
- many vehicle type parameters that you would use to calibrate a microscopic simulation have little or no effect in meso: https://sumo.dlr.de/docs/Simulation/Meso.html
- if you call duarouter just once then you will likely get lots of congestion due to lack of proper traffic assignment: https://sumo.dlr.de/docs/Demand/Dynamic_User_Assignment.html
- many forms of congestion and deadlock arise due to network model issues. It doesn't make much sense to calibrate traffic parameters before solving the network issues first

regards,
Jakob
re

Am Mo., 24. Mai 2021 um 21:20 Uhr schrieb mehdi <kpk628@xxxxxxxxxxxx>:
Dear All,

I am working on a large-scale simulation. I have all the OD matrics; using
OD2TRIPS, I generated the trips file, and using Duarouter, I generated the
rou.xml file. I also arranged about 200 E1 detectors in my networks. At the
same point, I have the real-world annually average speed data for one day.
Now I want to calibrate the model with real-world data. I want to change
some parameters such as acceleration, deceleration, Tau, Sigma, max speed,
gap distance. This speed data is the only data that I have from the real
world.
To change these parameters, is it accurate to change with
"vehtypedistributionpy"? How vehtype distribution works? Do I need to adjust
parameters one by one? Because I think vehtype distribution shift the
parameters of the car randomly, I believe.
Each time I need to change the parameters and run theSUMO, then calculate
the RMSE between model speed data and observed data. And adjust until reach
minimum square error is reached. Is it a proper way for calibration? Are
there other ways or scripts to calibrate a large-scale network?
I also using --meso to run faster. But there is congestion and deadlock
traffic and cars just stock on streets. Changing these parameters does not
have any effect on my traffic flow.
I appreciate it if you could help me.

Best wishes
Mehdi



--
Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

Back to the top