Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] which parameter has the most effect on mean headway?
Thank you for your help.

Actually, I'm looking for proper parameters that have the most effect on my results, in order to use them in the calibration process. my problem is not a single purpose, so when I try to fit headway with real data, I also should take care of other indicators such as average speed and etc. this is why I change SUMO parameters with a reasonable step to see their effects on my indicators. this is a sample of my simulation scenario:
https://filebin.net/ur65tc7o7yxvgftg/SUMO.zip?t=t90azruf

1) how should I configure SUMO, in order to be able to use sub-second departure time? setting step-length to 0.1s could solve the problem?
2) if I use tau smaller than action-step-length, is tau parameters rounded to the nearest multiple of action-step-length or not?


With Best Regards,
Mohsen

On Mon, Sep 9, 2019 at 5:10 PM Jakob Erdmann <namdre.sumo@xxxxxxxxx> wrote:
I'm not quite sure how you compute your "absolute magnitude of mean difference". Possibly it is invariant to a constant headway change by all vehicles. In this case you should use a tau distribution instead of setting the same tau for all vehicles. See https://sumo.dlr.de/wiki/Tools/Misc#createVehTypeDistributions.py

1) the departTime is rounded down (e.g. depart 1.7 actually departs at time 1)
2) the speed adapts to the safe one. If there is no safe speed, insertion is delayed.

regards,
Jakob

Am Mo., 9. Sept. 2019 um 14:01 Uhr schrieb Mohsen Rahmati <mohsen.eng74@xxxxxxxxx>:
Thanks, Jakob.
the plots were I mentioned before:
https://filebin.net/bhvv8wow4cx0e9x8/Headway.zip?t=h3x55hiu
as you can see in the plots, tau was not effective to the mean headway. (all other parameters left at their SUMO default values)


I have real data aggregated with a time resolution of 5 minutes, so the exact time pass of vehicles in every 5 minutes is unknown to me.
in order to generate vehicles in the simulation scenario, I use exponential distribution for the departure time of each vehicle. (I've tried some distributions and exponential seems to work relatively better than others)

now my questions are:

1) by leaving step-length, action-step-length and tau at their default values, what would happen if I use a subsecond time for the departure value of vehicles (for e.g, the first vehicle depart at 1.2s and the second one depart at 1.7s)? dose SUMO take care of that?

2) I use departLane="best" and departSpeed="max", what would happen if the departure time isn't safe? the departure time is delayed or the speed adapts to the safe one?

3) also, what is a safe situation in vehicles generation? the distance greater than the minGap and the headway greater than the tau or you mean something else by stating safe situation?

With Best Regards,
Mohsen


On Mon, Sep 9, 2019 at 1:47 PM Jakob Erdmann <namdre.sumo@xxxxxxxxx> wrote:
Your plot attachments are missing (possibly blocked by the mailing list).
Mean headway is mostly influenced by the 'tau' parameter.
However, it also depends on the desired speeds (speedFactor, speedDev) and traffic density:
- if all vehicles have the exact same desired speed (speedDev= 0) then they will never close the gap to their leader and the headways are fully determined by the insertion procedure.
- if the desired speeds vary and the density is high, then most vehicles will be in a queue following a slower leader with headways determined by tau
- if the desired speeds vary and the density is low, then the headways are determined by insertion as well as the width of the speed distribution (and of course tau)

Am Sa., 7. Sept. 2019 um 20:12 Uhr schrieb Mohsen Rahmati <mohsen.eng74@xxxxxxxxx>:
Hello,
I'm working on calibrating a highway section based on real data gathered from detectors.
I've changed 12 parameters* in car-following and lane-changing model to decrease the "Absolute Magnitude of Mead Headway Difference" but it seems none of them works. Plots were attached.
I wonder if you could introduce me the parameter that has the most effect on mean headway?

by the way:
SUMO version: 1.3.1
car-following model: KraussPS
lane-changing model: LC2013

Note*: the effect of simultaneous change of parameters was not considered. so by changing a single parameter, all other parameters were left at their default values.


With Best Regards,
Mohsen
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/sumo-user
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/sumo-user
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/sumo-user
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/sumo-user

Back to the top