automate calibration procedure for dinural curves (BAST)

[wese-da]

Okay this was on the table for a long time.

Calibration of the dinural curves is pretty straightforward once the automatic counting stations of the BASt are integrated in the database. This could then be part of the automatic model generation:

• check for counting stations in the survey area
• create a MATSim counting station for each of them
• write them to a file and reference it in the config

Maybe we can even generate those counting stations dynamically which would be

• great because we won't end up with a ton of files
• bad for documentation of the procedure

I didn't quite make up my mind about this matter. Opinions?

The first point (automatic modal split calibration) is trivial to implement since we only need an outer loop around the MATSim loop that calculates the modal split of the last simulation run and checks the deltas of the simulated vs the real modal split. More interesting is the question of the termination criterion which is classically sth. like:

• number of iterations exceed a limit
• delta is below a certain threshold (my favorite)

We just have to set the delta to a value that's small enough to produce plausible results without ending up with over, say 20 simulation runs...

These are my thoughts. There's nothing new or revolutionary, I just wanted to flush them all out :smiley:

[bsmoehring]

Just as a hint for the first point: as far as I know Nico Kuehnel tried to integrate all BASt counting stations attached to links in OSM during his internship at DB Analytics. Don't know if this data really exists or if you're already using it.

[wese-da]

Okay, thanks for the hint. Actually, I have no idea what the status of this project is, but the following discussion looks not so promising. Maybe let's talk with Nico about this.

[00Bock]

still open for BAST data.

[00Bock]

bast data now in db surveyed_mobility schema bast. compare: smb://192.168.0.3/gisdata/MOBILITYDATA/TRAFFIC_COUNT_DATA