visualize: fix timestamps on diagramms

[celle1234]

I tried to fix the problem with the timestamps in the diagrams. The discussion you can find in #387.

[NormannK]

visualization_results.pdf
from single_test_optimization.py for discussion

[Lasall]

Thank you for your PR! Currently the optimization endpoint (and therefore the visualization) expect the forecast data starting at 00:00 of the current day + x prediction hours (currently 48). The optimization starts at start_hour (default rounded down current hour, e.g. 16:42 -> 16) and goes until optimization hours (currently 48) which starts at 00:00 of the current day just like the predictions (e.g. start_hour = 16, optimization_hours = 48 - > optimization will go until end of the next day).

[celle1234]

@Lasall Yes I understand this in the same way.

For the PR it is important, that the timezone in EOS.config.json is set. Otherwise the axis timeline does not match.

[Lasall]

new_optimize_result_2_full.pdf
from test artifact.

Maybe you could also adjust the relative times "hours since start" to be the same between predictions and optimization?Currently it could be confusing the prediction goes 0-48 and optimization 0-38 (even though optimization_hours is set to 48).

[celle1234]

Did you mean something like this?
visualization_results.pdf

[NormannK]

Did you mean something like this? visualization_results.pdf

after some code digging I think this is "more" correct.

EOS/src/akkudoktoreos/core/ems.py

Line 131 in 1a2da76

cls._start_datetime = start_datetime.set(minute=0, second=0, microsecond=0)

This line does set the start time to the start of the current hour (if not defined). But imo it should be the start of the next hour, but that's something for an other discussion.

[celle1234]

I added the changes.

[Lasall]

Looks good to me, probably need some fine tuning with the optimization diagram (offset +1) but in a different PR (I think that's also what you meant Normann).

PDFs from test can be checked out here: https://github.com/Akkudoktor-EOS/EOS/actions/runs/13120831990?pr=430