Create datamodel for PV data

[peterdudfield]

Detailed Description

Create datamodel for pv data

metadata

• pv system id
• name?
• capacity?
• orientation?
• lattitude,
• longtitude
• provider: str

pv_yield:

• id
• datetime
• solargeneration{units}
• pv system id

[JackKelly]

at the moment, none of our code makes use of capacity or orientation. But it might still be useful to store this data.

One quick question: How to distinguish between PV systems from PVOutput.org vs PassivSystems?

[peterdudfield]

at the moment, none of our code makes use of capacity or orientation. But it might still be useful to store this data.

One quick question: How to distinguish between PV systems from PVOutput.org vs PassivSystems?

Yea, Capacity might be a good way to normalize things, rather than max value pv value of this historic data.

Ill add a column for provider, then we know where is from

[JackKelly]

Capacity might be a good way to normalize things, rather than max value pv value of this historic data

In my limited experience, the nominal capacity is often wrong. So, if we divide by the nominal capacity, we're very likely to get values above 1. I'd recommend that we continue normalising by the max of the timeseries :slightly_smiling_face:

[peterdudfield]

Ah ok, maybe I can check this using the data set.

1. use max pv values to normalise data. Save Max value from historic data, so that in 'live' these values can be loaded
2. use pv capaicty value to normalise both historic and live data.

Just to show there is an extra piece to the puzzle then, but perhaps the extra price is work it, if the capacity values are often wrong

[JackKelly]

yeah, my understanding is that the nominal capacity values are so often wrong that we should mostly ignore them!