Using predicted sun irradiance (rather then external HW sensors)

[haxtibal]

Hi @basbruss,

ideally one would have pyranometers or brightness sensors mounted on the roof top and use their measured values to decide for how much to close a cover. IIUC this can already be done with your integration by means of sensor.irradiance and sensor.lux.

I don't have such hardware and don't want to buy/install one. Luckily there's solid research on the topic and ready to use code (pvlib.irradiance, sun_daily, Global horizontal irradiance clear sky models : implementation and analysis) for how to predict sun irradiance based on location, current time and surface orientation. It models a good amount of reality, e.g. atmospheric state (air pressure, temperature, relative humidity, aerosol content, Rayleigh scattering), ground reflexion, distance of earth from sun.

So I drafted a custom integration (unpublished, too dirty yet) that calculates how much direct and diffuse irradance hit the tilted window and provide the calculated W/m² as HA sensor entities. Afterwards I found your very nice project and now I wonder if it would make sense to implement such irradiation prediction directly into your adaptive-cover project? Guess we could present it as optional alternative to sensor.irradiance and sensor.lux in the config flow ("Use predicted irradiance?" y/n). If so I'd stop my current approach and rather try to make a PR for your project.

Otoh, keeping the predicted values in a separate integration could support use cases unrelated to adaptive cover control (maybe photovoltaic forecast and motion control).

What do you think?

Tobias