I’m using the RPis in a few roles in relation to the solar panels. I collect daily solar production data from the inverter. The data is energy produced per hour. I also collect from Met Office model weather forecast data for our home location. This data is also per hour and I’m using 10+ parameters the forecast model provides. I learn every night a GP Regression model from the solar and weather forecast data. Learning every night ensures I always use the lates data available in the model. The GPR model allows me to estimate solar production tomorrow given a Met Office model forecast. I then use this estimate in various ways in decision-making, for example how much to use cheap electricity overnight to charge house and car batteries, i.e. how much headroom to leave in the batteries so that I’m unlikely to waste excess solar energy.