1.7 KiB
1.7 KiB
Optimizer
Optimizer uses mixed integer linear programming (MILP) to minize a cost function. The optimizer model implementation honors:
- energy consumption/ feed-in costs
- solar forecast
- base load (aka home) energy demand
- end of forecast commercial value
- strategy- either "charge before export" (charge loads as soon as possible) or attenuating grid peaks on the demand side, the feed-in side or both
- home battery or loadpoint/vehicle...
- capacity, soc and charge goals
- charge/discharge power limits and efficiency
Optimization spans N slots with N being minimum of available forecast data.
Parameters
Energy consumption/ feed-in costs/ Solar forecast
Grid/ feed-in/ solar tariff data.
TODO
- make feed-in optional
- make solar optional
Base load energy demand
Collected 15min energy profile averaged over the last 30 days.
Measured value blending
The solar forecast and the base load profile are anchored to the current situation using the last completed 15min metrics slot, decaying linearly over 4 slots:
- base load: the measured home consumption replaces the first slot and decays into the profile
- solar: the scale factor measured production/forecasted production is applied to the first slot and decays towards 1
End of forecast commercial value
Use minimum of energy consumption cost.
Home Battery
Loadpoint and Vehicles
- home battery or loadpoint/vehicle...
- capacity, soc and charge goals
- charge/discharge power limits and efficiency
Without vehicle capacity or soc a configured session energy limit is modelled instead: state is the session's charged energy, goal is the limit. Loadpoints with neither are not modelled at all- their power is added to the base load.