Skip to content

Home

Jump to bottom
David edited this page Jan 10, 2023 · 27 revisions

Welcome to the emhass wiki!

This is the roadmap of this project. The Add-On for Home Assistant OS users: https://github.com/davidusb-geek/emhass-add-on

TODO

New functionalities

  • Create a plotting script to visualize the optimization results.
  • Propose multiple types of cost functions: profit maximization, self-consumption maximization, etc
  • Integrate the possibility of variable tariffs, for purshasing and selling energy to the grid.
  • Support for list of inverters, see: get_power_from_weather in forecast class.
  • Implement an energy management with a Model Predictive Control approach. Consider implementing the receiding horizon approach.
  • Add total energy constraint for each deferrable load.
  • Battery SOC should be an input. For now only SOCinit=SOCend simulations are implemented
  • Add a web server app to run the complete module. This will be the webserver used in the add-on. Separate usage in standalone mode in docker container or in HA add-on mode.
  • Add simple graphics explaining what is this?
  • Introduce the modeling of constraints during optimization for a thermal energy storage (see Peter Pflaum Thesis or Langer et al paper https://arxiv.org/pdf/2009.02349v2.pdf).
  • Support for EVs (see Peter Pflaum Thesis).
  • Support for virtual batteries.
  • Add elasticity to LP formulation in case on infeasible solution.
  • Support for functioning hour period for each deferrable load.
  • Expand use case, add battery, test EV as battery with P_discharge=0
  • Let the user choose the type of mount: TEMPERATURE_MODEL_PARAMETERS['sapm']

Related to forecasting improvement

  • Define the type of forecast that should be used from the configuration file.
  • Move get_load_unit_cost from optimization to forecast class: define forecast methods for load and PV production prices.
  • Add simple integration of current/now values for PV and load forecast. This is important for MPC applications with high optimization frequencies. The new forecast can be computed using a mixed one-observation presistnace model and the forecasted values from the current method. This can be the equation for this: $P^{mix}{PV} = \alpha \hat{P}{PV}(k) + \beta P_{PV}(k-1)$
  • Improve load forecasting using a time series forecast algorithm. Some tests were made with fbprophet but results are not completly satisfactory. The model needs some regressors for more accuracy.
  • Test with LTSM with or without Autoencoders.

Packaging, HA integration, testing

  • EMHASS hass been tested in Home Assistan Core. It need to be tested on Home Assistant Operating System and Home Assistant Container.
  • Create an EMHASS add-on for even easier installation on Home Assistant Operating System and Home Assistant Supervised.
  • Improve testing to be used with no running hass instance.
  • Package everything in a docker container.
  • Package as a docker container with a new Dockerfile simpler than in the add-on
Clone this wiki locally