EOS Connect is an open-source tool for intelligent energy management and optimization. It uses the EOS (Energy Optimization System) and connects to various smart home platforms (like Home Assistant and OpenHAB) to monitor, forecast, and control your energy flows. EOS Connect fetches real-time and forecast data (PV, load, prices), processes it via EOS, and automatically controls devices (directly FRONIUS or inverters and batteries that are supported by evcc) to optimize energy usage and costs. It features a dynamic web dashboard for live monitoring and manual control, and supports integration with MQTT, EVCC, and more.
Key features in short:
- Automated energy optimization using real-time and forecast data
- Battery and inverter management (charge/discharge, grid/PV modes)
- Integration with Home Assistant, OpenHAB, EVCC, and MQTT
- Dynamic web dashboard for monitoring and control
- Cost optimization using dynamic energy prices (e.g., Tibber, smartenergy.at)
- Easy configuration and extensibility
EOS Connect helps you maximize self-consumption, minimize grid costs, and get the most out of your solar and storage systems.
- EOS Connect
- Dynamic Energy Flow Control:
- Automatically optimizes energy usage based on system states and external data.
- Supports manual override modes for precise control.
- Battery Management:
- Monitors battery state of charge (SOC) and remaining energy.
- Configures charging and discharging modes, including:
- Charge from grid.
- Avoid discharge.
- Discharge allowed.
- EVCC-specific modes (e.g., fast charge, PV mode).
- Dynamic Charging Curve:
- Dynamically adjusts the maximum AC charging power based on system conditions.
- Ensures efficient and safe battery charging by adapting to real-time energy availability and battery state.
- Cost and Solar Optimization:
- Aligns energy usage with real-time electricity prices (e.g., from Tibber or smartenergy.at) to minimize costs.
- Incorporates PV forecasts to prioritize charging during periods of high solar output.
- Reduces grid dependency and maximizes self-consumption by combining cost and solar production data.
- Energy Optimization Scheduling:
- Displays timestamps for the last and next optimization runs.
- Tracks system performance and optimization results.
- Real-Time Monitoring:
- View current system states, including battery SOC, grid charge power, and EVCC modes.
- Dynamic icons and color-coded indicators for easy visualization.
- User Controls:
- Set grid charge power and override system modes directly from the interface.
- Configure EVCC charging behavior with intuitive controls.
- Home Assistant:
- Full MQTT integration with Home Assistant Auto Discovery.
- Automatically detects and configures energy system entities.
- OpenHAB:
- Integrates with OpenHAB for monitoring and controlling energy systems.
- Publishes system states and subscribes to commands via MQTT.
- EVCC (Electric Vehicle Charging Controller):
- Monitors and controls EVCC charging modes and states.
- Supports fast charge, PV charging, and combined modes.
- Inverter Interfaces:
- OPTION 1: Communicates directly with a Fronius GEN24 to monitor and control energy flows.
- OPTION 2: Use the evcc external battery control to interact with all inverter/ battery systems that are supported by evcc (hint: the dynamic max charge power is currently not supported by evcc external battery control)
- OPTION 3: using without a direct control interface to get the resulting commands by EOS connect MQTT or web API to control within your own environment (e.g. Integrate inverter e.g. sungrow SH10RT #35 )
- Retrieves real-time data such as grid charge power, discharge power, and battery SOC.
- MQTT Broker:
- Acts as the central hub for real-time data exchange.
- Publishes system states and subscribes to control commands.
This project is in its early stages and is actively being developed and enhanced.
2025-04-10
- EOS made a breaking change - see here https://github.com/Akkudoktor-EOS/EOS/discussions/513
- there were also changes in the API at '<your_ip>:8503' - unfortunately the API is not versioned (ping ;-) )
- to fullfil both versions there is small hack to identify the connected EOS
- finally the current version can run with both EOS versions
EOS Connect is a self-running system that periodically collects:
- Local energy consumption data.
- PV solar forecasts for the next 48 hours.
- Upcoming energy prices.
Using this data, a request is sent to EOS, which creates a model predicting the energy needs based on different energy sources and loads (grid, battery, PV).
EOS Connect waits for the response from EOS (e.g., ~2 min 15 sec for a full 48-hour prediction on a Raspberry Pi 5). After receiving the response, it is analyzed to extract the necessary values.
Finally, the system sets up the inverter based on the following states:
MODE_CHARGE_FROM_GRIDwith a specific target charging power (based on your configuration).MODE_AVOID_DISCHARGE.MODE_DISCHARGE_ALLOWEDwith a specific target maximum discharging power (based on your configuration).
The system repeats this process periodically, e.g., every 3 minutes, as defined in the configuration.
Data collection for load forecasting is based on your existing load data provided by an OpenHAB or Home Assistant instance (using the persistence of each system). EOS requires a load forecast for today and tomorrow.
Load data is retrieved from:
- Today one week ago, averaged with today two weeks ago.
- Tomorrow one week ago, averaged with tomorrow two weeks ago.
- Car Load Adjustment: If an electric vehicle (EV) is/ was connected, its load is subtracted from the household load to ensure accurate forecasting of non-EV energy consumption.
(See Home Assistant Persistance for more details.)
Load data is retrieved from the last two days:
- From two days ago (00:00) to yesterday midnight.
- Car Load Adjustment: Similar to Home Assistant, the EV load is subtracted from the household load to isolate non-EV energy consumption.
PV forecasts are retrieved directly from the AKKUDOKTOR API.
Energy price forecasts are retrieved from the chosen source (TIBBER or AKKUDOKTOR API). Note: Prices for tomorrow are available earliest at 1 PM. Until then, today's prices are used to feed the model.
The dashbaord of EOS connect is available at http://localhost:8081.
With the first start of EOS connect a default config.yaml will be generated in the \src folder. For full documentation for the different entries go to CONFIG_README
Note: With the default config and a valid EOS server IP/DNS name entry ('eos -> server') - EOS connect should be running out of the box with some static defaults as a start point for a step-by-step commissioning.
The tool will use historical data from Home Assistant's local database. By default, this database is configured with a retention period of 10 days.
To improve the accuracy of load forecasts, it is recommended to use data from the last 2 weeks.
You can extend the retention period by modifying the recorder configuration in Home Assistant's configuration.yaml file. If the recorder section is not already present, you can add it as shown below:
recorder:
purge_keep_days: 15 # Keep data for 15 daysAfter making this change, restart Home Assistant for the new retention period to take effect.
Note: Increasing the retention period will require more storage space, depending on the number of entities being recorded.
If you do not change the retention period, the tool will still work, but it will use the available 10 days of data, which may result in less accurate load forecasts.
No specific info yet.
The application will start fetching energy data from OpenHAB or HomeAssistant and processing it. You can access the web interface at http://localhost:8081. For local usage the port is configurable see CONFIG_README. For docker usage change the mapped port in docker-compose.yml.
To run this project, you need to have the following installed:
- Python 3.x
You need to have the following additionally installed:
- Docker
- Docker Compose
Pull existing latest snapshot and run the application in the background using Docker Compose (--pull always ensures the latest image is pulled):
docker-compose up --pull always -d
see https://github.com/ohAnd/ha_addons
or directly
-
Clone the repository:
git clone <repository-url> cd EOS_connect -
Install the required Python packages (for local usage):
pip install -r requirements.txt
Run the application locally without Docker, you can execute the following command:
python src/eos_connect.py
Contributions are welcome! Please submit a pull request or open an issue for any suggestions or improvements.
This project is licensed under the MIT License. See the LICENSE file for more details.
