Solar Load Controller

Automatically controls Anker Solarbank 2 E1600 AC home load (0–800 W) based on real-time IOMeter smart meter readings to minimize grid electricity usage.

Features

• Auto load control — Reads the grid meter in real-time and adjusts the Solarbank home load to keep grid import near zero
• Manual control — Slider + buttons to manually set load (0–800 W in 10 W steps)
• Real-time dashboard — Solar production, battery SOC, meter reading, power flow visualisation
• Load history — 24 h chart of load changes vs meter readings
• Daily energy summary — Solar production, grid import/export, cost saved
• All parameters tunable from UI — No restart needed to change strategy settings

Architecture

FastAPI (REST API + lifespan tasks)
  ├── IOMeter Service  →  polls local meter every 5 s
  ├── Anker Service    →  polls cloud API every 30 s, sets load
  ├── Strategy Engine  →  3-layer algorithm (spike detect → adjust → cooldown)
  └── Dash Dashboard   →  mounted at /dashboard, Bootstrap 5 dark theme
         └── SQLite    →  config, meter readings, load changes, profiles

Motivation

This project was originally created to solve a very common pain point. After installing solar panels and the Anker Solix system, I realized that, like many other users, I did not have a smart meter installed — either because it wasn’t possible or due to cost considerations.

Without a smart meter, it becomes difficult to monitor and control power output and household loads in real time. Even when using smart plugs, they cannot effectively manage all devices in a home. For example, regular lighting or high-power appliances (like heating devices) are usually not connected to smart plugs, so they cannot be intelligently controlled.

To address this issue, I discovered a device called iometer, which can be easily integrated into a home environment. It also allows local access to its interface, making it very convenient to read real-time electricity usage data. Because of this, I highly recommend iometer as a solution for real-time energy monitoring.

Based on these considerations, I integrated these components into a system that enables real-time monitoring and automated control through a dashboard. This dashboard can be accessed both on a computer and on a mobile device.

In addition, in another project, I developed an ESP32-based real-time display terminal, which connects to the same web API to show live data.

Overall, this integrated solution is convenient, accurate, and practical for daily use. Anyone who is interested is welcome to download and try it out.

Screenshots

Live Dashboard

Real-time view: solar production, battery SOC, grid reading, per-channel PV breakdown, power flow, and load control — all on one page.

Load Control & Strategy Selection

Manual load slider (0–800 W) alongside the auto-control panel with 5 selectable strategies.

Live Grid Chart

24-hour grid meter history. Spikes are appliance events; the strategy keeps the baseline near zero.

Strategy Configuration

All control parameters tunable live from the UI — no restart required.

Analytics — Load vs Grid Balance

Overlay of Solarbank load output and grid meter over time, with colour-coded import/export zones and self-sufficiency stats.

Analytics — Daily Energy Flow

Multi-day view of solar self-use, grid import/export and self-sufficiency percentage.

Analytics — Energy Flow (30 days)

30-day stacked bar chart: self-use, grid export, and grid import at a glance.

Analytics — Daily Detail Table

Per-day breakdown of solar production, home consumption, import, export, electricity cost and savings.

Analytics — Hourly Pattern (7-day avg)

Average grid usage by hour of day over 7 days — useful for spotting recurring load patterns.

Solar Forecast (7-day)

7-day solar yield forecast (Open-Meteo) with estimated kWh output, sun hours, cloud cover and temperature.

Forecast vs Actual Solar

Comparison of forecast vs actual solar production with over/under-estimation tracking.

Quick Start

1. Configure credentials

cp .env.example .env
# Edit .env with your Anker credentials and IOMeter IP

2. Run with Docker

docker compose up -d

Open http://localhost:8000 (redirects to /dashboard/).

3. Run locally (development)

pip install -r requirements.txt
# Set environment variables or create .env
python -m app.main

Strategy Algorithm

Goal: Keep grid meter reading ≈ 0 W. Prefer slight export over any import.

Layer 1 — Spike Detection

• Maintains rolling 120 s window of meter readings
• New spike detected when grid import exceeds grid_target + min_change_threshold
• Spike is observed for up to spike_observation_period (default 90 s)
• Matched against appliance profiles:
  • ignore (e.g. microwave ≤ 120 s): no load change
  • adjust (e.g. oven > 3 min): triggers load increase

Layer 2 — Load Adjustment

target_load = current_load + (meter_reading − grid_target)

Clamped to [0, 800] W, rounded to 10 W steps. Only applied if change > min_change_threshold.

Layer 3 — Cooldown

• cooldown_period (default 45 s) after each change prevents oscillation
• Emergency threshold (default 500 W) overrides cooldown for large spikes

Configuration

All parameters are editable from the dashboard UI under Strategy Configuration:

Parameter	Default	Description
polling_interval_s	5	IOMeter read frequency
reaction_delay_s	15	Wait before first adjustment
spike_observation_period_s	90	Observation window
cooldown_period_s	45	Post-change cooldown
min_change_threshold_w	30	Hysteresis
emergency_threshold_w	500	Override cooldown threshold
grid_target_w	−10	Target meter reading
max_load_w	800	Maximum allowed load
electricity_price_eur_kwh	0.28	For cost calculation

API Endpoints

Method	Path	Description
GET	/api/data	Combined dashboard snapshot
GET	/api/status	System health
POST	/api/set-load	Manual load control {"load": 400}
POST	/api/auto-mode?enabled=true	Toggle auto mode
GET/POST	/api/config	Read/write strategy parameters
GET/POST/DELETE	/api/profiles	Manage appliance profiles
GET	/api/load-history	Load change log
GET	/api/meter-history	Meter reading history
GET	/api/power-events	Detected power events
GET	/api/daily-energy	Daily energy summary
GET	/health	Health check

Deployment

Designed for a local Mac Mini with Docker. Uses network_mode: host so the container can reach the IOMeter device on the local network.

docker compose up -d --build

Data (SQLite + auth cache) persists in the solar-data Docker volume.

Acknowledgements

The Anker Solix cloud API client (app/api/) is adapted from the reverse-engineering work by Thomas Luther. Without that foundational work this project would not be possible. His project is also MIT-licensed — if you find this useful, consider starring his repository too.

License

MIT — see LICENSE.