Energy Hub MILP/MIQP Optimization

Pyomo-based optimization system for multi-device energy hub scheduling. Supports both linear (MILP) and quadratic (MIQP) optimization using IPOPT solver.

Features

• Multi-device energy hub optimization (CHP, PEM, Battery, Methanation, Heat Pump)
• MILP and MIQP solving via IPOPT
• MQTT integration for real-time scheduling
• Comprehensive test suite with 80%+ code coverage
• All configuration centralized in project_config.json

Requirements

• Python 3.8+
• IPOPT solver (install separately, e.g., via conda: conda install -c conda-forge ipopt)

Installation

Using uv (recommended)

# Clone the repository
cd energy-hub-milp

# Create virtual environment and install dependencies
uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -e .

# Install development dependencies (for testing)
uv pip install -e ".[dev]"

Dependencies

All dependencies are listed in pyproject.toml:

• pyomo - Optimization modeling
• numpy - Numerical computing
• pandas - Data handling
• matplotlib, seaborn - Visualization
• paho-mqtt - MQTT integration
• requests - HTTP requests

Dev dependencies:

• pytest - Testing framework
• pytest-cov - Coverage reporting

Configuration

All configuration is managed through project_config.json. Key configuration sections:

Paths

• paths.data_dir - Data directory (default: "Daten")
• paths.output_dir - Output directory (default: "output")
• paths.logs_dir - Logs directory (default: "logs")

Solver

• solver.ipopt_executable - IPOPT executable path

MQTT

MQTT credentials are stored in credentials.json. Update this file with your broker details before running the MQTT runner.

Running Tests

# Run all tests with coverage
pytest tests/ --cov=. --cov-report=term-missing

# Run tests quietly
pytest tests/ -q

# Run specific test file
pytest tests/test_schedule_generator.py -v

Running the Optimization

Full workflow (quadratic + linear + statistics)

python main.py

Quadratic optimization only (MIQP)

python main.py --quadratic

Single-day scheduling

from schedule_generator import connect_and_schedule
connect_and_schedule(
    timeframe=96,
    step_length=900,  # 15 minutes in seconds
    all_results_file="output.json",
    result_schedule_file="schedule.csv",
    quadratic=True,
    load_ts="TargetFromResults.pkl",
    days=5
)

MQTT Runner

python mqtt_runner.py

Project Structure

milp/
├── config.py                 # Configuration loader
├── project_config.json       # Project configuration
├── main.py                   # Main optimization script
├── schedule_generator.py     # Optimization orchestration
├── mqtt_runner.py           # MQTT integration
├── indexed_model.py         # Pyomo model
├── components/              # Device models
│   ├── converter.py
│   ├── storage.py
│   ├── grid.py
│   ├── target.py
│   └── heat_target.py
├── tests/                   # Test suite
│   ├── test_indexed_model.py
│   ├── test_schedule_generator.py
│   ├── test_mqtt_runner.py
│   ├── test_components.py
│   └── test_utils.py
├── Daten/                   # Data and results
│   └── results/
└── .gitlab-ci.yml          # CI/CD pipeline

Testing in GitLab CI

The .gitlab-ci.yml includes a test stage that runs automatically with coverage enforcement:

test:
  stage: test
  script:
    - pytest tests/ --cov=. --cov-report=term-missing --cov-report=html --cov-fail-under=80

Tests run automatically on every push to the repository. The pipeline enforces minimum 80% code coverage.

License

MIT