Data-driven MPC for Micro-grid operation

This folder contains the code the master thesis Data-driven optimization of Micro-grid operation.

Authors:

• Nicolai Hoel
• Theodor Husefest

Supervisors:

• Sebastien Gros
• Phillip Maree

All data used in the experiments can be found in src/data.

Installation

Install the packages needed with ´pip install -r requirements.txt´.

Running the code

The code runs by using the main file, ´python3 main.py´

The configuration of the code happens in config.yml.

• Simulation horizon - number of hours to simulate the MPC for
• Prediction horizon - horizon of the MPC, in hours
• N_scenarios - Number of scenarios to include. Only 1, 3, 7 and 9 are valid
• Perfect predictions - Bool. If False, prediction methods are used.
• Plot - Bool. If True, figures are plotted.
• Foldername - If provided, a folder with the name will be created with all plots. If black, nothing happens
• Year, month, day - Which day to start the simulation
• Actions per hour - Sampling time of the MPC. Should always be 6
• Logpath - Which folder the logs are saved
• files - Where to find different files.
• Battery - Specifies the parameters of the battery. Read thesis for details.
• System - Specifies the parameters of the MPC. Read thesis for details.