Welcome to the repository of my master thesis.
Heating in private households accounted for 26% of total energy consumed in Germany in 2020, which is a major contributor to the emissions generated today. Heat pumps are a promising alternative for heat generation and are a key technology in achieving our goals of the German energy transformation which includes the reduction of gas emissions by 55% until 2030, compared to 1990. Today, the majority of heat pumps in the field are controlled by a simple heating curve, which is a naive mapping of the current outdoor temperature to a control action. An alternative approach is Model Predictive Control (MPC) which was applied in multiple research works to heat pump control. However, MPC is heavily dependent on the building model, which has several disadvantages. Motivated by this and by recent breakthroughs in the field, this work applied deep reinforcement learning (DRL) to heat pump control in a simulated environment.
Unfortunately, the simulation and the price data which were used in the thesis could not be made public (yet). Therefore, only the deep reinforcement learning part of the project could be published here.
You can find the thesis here
The repository is structured as follows:
This folder contains the data of the baseline methods wich were used for comparison.
This folder contains the code which was used in the master thesis. Further information can be found in the provided jupyter notebooks.
The models trained are stored in this folder