![Institutional Repository [SANDBOX]](images/dias.logo.png "Institutional Repository [SANDBOX]"){kind=logo}
Institutional Repository [SANDBOX]
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/DD42EAA9-0441-450D-A315-3FB710266E61 | ||
| Year | 2021 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Dimitrios Apostolakis, "Optimization of electric energy management in buildings based on machine learning techniques", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2021 https://doi.org/10.26233/heallink.tuc.91055 | ||
| Appears in Collections |
The continuously increasing energy demands of household consumers are causing higher electricity costs and environmental impact as well. This work concerns the implementation of machine learning algorithms, aiming at the management of electricity consumption in buildings. The system under consideration includes a photovoltaic system for generating electricity, a battery storage system and the electrical grid with which they are interconnected. An optimization algorithm has been developed to calculate the optimal time series of electricity exchanged with the grid to minimize the cost of electricity for the household consumer. At the same time, the cost of electricity is further reduced by applying an algorithm for scheduling the operating hours of individual electrical loads. The basic machine learning algorithm that is integrated in the optimization process is Q-Learning, while two more different versions are implemented, which are the Sarsa and the Double-Q-Learning algorithms, in order to compare their performance in the optimization problem. The results show that the energy management algorithm can reduce significantly the total electricity cost for the consumer.
Copyright © DIAS 2013
