Casagrande, Vittorio;
(2024)
Learning-based and distributed model predictive control architectures for resilient microgrid energy management.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
PhD_thesis_final.pdf - Accepted Version Download (963kB) | Preview |
Abstract
In this thesis innovative solutions for microgrid Energy Management System (EMS) are proposed to deal with uncertainty, privacy issues and the possible presence of faults. In recent decades, the increasing penetration of distributed energy resources, often coupled with storage systems, have boosted the segmentation of the traditional power distribution system into microgrids. At the top-level control of microgrids the EMS coordinates the microgrid agents, mainly for peak shaving and economic optimisation. In this work the three aforementioned problems are considered, while optimising traditional EMS objectives. The stochastic nature of renewable generators makes uncertainty one of the primary challenges in the EMS design. Secondly, the coordination of the agents requires sharing private information, such as future load power demands, and this can reveal details of a manufacturing process or controller insights which might be undesirable in a competitive economic environment or may lead to potential cyber threats. Thirdly, strategies to increase fault resilience must be developed to ensure power delivery during events events like blackouts. The proposed EMS is designed as a model predictive control algorithm. Uncertainty is addressed by designing an online learning method for neural networks, ensuring adaptability to possible changes in the environment, and by embedding the scheduling optimisation problem within the neural network. Hence, the controller is trained to directly optimise control performance, while implicitly learning system uncertainties and guaranteeing operational constraints. Privacy issues are tackled by leveraging distributed optimisation that solutions of the scheduling problem while keeping local information private. By implementing proactive scheduling and outage management strategies, the effect of possible faults is mitigated. Since faults usually are unlikely events, a scenario-based MPC method is designed to reduce conservativeness of control actions. The proposed algorithms are validated through extensive simulations employing real-world data and realistic microgrid configurations. The simulation code for the learning-based controller has been published in a GitHub repository.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Learning-based and distributed model predictive control architectures for resilient microgrid energy management |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2024. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Electronic and Electrical Eng |
| URI: | https://discovery.ucl.ac.uk/id/eprint/10187385 |
Loading...
Loading...Loading...Loading...Archive Staff Only
 |
View Item |
