Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/A6E9700D-8942-4BAC-B3CF-A612E2BBB63E | ||
| Year | 2022 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Georgios Charalampidis, "A Dataset generator for smart grid ecosystems populated with electric vehicles", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2022 https://doi.org/10.26233/heallink.tuc.91590 | ||
| Appears in Collections |
The widespread use of electric mobility technologies such as the Electric Vehicles(EVs), poses certain challenges both from the technical, and the socioeconomic points of view. In order to address these, research must utilize related data that originates from realistic sources. However, in the typical case, such data contains private and sensitive information that cannot be made available to the researchers. As a result, research makes excessive use of the few datasets that are publicly available. At the same time, the majority of the available data, contain only a few customers, while hundreds of thousands are required. To overcome this obstacle, synthetic data can be used, which nevertheless originates from models with relationships that sufficiently capture the properties of the actual real-world datasets. In this thesis, we design a dataset generator for the domain of EVs charging management in Smart Grid settings. The generator (i) takes as input anonymized data, describing different energy generation and demand types, as well as charging profiles of EVs and corresponding trip and type information; (ii) employs a variety of models—in particular Histograms, Kernel Density Estimation, Generative Adversarial Networks, and Frequency Tables—using this data as training sets; and thus (iii) generates new synthetic data, not identical to the input, but adhering to the same principles, and relationships. The proposed dataset generator also produces respective summarizations, which includesbarplots and histograms to visualize the results, and different metrics in order toquantify a ‘distance’ between the distributions under comparison. These summarizations give us a complete picture of the generated data and they are particularly useful for detecting correspondence issues. Last but not least, the dataset generator is available via an online repository and it can readily be incorporated by third parties in their research activities.
Copyright © DIAS 2013
