| Meno: | Júlia
|
|---|
| Priezvisko: | Lichmanová
|
|---|
| Názov: | Deep Learning applied to Anomaly Detection in Meteorological Time Series
|
|---|
| Vedúci: | MSc. Philipp Roberto Miotti
|
|---|
| Rok: | 2023
|
|---|
| Kµúčové slová: | deep learning, anomaly detection, quality control, transformer, meteorological time series
|
|---|
| Abstrakt: | Quality control is an important part that ensures certain standards, accuracy and reliability in data of various data-driven domains. Particularly in the context of solar energy systems quality control plays a key role in validating satellite-based and ground-measured data. Current quality control methods primarily employ statistical checks and limit-based outlier detection. The advancement of computational resources and availability of large amounts of data have created the foundation for machine learning and deep learning tools that can potentially improve the quality control process. This thesis explores the application of deep learning methods, specifically Anomaly Transformer and TranAD based on transformer architecture, for detecting unrealistic values in ground-measured meteorological data.
|
|---|