Exploring Daily Variation Patterns on Total Harmonic Distortion Long-Term Measurements in Traction Converter Stations using Data Analytics

In Sweden, traction converter stations supply the traction system and deliver power to the trains. Those stations have the role of converting the 50 Hz public grid frequency to the railway operating frequency, which in Sweden is 16 ⅔ Hz. Measurements regarding harmonic parameters, including total harmonic distortion (THD), are imperative to keep the power quality in good performance and provide the operators with more information for investigation or compliance. This work proposes methods for assessing patterns on THD daily variation measurements in traction converter stations. For that, statistical analysis of data is explored together with a technique based on deep autoencoder (DAE). The methods are applied in measurements from five different traction converter stations, considering both the public and railway grid sides. The outcomes provide insights into the railway operation and distortion levels for long-term periods, also illustrating how different methodologies can be used for time-varying harmonic distortion, which can be deployed in any system with similar characteristics. The process brings up several good pieces of information and successfully finds patterns for daily variation in harmonic distortion. Such application can be extended to other measurements in electrified road systems, electric vehicle charging infrastructure, microgrids, wind parks, etc.