Time-Domain Power Theory Applied to Waveform Distortion Assessment of AC Railways

The distortion patterns of an AC locomotive are considered based on a large set of experimental data. The analysis for harmonics, interharmonics and supraharmonics up to 25 kHz is carried out using Fryze's power theory and separating active and non-active current signals. Larger non-active components in the harmonic interval are found at low power levels, whereas active current becomes prevalent when tractioning at high power. Supraharmonics instead have a prevalence of active current, that is relevant to the estimate of power losses and energy efficiency. The current signals were also subject to clustering using unsupervised deep learning, synthesizing the behavior of more than 5000 current snapshots of 300 ms duration with only 5 distinct patterns.