Application of EMR Signature in Health Assessment and Monitoring of IGBT-Based Converters

In this paper, a non-invasive method based on electromagnetic radiation (EMR) signature is proposed for the health assessment of the converters. Converters use insulated gate bipolar transistor (IGBT) because of its robustness. However, it suffers from internal degradation due to the rapid power cycle and thermal stress. Degradation of IGBT primarily increases the turn-
off
time as a consequence of elevated junction temperature. The coupling of IGBT with the parasitics of the circuit results in EMR and depends on the turn-
off
time of IGBT. It has been found that the EMR signature reduces with the increase in turn-
off
time. This concept is taken forward for the health assessment of the converters. However, in a practical scenario, the EMR generated from multiple converters get mixed up. This is commonly known as a problem of near-field source localization. We use a uniform linear array to capture the EMR signals near the converters. Subsequently, ESPRIT and MUSIC algorithms are used to localize all the converters. An inverse transformation of the localization algorithm separates the EMR signature of all the individual converters. The proposed health assessment algorithm computes the degradation level of converters, and the experimental results validate the proposed approach.