Degradation assessment of an IGBT with recurrence analysis and Kalman filter based data fusion

Degradation is a nonstationary process that necessitates using a specialized measurement tool. In light of this, a degradation assessment tool has been presented in this study through the use of recurrence analysis. A fused signal has been constructed through the linear Kalman Filter Fusion Application (KFFA) to assess the growing degradation within the system. This approach aims to combine the features provided by different mediums of the system. A case study is presented using the collector-emitter current and package temperature signals collected during an IGBT's accelerated aging experiment. Time-frequency and recurrence analyses of fused and collected signals are evaluated to interpret the degradation through the nonstationary characteristics of the signals. The study has two main purposes: to investigate the combination capability of KFFA deeply for nonstationary processes and to define recurrence analysis as a degradation assessment tool. The study explores the relationship between degradation mechanisms and chaotic tendencies. Employing recurrence analysis successfully captures the decay moment of the degradation process. The study is distinctive in its investigation of the recurrence characteristics of a fused signal, which is the mixed representation of multiple sensor data. Additionally, the study demonstrated that the linear KFFA is highly effective in providing integrated and more reliable information even for nonstationary signals. With these contributions, the paper proposes a novel degradation assessment tool driven by recurrence analysis and supported by KFFA.