Multi-parameter detection of bond wire lift-off, current, and temperature in IGBT modules via gate voltage waveforms and CNNs with digital gate control

This is the authors' abstract. We don't add key points for this paper.

A new method for multi-parameter detection of bond wire lift-off, emitter current, and junction temperature using gate voltage waveforms and a convolutional neural network is proposed for the condition monitoring of power modules. This method was demonstrated to classify 80 levels for the full combination of various parameters. In addition, digital gate control (DGC) was utilized to improve not only the switching characteristics but also the detection accuracy. The experimental results show that the sensitivity of the gate voltage waveforms changed significantly due to the influence of the combined parameters. The detection accuracy depends on the control conditions of DGC, and optimized conditions achieved a high accuracy of over 96%, even for multi-parameter detection.