China International Youth Conference on Electrical Engineeing (CIYCEE 2025)

Insulated gate bipolar transistors (IGBTs) are critical in high-power systems, where reliable monitoring of switching transients is essential for condition assessment. Conventional approaches require high sample-level acquisition, leading to high cost and data burdens. This paper proposes a wavelet-optimized orthogonal matching pursuit (WO-OMP) framework for compressive reconstruction of IGBT switching transients. By leveraging wavelet-domain sparsity and adaptive reconstruction, the method reduces sampling rates while maintaining fidelity. Experiments on a double-pulse test platform show that WO-OMP achieves nanosecond-level accuracy in turn-off time toff, with relative errors below 1% under compression ratios up to 100. Performance metrics including RMSE, PRD, and PSNR further verify that waveform fidelity and timing precision are preserved under sub-Nyquist sampling. The results demonstrate the feasibility of cost-effective and real-time IGBT monitoring.

Compressed sensing,Condition monitoring,Sparse reconstruction,Wavelet transform,IGBT switches