A gate voltage monitoring method and device for an IGBT module under high-speed switching transient

By combining a physically guided hybrid kernel function with Gaussian process regression, the problem of gate voltage signal distortion in IGBT modules at low sampling rates is solved, achieving high-precision online monitoring and health assessment. It is applicable to IGBT modules and devices such as SiC and GaN, and has good generalization ability and low cost.

CN122260065APending Publication Date: 2026-06-23HARBIN ENG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HARBIN ENG UNIV
Filing Date
2026-03-14
Publication Date
2026-06-23

AI Technical Summary

Technical Problem

In the high-speed switching transients of IGBT modules, the low sampling rate in existing technologies leads to distortion of the gate voltage signal. Traditional compressed sensing methods cannot achieve high-precision reconstruction, and deep learning algorithms have weak generalization ability under different operating conditions, failing to provide reliable health assessments.

Method used

By combining a physics-guided hybrid kernel function with Gaussian process regression, a Gaussian process prior model is constructed. Induced variables and constrained variational inference are introduced to optimize hyperparameters to reconstruct the gate voltage waveform at a high sampling rate and provide confidence quantification.

Benefits of technology

It achieves high-fidelity reconstruction at extremely low sampling rates, extracts health-sensitive parameters with sub-nanosecond accuracy, possesses unsupervised characteristics, is suitable for different operating conditions and devices, reduces hardware costs, and improves monitoring efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122260065A_ABST
    Figure CN122260065A_ABST
Patent Text Reader

Abstract

This invention discloses a method and device for monitoring the gate voltage of an IGBT module during high-speed switching transients. By constructing a physically guided hybrid kernel function and embedding a Gaussian process-constrained variational inference framework, it achieves real-time and accurate reconstruction of low-sampling-rate gate voltages into extremely high-sampling-rate waveforms without pre-training. Addressing the complex physical characteristics exhibited by the gate voltage during high-speed switching, such as the quasi-steady-state plateau, rapid jumps caused by large switching transient gate voltage change rates, and accompanying damped oscillations, the designed physically guided hybrid kernel function explicitly encodes these multimodal dynamic characteristics. Gaussian process modeling is employed to capture smooth transient dynamic features and provide post-calibration uncertainty assessment. This invention enables accurate IGBT module status monitoring information at extremely low sampling rates, improving signal transmission and storage efficiency by more than a hundred times and providing a reliable technical means for remote online status monitoring systems.
Need to check novelty before this filing date? Find Prior Art