Conference proceeding
PINN-Assisted Physical Model of SiC MOSFETs: A Leap in Efficiency and Accuracy
IEEE Energy Conversion Congress and Exposition, pp 7062-7067
20 Oct 2024
Abstract
Accurately modeling the behaviors of Silicon Carbide (SiC) MOSFETs is crucial for power electronics system analysis. However, conventional SiC MOSFET models often lack sufficient physical detail, failing to accurately depict the device's physical behavior. Surface-potential-based models provide a more granular physical schema, facilitating enhanced accuracy regarding devices' characteristics. Yet, the traditional formula for solving surface potential is implicit and intricate, which hinders achieving an equilibrium between simulation fidelity and efficiency. To address these issues, this paper proposes a Physics-Informed Neural Network (PINN) to achieve rapid calculation of surface potentials in SiC MOSFETs. This model requires minimal training data to deliver precise and rapid characterizations of SiC MOSFETs at the microscopic level. Simulation results show that the proposed model can be speeded up by 62% over the conventional iterative-based method with the same accuracy.
Metrics
7 Record Views
Details
- Title
- PINN-Assisted Physical Model of SiC MOSFETs: A Leap in Efficiency and Accuracy
- Creators
- Xinlian Li - Shanghai Jiao Tong UniversityChong Zhu - Shanghai Jiao Tong UniversityYansong Lu - Shanghai Jiao Tong UniversityXu Lu - Shanghai Jiao Tong UniversityFei Lu - Drexel UniversityXi Zhang - Shanghai Jiao Tong University
- Publication Details
- IEEE Energy Conversion Congress and Exposition, pp 7062-7067
- Publisher
- IEEE
- Grant note
- National Natural Science Foundation of China (10.13039/501100001809)
- Resource Type
- Conference proceeding
- Language
- English
- Academic Unit
- Electrical and Computer Engineering
- Scopus ID
- 2-s2.0-86000452170
- Other Identifier
- 991022028225304721