Scholarlink

<p>Power electronic devices, such as metal–oxide–semiconductor field-effect transistors (MOSFETs), play a significant role in a wide range of applications. However, the remaining useful life (RUL) of MOSFETs has always been difficult to predict accurately. First, the interaction principle between the degradation state of MOSFETs and the sensitive electrical parameters is organized on the basis of structural and packaging characteristics, and the feasibility of using conduction resistance as a precursor parameter for the failure of MOSFETs is clarified. Second, in contrast with traditional RUL prediction methods, a MOSFET RUL prediction method based on degradation models is proposed. By combining with the MOSFET thermal overload aging dataset that is publicly available from NASA- Prognostics Center of Excellence in the United States, the performance of both algorithms in predicting the RUL of MOSFETs is systematically evaluated and compared. The results demonstrate that the extended Kalman filter algorithm not only achieves higher prediction accuracy (≈ 95%) but also exhibits better stability and significantly faster computational speed (one order of magnitude less time per iteration) compared with the nonlinear regression algorithm. The findings underscore the importance of selecting appropriate algorithms for prognostics and highlight the growing need for interdisciplinary expertise that combines power electronics, data modeling, and algorithm implementation in advanced engineering research and education.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Estimation method for the remaining useful life of silicon mosfets based on degradation models and deep reflections on its technical educational requirements

  • 期刊论文

  • 发表日期：

    2026年3月4日

查看全文
  • Lingfeng Shao,
  • Guoqing Xu,
  • Weiwei Wei

摘要

Power electronic devices, such as metal–oxide–semiconductor field-effect transistors (MOSFETs), play a significant role in a wide range of applications. However, the remaining useful life (RUL) of MOSFETs has always been difficult to predict accurately. First, the interaction principle between the degradation state of MOSFETs and the sensitive electrical parameters is organized on the basis of structural and packaging characteristics, and the feasibility of using conduction resistance as a precursor parameter for the failure of MOSFETs is clarified. Second, in contrast with traditional RUL prediction methods, a MOSFET RUL prediction method based on degradation models is proposed. By combining with the MOSFET thermal overload aging dataset that is publicly available from NASA- Prognostics Center of Excellence in the United States, the performance of both algorithms in predicting the RUL of MOSFETs is systematically evaluated and compared. The results demonstrate that the extended Kalman filter algorithm not only achieves higher prediction accuracy (≈ 95%) but also exhibits better stability and significantly faster computational speed (one order of magnitude less time per iteration) compared with the nonlinear regression algorithm. The findings underscore the importance of selecting appropriate algorithms for prognostics and highlight the growing need for interdisciplinary expertise that combines power electronics, data modeling, and algorithm implementation in advanced engineering research and education.