Scholarlink

<p>Electronic components such as capacitors and transistors play critical roles in aerospace, military, and communication systems, where long-term reliability is essential. However, long-life components often lack complete degradation data, and traditional accelerated life testing (ALT) methods are costly and time-consuming. Previous research has explored FEM-based simulations for stress analysis, ALT for empirical degradation tracking, and machine learning for predictive modeling. Yet, each approach faces limitations in accuracy, data availability, or physical interpretability. In this study, we investigate the failure mechanisms of solid tantalum capacitors and transistors and identify sensitive parameters affecting lifetime performance. Using the Arrhenius model in combination with Icepak thermal simulations, we generate missing life data under stress conditions. These are then integrated with electrical parameters as inputs to a back-propagation (BP) neural network to train a life prediction model. Experimental results demonstrate that our model achieves prediction errors below 5%, offering an efficient, generalizable framework for component reliability forecasting.</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

Research on reliability of capacitors and transistors based on BP neural network and Icepak simulation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年6月5日

查看全文
  • Yi He,
  • Xiaoqing Yang,
  • Xiaoyu Zhai

摘要

Electronic components such as capacitors and transistors play critical roles in aerospace, military, and communication systems, where long-term reliability is essential. However, long-life components often lack complete degradation data, and traditional accelerated life testing (ALT) methods are costly and time-consuming. Previous research has explored FEM-based simulations for stress analysis, ALT for empirical degradation tracking, and machine learning for predictive modeling. Yet, each approach faces limitations in accuracy, data availability, or physical interpretability. In this study, we investigate the failure mechanisms of solid tantalum capacitors and transistors and identify sensitive parameters affecting lifetime performance. Using the Arrhenius model in combination with Icepak thermal simulations, we generate missing life data under stress conditions. These are then integrated with electrical parameters as inputs to a back-propagation (BP) neural network to train a life prediction model. Experimental results demonstrate that our model achieves prediction errors below 5%, offering an efficient, generalizable framework for component reliability forecasting.