Scholarlink

<p>Electrochemical impedance spectroscopy (EIS) is a widely used experimental technique whose interpretation depends heavily on the model selected by its operator. Here, we develop an approach for EIS analysis by integrating convolutional neural network (CNN) methodologies. Our technique utilizes the versatility of CNNs to extract characteristic feature shapes from Nyquist plots for equivalent circuit classification. By employing CNNs, we overcame the subjective bias inherent in manual EIS analysis, providing a more standardized and reproducible method. The adaptability of our model was enhanced by incorporating transfer learning, which allows the efficient application of pretrained kernels to new data. This resulted in a robust model capable of maintaining high accuracy levels, as demonstrated by an accuracy of &gt;80% in classifying electrochemical impedance spectra prepared with random parameters. The practicality of our approach was verified by analyzing the electropolymerization of pyrrole, indicating its potential for a wide range of electrochemical applications. This advancement in the application of machine learning to EIS represents a transformative step toward automating and refining electrochemical analysis.</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

Determination of equivalent circuits for electrochemical impedance spectra using convolutional neural networks

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月11日

查看全文
  • Long Duong Ha,
  • Sangkyu Park,
  • Donghyun Lee,
  • Hyeonsu Je,
  • Kwok-Fan Chow,
  • Doran L. Pennington,
  • Heeso Noh,
  • Christopher H. Hendon,
  • Seongpil Hwang,
  • Byoung-Yong Chang

摘要

Electrochemical impedance spectroscopy (EIS) is a widely used experimental technique whose interpretation depends heavily on the model selected by its operator. Here, we develop an approach for EIS analysis by integrating convolutional neural network (CNN) methodologies. Our technique utilizes the versatility of CNNs to extract characteristic feature shapes from Nyquist plots for equivalent circuit classification. By employing CNNs, we overcame the subjective bias inherent in manual EIS analysis, providing a more standardized and reproducible method. The adaptability of our model was enhanced by incorporating transfer learning, which allows the efficient application of pretrained kernels to new data. This resulted in a robust model capable of maintaining high accuracy levels, as demonstrated by an accuracy of >80% in classifying electrochemical impedance spectra prepared with random parameters. The practicality of our approach was verified by analyzing the electropolymerization of pyrrole, indicating its potential for a wide range of electrochemical applications. This advancement in the application of machine learning to EIS represents a transformative step toward automating and refining electrochemical analysis.