Scholarlink

<p>The damage evolution concept is based on different coating performance stages for barrier coatings degradation, the fact that real-time monitoring gives rise to the characterization of such an electrochemical system, and the prediction and extrapolation of events around time series prediction (TPS) can occur. The nature of the acquired information has considered artificial neural network models (ANN) for quantitative characterization of parameters correlating performance or coating failure modes; the ANN learns features and dynamics directly from the data rather than depending on parameter optimization that considers domain experience. This work considers a unique set of time series data monitoring of a physical barrier coating performance with electrochemical impedance spectroscopy (EIS) measurements over 1000&#xa0;days. The algorithm built made predictions about different degradation stages for coating barriers. Different coating thickness samples at 0%, 30%, and 100% (full failure) of the original coating thickness permit the characterization by simulating natural degradation. EIS parameters helped to quantify the state of the coating by analyzing the time series of the impedance magnitude at different frequencies. Using three distinct ANN modeling modes, we minimized the error by selecting the most appropriate ANN approach. The low-frequency data was a critical parameter not only for the physical mechanistic interpretation but also due to the dynamic changes occurring in a specific time interval. To ensure a comprehensive understanding of corrosion, we integrate the underlying mechanisms into our analysis. The designed model was able to predict the real and imaginary impedance along several frequencies after 1000&#xa0;days. As a result, we could generate a comprehensive experimental-theoretical framework, including electrochemical impedance spectroscopy testing, and verification by using the Lin-KK validity approach.</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

Corrosion prediction based on damage evolution framework by using artificial neural networks for multilayer coating/substrate interface

  • 期刊论文

  • 发表日期：

    2025年8月14日

查看全文
  • Victor Ponce,
  • Seongkoo Cho,
  • Changkyu Kim,
  • Seungho Ahn,
  • Homero Castaneda

摘要

The damage evolution concept is based on different coating performance stages for barrier coatings degradation, the fact that real-time monitoring gives rise to the characterization of such an electrochemical system, and the prediction and extrapolation of events around time series prediction (TPS) can occur. The nature of the acquired information has considered artificial neural network models (ANN) for quantitative characterization of parameters correlating performance or coating failure modes; the ANN learns features and dynamics directly from the data rather than depending on parameter optimization that considers domain experience. This work considers a unique set of time series data monitoring of a physical barrier coating performance with electrochemical impedance spectroscopy (EIS) measurements over 1000 days. The algorithm built made predictions about different degradation stages for coating barriers. Different coating thickness samples at 0%, 30%, and 100% (full failure) of the original coating thickness permit the characterization by simulating natural degradation. EIS parameters helped to quantify the state of the coating by analyzing the time series of the impedance magnitude at different frequencies. Using three distinct ANN modeling modes, we minimized the error by selecting the most appropriate ANN approach. The low-frequency data was a critical parameter not only for the physical mechanistic interpretation but also due to the dynamic changes occurring in a specific time interval. To ensure a comprehensive understanding of corrosion, we integrate the underlying mechanisms into our analysis. The designed model was able to predict the real and imaginary impedance along several frequencies after 1000 days. As a result, we could generate a comprehensive experimental-theoretical framework, including electrochemical impedance spectroscopy testing, and verification by using the Lin-KK validity approach.