Scholarlink

<p>In this work, a backpropagation artificial neural network (BP-ANN) was successfully built to achieve the accurate prediction of corrosion inhibition efficiency of amino acid-derivate corrosion inhibitor for Q235 carbon steel in 3.5% NaCl solution. The principal component analysis was performed among eight calculated quantum chemical values to obtain four main parameters affecting the corrosion inhibition efficiency, including the highest occupied molecular orbital (HOMO), the lowest non-occupied molecular orbital (LUMO), adsorption energy, and hardness. A dataset comprising 20 kinds of amino acid corrosion inhibitors is employed as training data to construct model. The optimal BP-ANN model is established based on 20-10-5-1 network architecture and BFGS learning algorithm. Values of corrosion inhibition efficiencies for five kinds of amino acid corrosion inhibitors were predicted using the established model; deviations between predicted and experimental data were less than 8%. This work provides a novel method to efficiently determine appropriate corrosion inhibitor from numerous organic molecules.</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

Machine Learning Approach to Predict Corrosion Inhibition Performance of Amino Acid-Derivate Corrosion Inhibitor for Carbon Steel in Neutral Medium

  • 期刊论文

  • 发表日期：

    2025年6月17日

查看全文
  • Bo-Kai Liao,
  • Shi-Quan Ma,
  • Hao-Lan Huang,
  • Hui-Zhen Li

摘要

In this work, a backpropagation artificial neural network (BP-ANN) was successfully built to achieve the accurate prediction of corrosion inhibition efficiency of amino acid-derivate corrosion inhibitor for Q235 carbon steel in 3.5% NaCl solution. The principal component analysis was performed among eight calculated quantum chemical values to obtain four main parameters affecting the corrosion inhibition efficiency, including the highest occupied molecular orbital (HOMO), the lowest non-occupied molecular orbital (LUMO), adsorption energy, and hardness. A dataset comprising 20 kinds of amino acid corrosion inhibitors is employed as training data to construct model. The optimal BP-ANN model is established based on 20-10-5-1 network architecture and BFGS learning algorithm. Values of corrosion inhibition efficiencies for five kinds of amino acid corrosion inhibitors were predicted using the established model; deviations between predicted and experimental data were less than 8%. This work provides a novel method to efficiently determine appropriate corrosion inhibitor from numerous organic molecules.