Machine Learning-Enhanced Classification and Analysis of Modified Carbon Paste Electrodes for Methanol Fuel Cell Applications

Oukbab, Mustapha; Chtaini, Imane; Chtaini, Yassine; Najih, Rachida; Chtaini, Abdelilah

doi:10.22034/abec.2026.741068

Machine Learning-Enhanced Classification and Analysis of Modified Carbon Paste Electrodes for Methanol Fuel Cell Applications

Document Type : Original Article

Authors

Mustapha Oukbab ¹

Imane Chtaini ²

Yassine Chtaini ²

Rachida Najih ¹

Abdelilah Chtaini ¹

¹ Molecular Electrochemistry and Inorganic Materials Team, Beni Mellal Faculty of Science and Technology, Morocco

² Independent Researcher

10.22034/abec.2026.741068

Abstract

This study presents a comprehensive machine learning approach for the classification and analysis of modified carbon paste electrodes (CPE) designed for methanol fuel cell anodes. Four electrode configurations were systematically investigated: unmodified CPE, zinc-modified CPE (CPE/Zn), polymer-enhanced CPE/Zn with 1,4-cis polymyrcene (CPE/Zn/Polymer), and bio-modified CPE/Zn/Polymer/Bacteria electrodes using Escherichia coli biofilms. Advanced machine learning techniques including Convolutional Neural Networks (CNN), Random Forest, Support Vector Machines, Long Short-Term Memory networks, and Bayesian analysis were employed to analyze cyclic voltammetry data and predict electrode performance for methanol oxidation. The deep CNN achieved superior classification accuracy of 98% (AUC = 0.98) compared to Random Forest (91%) and SVM (87%). Principal Component Analysis revealed that electrode modifications systematically altered electrochemical properties, with methanol oxidation peak current and potential being the most discriminative features (47% combined importance). Autoencoder-based denoising successfully reconstructed CV curves with minimal error, while LSTM networks demonstrated predictive capability for temporal electrode behavior and methanol oxidation efficiency. The integration of machine learning with fundamental electrochemical principles provides a powerful framework for automated electrode characterization, quality control, and performance optimization in methanol fuel cell applications.

Keywords

Machine learning

Methanol fuel cells

Carbon paste electrodes

Cyclic voltammetry

Deep learning

Electrode modification

Bioelectrochemistry

Zinc catalysis

Subjects