A method for optimizing catalyst preparation conditions based on machine learning and genetic algorithm

Junhui Ao

doi:10.1016/j.mcat.2026.116334

A method for optimizing catalyst preparation conditions based on machine learning and genetic algorithm

2026-09-12

0

PRE

AI

DOI：10.1016/j.mcat.2026.116334

Original

Original request for help

Save

Abstract

En 中文

• An RF-GA coupled intelligent framework is proposed to optimize Mn-Ca/γ-Al₂O₃ catalyst for xylene catalytic oxidation. • Random Forest achieves superior prediction performance with test set R² = 0.949 for pollutant removal efficiency. • SHAP analysis reveals the Ca/Mn molar ratio (43.94%) and calcination temperature (21.96%) are the dominant influencing factors. • GA global search avoids local optima of traditional experiments, obtaining optimal preparation parameters: 5.4% loading, Ca:Mn = 1:4, 423 °C, 3.75 h. • The optimized catalyst reaches 93.7% xylene removal rate, higher than the 91.2% of single-factor local optimum conditions. This data-driven strategy offers a low-carbon reusable paradigm for non-noble heterogeneous catalyst development.

Keywords:

Catalyst optimization

Machine learning

Multiphase catalytic technology

A method for optimizing catalyst preparation conditions based on machine learning and genetic algorithm

Abstract

Journal

Organization

Cited Papers