(569ba) Exploring the Role of Process Control and Catalyst Design in Methane Catalytic Decomposition: A Machine Learning Perspective

Methane catalytic pyrolysis offers a promising route for producing CO₂-free hydrogen, yet optimizing catalyst composition and process parameters through experimental studies has been resource-intensive. Machine learning techniques are increasingly employed to deepen the understanding of process mechanisms. In this work, machine learning models were developed to enhance understanding of the effects of process control and catalyst design on CH₄ conversion performance. Input parameters included CH₄ concentration in the supplied gas (%), gas hourly space velocity (GHSV) of the supplied gas (ml/(g×h)), pyrolysis temperature (°C), and reaction time (min). For catalyst design, the weight contents of the four most prevalent catalytic metals (Fe, Ni, Co, and Cu) and the four most utilized supports (Al₂O₃, SiO₂, TiO₂, and MgO), as well as the calcination temperature (°C) were chosen. The optimal CH₄ conversion machine learning model achieved a testing R² of 0.9999. To validate the model's practical utility, an initial verification experiment was conducted, employing a catalyst composition and process parameters different from those in the dataset. The results confirmed the model's high accuracy in real-world scenarios. This study further explored the intricacies of methane pyrolysis by analyzing the Shapley values and partial dependence plots generated by the model. These analyses provided valuable insights into the impact of process control and catalyst design parameters on methane catalytic decomposition, offering a fresh perspective on the subject.