As an emerging route to achieving carbon neutrality, plasma-catalyzed CO2 hydrogenation to CH3OH faces challenges in defining suitable operating conditions and achieving industrial implementation. To address these issues effectively, dimensional analysis provides a promising quantitative approach for optimizing such reaction systems. Dimensional analysis enables the extraction of dimensionless numbers from complex physical systems, providing a compact representation of coupled operating variables. This research proposes a dimensionless automated machine learning (AML) framework based on Bayesian optimization (BO) to identify candidate dimensionless groups and empirical operating correlations from heterogeneous literature-derived experimental data. Datasets are constructed by data augmentation and completion for dimensionless analysis. Models built on the selected dimensionless groups achieve a maximum R2 of 0.92, and their residuals are centered near zero with a bias of 0.01. The selected dimensionless numbers show clear correlations with CO2 conversion, CH3OH selectivity and CH3OH yield in plasma-catalyzed CO2 hydrogenation to CH3OH. Overall, the framework provides compact dimensionless operating maps for methanol-oriented operating-condition screening and offers a quantitative reference for prioritizing conditions in subsequent experiments.


扫码关注我们
求助内容:
应助结果提醒方式:

