Abstract: This study presents a genetic programming-based symbolic regression approach for the automated discovery of chemical reaction kinetic differential equations from experimental data. Using three distinct sets of literature-reported optical rotation data from sucrose hydrolysis reactions, we validated the method's efficacy in identifying chemical reaction differential equations while examining the influence of numerical errors on equation recognition. Our findings demonstrate that this approach not only accurately identifies the mathematical expressions of sucrose hydrolysis kinetic differential equations that align with theoretical derivations, but also optimizes the estimation of kinetic parameters. This research facilitates students' understanding of the process of deriving scientific laws from experimental data, bridging the conceptual gap between experimental observation and scientific discovery. Furthermore, it provides robust technical support for implementing the emerging paradigm of “AI-assisted scientific discovery” in university-level chemistry education.

Key words: Symbolic regression, Differential equations, Sucrose hydrolysis, Rate constant