Abstract: Chemical oscillatory reactions constitute a distinctive category of non-equilibrium chemical processes that not only integrate multidisciplinary knowledge from chemistry, fluid mechanics, and computational science, but also provide an excellent platform for students to understand dynamic chemical processes and develop model-building competencies. Conventional university-level experiments on chemical oscillations encounter three major challenges: poor reproducibility, difficulties in controlling multiple variables, and the high complexity threshold for mathematical modeling, all of which impede students’ comprehension of complex reaction kinetics. This study presents an AI-augmented approach that establishes a novel pedagogical paradigm combining numerical simulation with experimental practice. Utilizing retrieval-augmented generation technology, a localized knowledge base, and large language models, we developed an interactive chemical oscillation simulator that accurately reproduces the dynamic behaviors of both Belousov-Zhabotinsky oscillation and iodine clock reactions. The simulator demonstrates remarkable consistency with experimental data, showing ≤ 2% relative deviation in oscillation periods and concentration-time profiles. Implementation results indicate a 92% system transfer success rate and reduction of parameter retrieval time from hours to approximately 2 minutes. The dual-mode “experimental-simulation parallel” teaching approach significantly enhanced students’ understanding of nonlinear dynamics, computational thinking, and AI application skills without requiring additional instructional hours. This methodology effectively lowers the technical barriers in teaching chemical oscillatory reaction kinetics while establishing a scalable, innovative framework for experimental education.

Key words: Chemical kinetics experiments, AI-augmented, Numerical simulation, Pedagogical innovation