Kinetic Modeling and Multiobjective Optimization of Ibuprofen Synthesis Using Machine Learning

Area of Science:

• Chemical Engineering
• Computational Chemistry
• Machine Learning Applications

Background:

• Ibuprofen synthesis requires precise control over reaction parameters for efficiency and cost-effectiveness.
• Traditional optimization methods can be time-consuming and may not capture complex parameter interactions.

Purpose of the Study:

• To develop and apply integrated machine learning tools for modeling and optimizing the ibuprofen synthesis process.
• To identify critical input variables and optimal operating conditions for ibuprofen production.

Main Methods:

• Creation of a large database (39,460 combinations) using chemical reaction theory, validated experimentally.
• Application of a CatBoost meta-model optimized with a snow ablation optimizer.
• Utilizing SHAP values for variable importance analysis and NSGA-II for multiobjective optimization.
• Performing Monte Carlo simulations for uncertainty analysis.

Main Results:

• The optimized CatBoost model accurately predicts reaction time, conversion rate, and production cost.
• Critical parameters identified include catalyst precursor (L2PdCl2), H+, and H2O concentrations.
• Optimal catalyst concentration range (0.002-0.01 mol/m3) found for high conversion and low cost.
• Reaction time shows high sensitivity to parameter fluctuations, with nonlinear behavior.

Conclusions:

• Integrated machine learning effectively models and optimizes ibuprofen synthesis.
• Data-driven insights provide quantitative guidance for rational process design.
• The study demonstrates a powerful approach combining physics-based modeling with machine learning for chemical process optimization.