This article presents a comprehensive Python framework for optimizing lithium battery manufacturing parameters using machine learning surrogate models, multi-objective genetic algorithms (NSGA-II), and SHAP explainability. The approach trains 11 different ML models to predict battery performance, then uses NSGA-II to find optimal process parameters that balance multiple objectives. SHAP analysis provides interpretability by identifying which parameters most influence outcomes. This pipeline is highly relevant for industrial AI applications where predictive modeling, optimization, and explainability must work together. The framework can be adapted to other manufacturing processes beyond battery production, making it a valuable reference for engineers working on process optimization with machine learning.
A Python framework combining 11 ML surrogate models, NSGA-II, and SHAP for lithium battery parameter optimization.