“A methodological approach to the optimization of EV Battery Specifications under manufacturability constraints”

This thesis investigates the optimization of electric vehicle battery parameters within a methodological framework that combines statistical analysis, clustering, and constraint-based optimization. The study begins with the construction of a comprehensive dataset, followed by systematic data preparation and codification, ensuring consistency and comparability across variables. Statistical tools such as Stepwise regression, LASSO, Pearson, and Spearman correlations are applied to identify dependencies between battery parameters and vehicle specifications, while a dependency matrix and hierarchical clustering are used to capture both the intensity and the structure of these relationships. Building on this analytical foundation, a custom optimization algorithm was developed to integrate both continuous and discrete variables into a unified model. The algorithm employs a utility-based objective function, where the relative importance of each parameter is derived from correlation patterns and adjusted to reflect manufacturability and feasibility considerations. A set of penalty functions ensures compliance with technical, economic, and regulatory constraints, while stochastic elements enhance the robustness of the search and prevent convergence toward local optima. The methodology is applied to two case studies demonstrating how the framework can refine battery configurations by balancing performance gains with realistic manufacturing and economic limitations.