Constraint-Aware and User-Specific Product Design: A Machine Learning Framework for User-Centered Optimization

This study presents a data-driven, multi-objective optimization framework for user-centric product form design, integrating affective response modeling with coupled constraint satisfaction. Initially, morphological analysis and aesthetic evaluation are employed to extract critical design elements, while cluster analysis segments users based on preference data. Dominance-based rough set theory is then applied to derive group-specific affective patterns, which are subsequently modeled using Genetic Algorithm-optimized Backpropagation Neural Networks (GA-BPNN). The framework leverages Non-dominated Sorting Genetic Algorithm II (NSGA-II) to generate Pareto-optimal solutions, balancing aesthetic preferences and engineering constraints across user groups. A case study on SUV form design validates the proposed methodology, demonstrating its efficacy in delivering optimal, user-group-targeted design solutions while accommodating individual variability and constraint interdependencies. The results highlight the framework’s potential as a generalizable approach for emotion-aware, constraint-compliant product design.