Enhanced Rule Generation in Product Design Through Rough Set Theory and Ant Colony Optimization

Abstract

Limitations inherent in conventional rule generation methodologies, particularly concerning knowledge redundancy and efficiency in product design, are addressed through the adoption of a rough set-based approach in this study. An enhancement to the Ant Colony Optimization (ACO) algorithm's information gain ratio is introduced by integrating a redundancy detection mechanism, which notably accelerates the convergence process. Furthermore, the application of a classification consistency algorithm effectively minimizes the number of attributes, facilitating the extraction of potential associative rules. Comparative validation performed on a public dataset demonstrates that the proposed attribute reduction approach surpasses existing methods in terms of attribute count reduction, reduction rate, and execution time. When applied to an automotive design case study, the approach yields rules with 100% coverage and accuracy, characterized by a reduced average number of attributes per rule. These findings underscore the superiority of the rough set-based methodology in generating product design rules, providing a robust framework that enhances both the precision and applicability of the design process.