Enhancing Design for Additive Manufacturing Through 3D Objects Clustering Using HICA and Segmentation Techniques

Abstract

A novel approach for 3D objects clustering to enhance detail design phase is developed with the Hierarchical Clustering Algorithm (HICA). This bigdata analytic extract features like vertex count, genus or convexity. The aim is to classify a 3D objects database and their compatibility with different manufacturing technologies (casting, milling and additive manufacturing), thus facilitating more informed decision-making for designers. The result is the identification of main features based on Variance Inflation Factor (VIF) that enables to evaluate clusters that share similar characteristics. Mean dihedral angles, minimal thickness, betti numbers, accessibility score are found. Then principal components are employed to offer specific information and enable interpretation. This latter can be directly applied to refine and adapt their designs for various manufacturing technologies or use specific segmentation tools.