Parallel non-dominated sorting genetic algorithm-II for optimal part deposition orientation in additive manufacturing based on functional features

Abstract

Surface finish, especially the surface finish of functional features, and build time are two important concerns in additive manufacturing. A suitable part deposition orientation can enhance the surface quality of functional features and reduce the build time. This article proposes a novel method to obtain an optimum part deposition orientation for industrial-grade 3D printing based on fused deposition modeling process by considering two objective functions at a time, namely adaptive feature roughness (the weighted sum of all feature roughnesses) and build time. First, mesh segmentation and level classification of features are carried out. Then, models for evaluation of adaptive feature roughness and build time are established. Finally, a non-dominated sorting genetic algorithm-II based on Compute Unified Device Architecture is used to obtain the Pareto-optimal set. The feasible of the algorithm is evaluated on several examples. Results demonstrate that the proposed parallel algorithm obtains a limiting solution that enhances the surface quality of functional features significantly and reduces average running time by 94.8% compared with the traditional genetic algorithm.