4.7 Article

Geometrical defect detection for additive manufacturing with machine learning models

Journal

MATERIALS & DESIGN
Volume 206, Issue -, Pages -

Publisher

ELSEVIER SCI LTD
DOI: 10.1016/j.matdes.2021.109726

Keywords

Additive manufacturing; Defect detection; Machine learning

Ask authors/readers for more resources

This study introduced a Machine Learning-based scheme for detecting geometric defects in additively manufactured objects, using synthetic 3D point clouds for training and outperforming the existing Z-difference method. Bagging and Random Forest were identified as the best prediction models, suitable for various conditions of point cloud densities and defect sizes.
This study proposed a scheme based on Machine Learning (ML) models to detect geometric defects of additively manufactured objects. The ML models are trained with synthetic 3D point clouds with defects and then applied to detect defects in actual production. Using synthetic 3D point clouds rather than experimental data could save a huge amount of training time and costs associated with many prints for each design. Besides distance differences of individual points between source and target point clouds, this scheme uses a new concept called patch to capture macro-level information about nearby points for ML training and implementation. Numerical comparisons of prediction results on experimental data with different shapes showed that the proposed scheme outperformed the existing Z-difference method in the literature. Five ML methods (Bagging of Trees, Gradient Boosting, Random Forest, K-nearest Neighbors and Linear Supported Vector Machine) were compared under various conditions, such as different point cloud densities and defect sizes. Bagging and Random Forest were found the two best models regarding predictability; and the right patch size was found to be at 20. The proposed ML-based scheme is applicable to in-situ defect detection during additive manufacturing with the aid of a proper 3D data acquisition system. (c) 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.7
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

No Data Available
No Data Available