0
Research Papers

Classifying the Dimensional Variation in Additive Manufactured Parts From Laser-Scanned Three-Dimensional Point Cloud Data Using Machine Learning Approaches

[+] Author and Article Information
M. Samie Tootooni, Ashley Dsouza, Ryan Donovan, Peter Borgesen

System Science and Industrial
Engineering Department,
Binghamton University (SUNY),
Binghamton, NY 13902

Prahalad K. Rao

Department of Mechanical
and Materials Engineering,
University of Nebraska–Lincoln,
Lincoln, NE 68588-0452
e-mail: rao@unl.edu

Zhenyu (James) Kong

Grado Department of Industrial
and Systems Engineering,
Virginia Tech,
Blacksburg, VA 24060

1Corresponding author.

Manuscript received February 4, 2017; final manuscript received April 24, 2017; published online June 22, 2017. Assoc. Editor: Zhijian J. Pei.

J. Manuf. Sci. Eng 139(9), 091005 (Jun 22, 2017) (14 pages) Paper No: MANU-17-1073; doi: 10.1115/1.4036641 History: Received February 04, 2017; Revised April 24, 2017

The objective of this work is to develop and apply a spectral graph theoretic approach for differentiating between (classifying) additive manufactured (AM) parts contingent on the severity of their dimensional variation from laser-scanned coordinate measurements (3D point cloud). The novelty of the approach is in invoking spectral graph Laplacian eigenvalues as an extracted feature from the laser-scanned 3D point cloud data in conjunction with various machine learning techniques. The outcome is a new method that classifies the dimensional variation of an AM part by sampling less than 5% of the 2 million 3D point cloud data acquired (per part). This is a practically important result, because it reduces the measurement burden for postprocess quality assurance in AM—parts can be laser-scanned and their dimensional variation quickly assessed on the shop floor. To realize the research objective, the procedure is as follows. Test parts are made using the fused filament fabrication (FFF) polymer AM process. The FFF process conditions are varied per a phased design of experiments plan to produce parts with distinctive dimensional variations. Subsequently, each test part is laser scanned and 3D point cloud data are acquired. To classify the dimensional variation among parts, Laplacian eigenvalues are extracted from the 3D point cloud data and used as features within different machine learning approaches. Six machine learning approaches are juxtaposed: sparse representation, k-nearest neighbors, neural network, naïve Bayes, support vector machine, and decision tree. Of these, the sparse representation technique provides the highest classification accuracy (F-score > 97%).

FIGURES IN THIS ARTICLE
<>
Copyright © 2017 by ASME
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Fig. 1

Flooded contour plots (dimensional deviation maps) of the test artifact used in FFF experiments detailed in Sec. 3.2. The material is ABS polymer. The first row (1) shows the top views and the second row (2) contains the bottom views of the parts. Figures (a)–(d) represent different parts printed under infill percentages of 70%, 80%, 90%, and 100% at 230 °C, respectively. The color scale shown on the right is in millimeter. The dimensional variations progressively increase with infill percentage (see color figure online).

Grahic Jump Location
Fig. 2

Design of the circle–square–diamond part—a simplified embodiment of the NAS 979 standard test artifact for testing accuracy of machining centers. The dimensions are in millimeters. Figures (a) and (b) are front and top views of the part, respectively, and figure (c) is an isometric projection of the part (The NAS 979 standard artifact was first used in the AM context by Cooke and Soons [26]).

Grahic Jump Location
Fig. 3

The eight points used for alignment of the scan points with the CAD model

Grahic Jump Location
Fig. 4

The design of experiments matrix and resulting RMS and in-tolerance percentage at different settings

Grahic Jump Location
Fig. 5

Micrograph showing the effect of infill percentage on the internal morphology of the circle–square–diamond part. Shown is the quarter cross section of the part; the circular section is at the top end. At 100% infill, the thermal residual stresses cannot be accommodated without deleteriously affecting the dimensional integrity since there is no vacant space for stress relief.

Grahic Jump Location
Fig. 6

Random samples taken from deviation coordinates of two FFF parts are mapped into unweighted, undirected graphs. These graphs are from parts printed with (a) 90% infill at 230 °C temperature and (b) 70% infill at 230 °C temperature. The distinctive connectivity structures of these graphs are indicative of the differences in geometric variation between parts created at 90% and 70% infill.

Grahic Jump Location
Fig. 7

(a) A fully connected representative graph, with the dark and light nodes analogous to different dimensional variations on the part. Edge distances are estimated using the radial basis function (Eq. (4)). (b) Edges connecting similar nodes are identified with dashed lines using the threshold function (Eq. (5)). (c) Edges connecting similar textured nodes are pruned [32].

Grahic Jump Location
Fig. 8

Scenario I sampling of point cloud data. In scenario I sampling, the circle–square–diamond part is divided into nine areas: (a) top view and (b) 3D view.

Grahic Jump Location
Fig. 9

Three-dimensional view of the designed part in scenario II sampling approach; in this scenario, the XY planes considered as the four areas. Samples are taken at different Z-heights (build direction).

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In