0
Research Papers

Improving Machined Surface Shape Prediction by Integrating Multi-Task Learning With Cutting Force Variation Modeling

[+] Author and Article Information
Chenhui Shao

Department of Mechanical
Science and Engineering,
University of Illinois at Urbana-Champaign,
Urbana, IL 61801
e-mail: chshao@illinois.edu

Jie Ren

Department of Industrial and
Manufacturing Engineering,
Florida State University,
Tallahassee, FL 32310
e-mail: jr14r@my.fsu.edu

Hui Wang

Department of Industrial and
Manufacturing Engineering,
Florida State University,
Tallahassee, FL 32310
e-mail: hwang10@fsu.edu

Jionghua (Judy) Jin

Department of Industrial and
Operations Engineering,
University of Michigan,
Ann Arbor, MI 48109
e-mail: jhjin@umich.edu

S. Jack Hu

Department of Mechanical Engineering,
University of Michigan,
Ann Arbor, MI 48109;
Department of Industrial and
Operations Engineering,
University of Michigan,
Ann Arbor, MI 48109
e-mail: jackhu@umich.edu

1Corresponding author.

Manuscript received April 23, 2016; final manuscript received August 5, 2016; published online September 29, 2016. Assoc. Editor: Radu Pavel.

J. Manuf. Sci. Eng 139(1), 011014 (Sep 29, 2016) (11 pages) Paper No: MANU-16-1244; doi: 10.1115/1.4034592 History: Received April 23, 2016; Revised August 05, 2016

The shapes of machined surfaces play a critical role affecting powertrain performance, and therefore, it is necessary to characterize the shapes with high resolution. State-of-the-art approaches for surface shape characterization are mostly data-driven by interpolating and extrapolating the spatial data but its precision is limited by the density of measurements. This paper explores the new opportunity of improving surface shape prediction through considering the similarity of multiple similar manufacturing processes. It is a common scenario when the process of interest lacks sufficient data whereas rich data could be available from other similar-but-not-identical processes. It is reasonable to transfer the insights gained from other relevant processes into the surface shape prediction. This paper develops an engineering-guided multitask learning (EG-MTL) surface model by fusing surface cutting physics in engineering processes and the spatial data from a number of similar-but-not-identical processes. An iterative multitask Gaussian process learning algorithm is developed to learn the model parameters. Compared with the conventional multitask learning, the proposed method has the advantages in incorporating the insights on cutting force variation during machining and is potentially able to improve the prediction performance given limited measurement data. The methodology is demonstrated based on the data from real-world machining processes in an engine plant.

Copyright © 2017 by ASME
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Fig. 1

Data loss in high-resolution measurement

Grahic Jump Location
Fig. 2

Potential knowledge transfer in manufacturing applications: (a) new plant and (b) low production station

Grahic Jump Location
Fig. 3

Difference between single task learning and multitask learning: (a) single task learning and (b) multitask learning

Grahic Jump Location
Fig. 4

Illustration of the EG-MTL scheme

Grahic Jump Location
Fig. 5

Flowchart for the iterative algorithm

Grahic Jump Location
Fig. 6

Implementation procedure of multitask learning

Grahic Jump Location
Fig. 7

An engine head surface example

Grahic Jump Location
Fig. 8

RMSE comparison for the EG-MTL model, GPMTL, and kriging

Grahic Jump Location
Fig. 9

Engine surface example for the hyperparameter study

Grahic Jump Location
Fig. 10

Effects of hyperparameters on RMSE: (a) RMSE versus τ, (b) RMSE versus π, and (c) RMSE versus δ2

Grahic Jump Location
Fig. 11

Effects of hyperparameter pairs on RMSE: (a) RMSE versus (τ, π), (b) RMSE versus (τ, δ2), and (c) RMSE versus (π, δ2)

Grahic Jump Location
Fig. 12

Effects of sample size on the prediction performance: (a) RMSE and (b) Δ+RMSE

Grahic Jump Location
Fig. 13

Effects of number of tasks on the prediction performance: (a) RMSE and (b) Δ+RMSE

Grahic Jump Location
Fig. 14

Effects of sample size and number of tasks on the prediction performance: (a) RMSE and (b) Δ+RMSE

Grahic Jump Location
Fig. 15

Effects of ρ(U, Z) on the prediction performance: (a) RM SE versus ρ(U, Z) and (b) Δ+RMSE versus ρ(U, Z)

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