neural networks research group
areas
•
people
•
projects
•
demos
•
publications
•
software/data
•
Efficient Sampling for Design Optimization of an SLS Product (2017)
Nancy Xu,
Cem C. Tutum
In this work an efficient constrained surrogate-based sampling algorithm is implemented to optimize Selective Laser Sintering (SLS) process parameters for maximizing the tensile strength of a tensile specimen. Two variations of the algorithm have been implemented and tested on a Farsoon HT251P machine using (polyamid) PA3300 polymer powder. The algorithm is based on building a statistical predictive model of the objective response (i.e., maximization of tensile strength), aggregating the constraint function (i.e., limited amount of warping), in an iterative manner by simultaneously improving the accuracy of the predictive model as well as searching for the optimum set of process parameters. The difference in two algorithmic variations is the number of samples to update at each iteration. While the first method is based on a single sample update, the latter searches for multiple simultaneous updates to let the manufacturer try several potentially good sets of parameters in the same machine to eventually speed up the experimental evaluation procedure.
View:
HTML
Citation:
In
Proceedings of the 28th Annual International Solid Freeform Fabrication Symposium
, 12, Austin, TX, August 2017.
Bibtex:
@inproceedings{tutum:sff17, title={Efficient Sampling for Design Optimization of an SLS Product}, author={Nancy Xu and Cem C. Tutum}, booktitle={Proceedings of the 28th Annual International Solid Freeform Fabrication Symposium}, month={August}, address={Austin, TX}, pages={12}, url="http://nn.cs.utexas.edu/?tutum:sff17", year={2017} }
People
Cem C Tutum
Research Scientist
tutum [at] cs utexas edu
Areas of Interest
Supervised Learning
Evolutionary Computation
Other Areas