期刊
ADDITIVE MANUFACTURING
卷 38, 期 -, 页码 -出版社
ELSEVIER
DOI: 10.1016/j.addma.2021.101836
关键词
Gaussian process; Multi-material; Functionally graded materials; Selective laser melting; Stainless steel; Copper; Surface roughness; Density; Process parameter maps
资金
- University of Wisconsin-Madison Discovery to Product through the grant Draper Technology Innovation Fund
- UW2020 Wisconsin Alumni Research Foundation Discovery Institute funds
This study proposes a methodology for rapidly predicting suitable process parameters for additive manufacturing of 316L-Cu multi-material parts. By using machine learning algorithm and high-throughput experimentation method, the study predicts part density and surface roughness and generates process parameter maps.
The aim of this work is to propose a methodology for rapidly predicting suitable process parameters for additive manufacturing of a 316L-Cu multi-material part with a compositional gradient by using machine learning. Specifically, an algorithm based on a multivariate Gaussian process is developed to predict part density and surface roughness for a given set of laser power, velocity, and hatch spacing values. The training data for the algorithm is collected using a high-throughput experimentation method that allows for rapid measurement of part density, and surface roughness. After the model is validated using leave-one-out cross validation method, process parameter maps are generated for 316L-Cu parts manufactured using selective laser melting with premixed powder at mass fractions of 0.25, 0.50, and 0.75. A set of suitable process parameters are predicted using the process maps. It is shown that process parameters are a nonlinear function of gradient composition and neither process parameters of 316L or Cu are suitable for the graded region of a 316L-Cu multi-material part. Generated process maps provide a firsthand knowledge of process-property relationships for regions of compositional grading in 316L-Cu parts.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据