Robust optimization for functional multiresponse in 3D printing process

Feng, Zebiao; Wang, Jianjun; Zhou, Xiaojian; Zhai, Cuihong; Ma, Yizhong

doi:10.1016/j.simpat.2023.102774

Cited by 5 publications

(2 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For ease of notation, let 𝐑 𝑁 be denoted as 𝐂 𝑁 + 𝐷𝑖𝑎𝑔(𝜎 2 𝜀 (𝐱 𝑖 )), yielding 𝐊 𝑁 + 𝚺 𝑁 = 𝑣𝐑 𝑁 . The logarithm of the likelihood function is defined as shown in Equation (8).…”

Section: Gaussian Process Surrogatementioning

confidence: 99%

“…As a result, optimizing process parameters for multiple conflicting objectives has been extensively researched. 7 Feng et al 8 investigated the relationship between process parameters (e.g., layer height and infill rate) and quality characteristics of 3D printing products (e.g., creep rate and rupture time) based on computer models. They pointed out that the quality characteristics of the products are mainly determined by the process parameters.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multi‐objective Bayesian modeling and optimization of 3D printing process via experimental data‐driven method

Ding,

Wang,

et al. 2024

Quality & Reliability Eng

Self Cite

View full text Add to dashboard Cite

The instability of product quality and low printing efficiency are the main obstacles to the widespread application of 3D printing in the manufacturing industry. Optimizing printing parameters can substantially improve product quality and printing efficiency. However, existing methods for optimizing process parameters primarily rely on computationally expensive numerical simulations or costly physical experiments, which cannot balance model accuracy and experiment cost. To the best of our knowledge, almost no relevant papers have been found to address the issues of product quality and printing efficiency in 3D printing from experimental data‐driven perspective. In this paper, we propose a method that integrates multiobjective Bayesian optimization (MOBO) with experimental data‐driven, aiming at obtaining more accurate optimization results at a lower cost. Distinguishing from previous studies, the proposed method utilizes experimental data instead of predicted values to update the model and find the optimal process parameters based on expected hypervolume improvement. The results of the 3D printing case study show that the proposed method can better model and optimize the highly fluctuating 3D printing process and obtain the optimal process parameters at a much lower cost. In addition, confirmatory experiments verify that the proposed method achieves higher printing efficiency while maintaining product quality.

show abstract

Section: Gaussian Process Surrogatementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%