A Hybrid Model Based on Jensen's Inequality Theory for 3D Printed Lattice Structures Maximum Stress Prediction. Issue 1 (1st December 2022)

• Record Type: Journal Article
• Authors: Zhang, Zhiwei; Zhang, Yuyan; Wen, Yintang; Ren, Yaxue
• Is Part Of: Journal of physics. Volume 2403 Issue 1(2022)
• Publication Date: 2022-12-01
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/2403/1/012005
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: Inherent defects in 3D-printed metal lattice structures are inevitable and significantly affect the structural part's maximum stress. To solve this problem, a hybrid model based on Jensen's Inequality Theory to predict the maximum stress of 3D printed lattice structures is proposed in this study. Jensen's inequality theory means that the average of the model predictions will not be lower than the predictions of the individual prediction models. Therefore, the optimal number of hybrid models is discussed and designed, and the optimal base estimators are trained and selected on the same dataset. A hyperparametric approach is used to optimise the base estimator parameters. The experimental results demonstrate that the hybrid model performs better than a single estimator's prediction, with an R 2 of 0.83. This research offers a novel approach for the prediction of 3D-printed lattice structures using machine learning.