Energy efficiency design for eco-friendly additive manufacturing based on multimodal attention fusion. (July 2022)

• Record Type: Journal Article
• Authors: Wang, Kang; Song, Youyi; Sheng, Hongsheng; Xu, Jinghua; Zhang, Shuyou; Qin, Jing
• Is Part Of: Journal of manufacturing processes. Volume 79(2022)
• Page Start: 720
• Page End: 730
• Publication Date: 2022-07
• Subjects: Energy efficiency design -- Additive manufacturing -- Energy consumption prediction -- Multimodal fusion -- Attentional feature; Production management -- Data processing -- Periodicals; Manufacturing processes -- Periodicals; Procestechnologie; Productietechniek; Production -- Gestion -- Informatique -- Périodiques; Fabrication -- Périodiques; Manufacturing processes; Production management -- Data processing; Periodicals; 670.5
• DOI: 10.1016/j.jmapro.2022.05.010
• Languages: English
• ISSNs: 1526-6125

Abstract

Abstract: Additive manufacturing (AM) holds an imponderable scope to reduce energy consumption (EC) and material waste in contrast to subtractive manufacturing methodologies. With the increasing concerns about sustainable production, there is an urgent demand to critically assess the EC, and integrate the EC assessment into AM design processes. To address this issue, an energy efficiency design method for eco-friendly additive manufacturing based on Multimodal Attention Fusion Network (MAFN) is proposed to take full advantage of the derived multiple modalities from the computer-aided design (CAD) and manufacturing process. In view of the operating characteristics in AM process, the mathematical EC model of AM is first constructed to seek the salient modalities for the following EC prediction task. The proposed MAFN then leverages a Multimodal Fusion Framework (MFF) as the backbone module, unifying the processing-, pixel-, and geometric-level data to adapt to the increasing part geometry complexity and the process parameters. An Attentional Feature Fusion Module (AFFM) is further proposed to subtly combine the inherent correlation and variation among attentional features, preventing only learning similar features from multiple sources. Extensive numerical and physical experimental results demonstrate that the proposed method consistently outperforms the state-of-the-art EC prediction approaches and improves the energy efficiency design without a mechanical performance loss for eco-friendly additive manufacturing.