Machine learning enables high-precision formation prediction of ultrasonic-assisted friction stir channeling

Shengnan Hu; Yuming Xie; Xiangchen Meng; Yilong Han; Shenglong Wang; Cheng Shan; Yongxian Huang

doi:10.1016/j.mfglet.2026.04.003

Machine learning enables high-precision formation prediction of ultrasonic-assisted friction stir channeling

Shengnan Hu
, Yuming Xie^*
, Xiangchen Meng
, Yilong Han
, Shenglong Wang
, Cheng Shan
, Yongxian Huang

^*Corresponding author for this work

School of Materials Science and Engineering

Harbin Institute of Technology
National University of Singapore
Beijing Hangxing Machinery Manufacture Company

Research output: Contribution to journal › Article › peer-review

Abstract

Ultrasonic-assisted friction stir channeling for ZL114 cast aluminum alloys was proposed to achieve high-rectangularity channels optimized by neural network. The machine learning model achieved prediction accuracies of 100% for surface index, 92% for width, 82% for height, and 85% for rectangularity, respectively, with rectangularity enhanced the range from 0.85 to 0.95. Pareto frontier analysis demonstrated that ultrasonic assistance increased channel height by 26% while significantly improving width and rectangularity.

Original language	English
Pages (from-to)	76-81
Number of pages	6
Journal	Manufacturing Letters
Volume	48
DOIs	https://doi.org/10.1016/j.mfglet.2026.04.003
State	Published - Jun 2026

Keywords

Cast aluminum alloys
Friction stir channeling
Neural network
Thermal management
Ultrasonic assistance

Access to Document

10.1016/j.mfglet.2026.04.003

Cite this

@article{1282ec02c8ed4be3bc6769a833c80ba3,

title = "Machine learning enables high-precision formation prediction of ultrasonic-assisted friction stir channeling",

abstract = "Ultrasonic-assisted friction stir channeling for ZL114 cast aluminum alloys was proposed to achieve high-rectangularity channels optimized by neural network. The machine learning model achieved prediction accuracies of 100\% for surface index, 92\% for width, 82\% for height, and 85\% for rectangularity, respectively, with rectangularity enhanced the range from 0.85 to 0.95. Pareto frontier analysis demonstrated that ultrasonic assistance increased channel height by 26\% while significantly improving width and rectangularity.",

keywords = "Cast aluminum alloys, Friction stir channeling, Neural network, Thermal management, Ultrasonic assistance",

author = "Shengnan Hu and Yuming Xie and Xiangchen Meng and Yilong Han and Shenglong Wang and Cheng Shan and Yongxian Huang",

note = "Publisher Copyright: {\textcopyright} 2026 Society of Manufacturing Engineers (SME)",

year = "2026",

month = jun,

doi = "10.1016/j.mfglet.2026.04.003",

language = "英语",

volume = "48",

pages = "76--81",

journal = "Manufacturing Letters",

issn = "2213-8463",

publisher = "Society of Manufacturing Engineers",

}

TY - JOUR

T1 - Machine learning enables high-precision formation prediction of ultrasonic-assisted friction stir channeling

AU - Hu, Shengnan

AU - Xie, Yuming

AU - Meng, Xiangchen

AU - Han, Yilong

AU - Wang, Shenglong

AU - Shan, Cheng

AU - Huang, Yongxian

PY - 2026/6

Y1 - 2026/6

N2 - Ultrasonic-assisted friction stir channeling for ZL114 cast aluminum alloys was proposed to achieve high-rectangularity channels optimized by neural network. The machine learning model achieved prediction accuracies of 100% for surface index, 92% for width, 82% for height, and 85% for rectangularity, respectively, with rectangularity enhanced the range from 0.85 to 0.95. Pareto frontier analysis demonstrated that ultrasonic assistance increased channel height by 26% while significantly improving width and rectangularity.

AB - Ultrasonic-assisted friction stir channeling for ZL114 cast aluminum alloys was proposed to achieve high-rectangularity channels optimized by neural network. The machine learning model achieved prediction accuracies of 100% for surface index, 92% for width, 82% for height, and 85% for rectangularity, respectively, with rectangularity enhanced the range from 0.85 to 0.95. Pareto frontier analysis demonstrated that ultrasonic assistance increased channel height by 26% while significantly improving width and rectangularity.

KW - Cast aluminum alloys

KW - Friction stir channeling

KW - Neural network

KW - Thermal management

KW - Ultrasonic assistance

UR - https://www.scopus.com/pages/publications/105039584918

U2 - 10.1016/j.mfglet.2026.04.003

DO - 10.1016/j.mfglet.2026.04.003

M3 - 文章

AN - SCOPUS:105039584918

SN - 2213-8463

VL - 48

SP - 76

EP - 81

JO - Manufacturing Letters

JF - Manufacturing Letters

ER -

Machine learning enables high-precision formation prediction of ultrasonic-assisted friction stir channeling

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this