Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

Multi-attention collaborative temporal-spatial hypergraph attention network for machinery fault diagnosis.

Dongdong Liu¹, Xianju Cheng², Zhichao Jiang³

¹Key Laboratory of Advanced Manufacturing Technology, Beijing University of Technology, Beijing 100124, China; Chongqing Research Institute of Beijing University of Technology, Chongqing 401121, China.

ISA Transactions

|May 22, 2026

Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Adsorption-desorption behavior perspective deciphers the photo-detoxification of high-concentration Cr(VI) by crystalline Bi<sub>4</sub>O<sub>5</sub>I<sub>2</sub>/amorphous Fe<sub>2</sub>O<sub>3</sub> junctions.

Journal of colloid and interface science·2026

Same author

Combining Network Pharmacology, Machine Learning, Molecular Docking, and Experimental Validation to Explore the Mechanism of Danggui-Shaoyao-San in treating Rheumatoid arthritis.

Current pharmaceutical design·2026

Same author

Comparing Efficacy of Oxycodone versus Sufentanil for Postoperative Recovery in Patients Receiving Bilateral Modified Thoracoabdominal Nerves Block Through Perichondrial Approach (M-TAPA) Following Laparoscopic Radical Distal Gastrectomy: A Study Protocol for a Prospective, Blinded, Randomized Controlled Trial.

Journal of pain research·2026

Same author

Allosteric Inhibition of Polycomb Repressive Complex 2 by an EZH2-Selective Small Molecule Inhibitor.

Advanced science (Weinheim, Baden-Wurttemberg, Germany)·2026

Same author

"Looking inward," creativity, and well-being: an SEM analysis of self-reflective rumination and creative self-efficacy as pathways linking self-beliefs in creativity and well-being in college students.

Frontiers in psychology·2026

Same author

Research Progress Review on the Activation of Bioactive Substances by Targeted Fermentation of Rice Bran.

Foods (Basel, Switzerland)·2026

This study introduces a novel hypergraph attention network for machinery fault diagnosis. The method effectively captures complex interactions and achieves high accuracy even with limited labeled data.

Area of Science:

Machinery fault diagnosis
Graph-based machine learning
Artificial intelligence in engineering

Background:

Traditional graph models struggle with high-order interactions and require extensive labeled data for machinery fault diagnosis.
Existing methods often fail to capture complex, non-pairwise relationships among multiple data samples.

Purpose of the Study:

To propose a novel multi-attention collaborative temporal-spatial hypergraph attention network (MA-TSHGAN) for enhanced machinery fault diagnosis.
To address the limitations of existing methods in handling high-order interactions and limited labeled data.
To improve the accuracy and efficiency of fault diagnosis systems.

Main Methods:

Developed an attention-aware hypergraph construction method to model high-order relationships with dynamic hyperedge weighting.

Keywords:

Fault diagnosis Hypergraph attention network Rotating machinery

Related Experiment Videos

Implemented a hypergraph label passing method (HLPM) for semi-supervised learning, propagating limited labeled data information.

Designed a multi-attention collaborative hypergraph attention network to capture temporal-spatial features and aggregate multi-scale neighborhood information.

Main Results:

The proposed MA-TSHGAN achieved high average recognition accuracies of 99.82%, 99.83%, and 97.63% on three datasets.
Demonstrated superior performance compared to several state-of-the-art methods, particularly with limited labeled samples.
Effectively captured high-order interactions and temporal-spatial features for accurate fault diagnosis.

Conclusions:

The MA-TSHGAN, incorporating hypergraph label passing and multi-attention mechanisms, significantly advances machinery fault diagnosis.
The method offers a robust solution for scenarios with limited labeled data, outperforming existing approaches.
This work highlights the potential of hypergraph neural networks in complex industrial diagnostic tasks.