Work

A Graph Isomorphic Network with Attention Mechanism for Intelligent Fault Diagnosis of Axial Piston Pump

Year: 2025

Type: article

Abstract: Axial piston pumps play a vital role in fluid power systems, which are widely employed in diverse fields such as aerospace, ocean engineering, and rail transit. It is essential to accurately diagnose faults in these pumps since their reliable operation hinges on it. A graph isomorphic network with a spatio-temporal attention mechanism (GIN-ST) is proposed in this paper for fault diagnosis of hydraulic axial piston pumps; GIN-AM addresses the problem of traditional intelligent fault diagnosis methods being limited to nonlinear mapping and transformation in Euclidean space. Initially, the weighted graphs are constructed from a univariate time series through K-nearest neighbor graph methods. Subsequently, a spatio-temporal attention-based module used to learn the graph representation of piston pump faults is presented, where a novel READOUT function and Transformer encoder provide spatial and temporal interpretability, respectively. Finally, the proposed (GIN-ST) model is compared against other intelligent fault diagnosis methods, and the superiority of the proposed method is proven.

Source: Applied Sciences

DOI: 10.3390/app15126586

Authors: Kai Li, Bofan Wu, Shiqi Xia, Xianshi Jia

Institution: Central South University

Language: English

Topic: Machine Fault Diagnosis Techniques

Subfield: Control and Systems Engineering

Field: Engineering

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $2,896

Funder: National Natural Science Foundation of China