DG-FuseNet: A Dual-Scale Dynamical Gated Fusion Framework for Cross-Domain Fault Diagnosis in Rotating Machinery

Area of Science:

• Mechanical Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Intelligent fault diagnosis of rotating components faces challenges like scale normalization, time-frequency localization, and multi-scale feature extraction.
• Varying operating conditions complicate accurate fault detection in machinery.

Purpose of the Study:

• To propose a novel convolutional neural network, DG-FuseNet, for improved intelligent fault diagnosis.
• To address limitations in existing methods for rotating component fault diagnosis.

Main Methods:

• Developed DG-FuseNet, a novel convolutional neural network architecture.
• Validated the model on real-world vibration signals from train vehicles and aero-engine systems.

Main Results:

• DG-FuseNet achieved diagnostic accuracies of 99.76% (train vehicles) and 94.32% (aero-engines).
• Outperformed eleven advanced intelligent models in convergence speed, accuracy, robustness, and generalization.
• Demonstrated superior performance and stability in complex industrial scenarios.

Conclusions:

• DG-FuseNet offers a significant advancement in intelligent fault diagnosis for rotating components.
• The model's robustness and generalization capabilities make it suitable for real-world industrial applications.
• DG-FuseNet represents a stable and high-performing solution for machinery health monitoring.