Intelligent Fault Diagnosis and Forecast of Time-Varying Bearing Based on Deep Learning VMD-DenseNet

Area of Science:

• Mechanical Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Rolling bearings are critical components in rotating machinery.
• Effective fault diagnosis is essential for preventing equipment failure and ensuring operational reliability.
• Traditional methods often require manual feature extraction, which can be complex and time-consuming.

Purpose of the Study:

• To propose a novel deep learning approach for automated bearing fault diagnosis.
• To develop a system capable of accurately identifying various bearing fault types.
• To enhance the efficiency and accuracy of mechanical intelligent fault diagnosis systems.

Main Methods:

• Utilizing Variational Mode Decomposition (VMD) to transform vibration signals into Hilbert spectrum images.
• Employing the lightweight DenseNet convolutional neural network for automated image classification and feature extraction.
• Integrating signal acquisition, feature extraction, fault identification, and prediction into a comprehensive system.

Main Results:

• The VMD-DenseNet model demonstrated a 92% prediction accuracy in identifying four common motor faults.
• The method eliminates the need for manual feature selection by leveraging deep learning for feature extraction directly from images.
• The system successfully established the state of bearings using data from a time-varying bearing experimental device.

Conclusions:

• The VMD-DenseNet approach offers a highly effective and accurate method for bearing fault diagnosis.
• This research provides a robust foundation for developing intelligent fault diagnosis systems with wide engineering applications.
• Future work aims to enable online and real-time diagnosis capabilities for enhanced machinery monitoring.