Ensemble Dilated Convolutional Neural Network and Its Application in Rotating Machinery Fault Diagnosis.

Area of Science:

• Mechanical Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Rotating machinery fault diagnosis is crucial for industrial maintenance.
• Existing methods struggle with complex fault feature extraction from vibration signals.

Purpose of the Study:

• To develop a novel intelligent fault diagnosis scheme for rotating machinery.
• To enhance the accuracy and efficiency of fault detection using deep learning.

Main Methods:

• Ensemble of one-dimensional dilated convolutional neural networks (1D-DCNNs) with varied dilation rates.
• Early stopping optimization for efficient model training and prevention of overfitting.
• Weighted ensemble mechanism to combine outputs from multiple 1D-DCNNs for final diagnosis.

Main Results:

• The proposed ensemble dilated convolutional neural network (EDCNN) model effectively extracts diverse fault features from vibration signals.
• The EDCNN method demonstrated superior performance compared to state-of-the-art classical machine learning and deep learning approaches.
• Accurate identification of multiple rotating machinery faults was achieved, outperforming existing fault detection techniques.

Conclusions:

• The novel EDCNN framework offers a robust and accurate solution for rotating machinery fault diagnosis.
• The approach effectively mines fault features, improving diagnostic capabilities.
• This intelligent scheme provides a significant advancement in condition monitoring and predictive maintenance.