Fault analysis on deep groove ball bearing using ResNet50 and AlexNet50 algorithms.

Area of Science:

• Mechanical Engineering
• Artificial Intelligence
• Predictive Maintenance

Background:

• Deep Groove Ball Bearings (DGBBs) are critical industrial components subjected to axial and radial loads.
• Bearing faults represent a significant risk factor, impacting machinery performance and safety.
• Existing methods for fault detection require robust classification strategies.

Purpose of the Study:

• To classify four distinct types of Deep Groove Ball Bearing faults: Case Fault (CF), Ball Fault (BF), Inner Ring Fault (IRF), and Outer Ring Fault (ORF).
• To evaluate the effectiveness of Artificial Neural Networks (ANNs) for automatic bearing fault classification.
• To identify the most significant features contributing to accurate fault diagnosis.

Main Methods:

• Utilized 14 input features for bearing fault evaluation.
• Implemented a feature ranking method to determine the contribution of each parameter.
• Employed Artificial Neural Networks (ANNs), including Resnet50, for automatic fault classification.
• Trained and compared various algorithms, assessing prediction probabilities.

Main Results:

• Achieved a high classification accuracy of 97.9% using the Resnet50 algorithm.
• The neural network classifier learner demonstrated 97% accuracy.
• The probability of correct predictions was observed to decrease with an increased number of fault samples.

Conclusions:

• Artificial Neural Networks, particularly Resnet50, are highly effective for accurate Deep Groove Ball Bearing fault classification.
• Feature ranking provides valuable insights into the parameters influencing fault diagnosis.
• The developed method offers a promising approach for predictive maintenance in industrial applications.